1359.0 - Methods changes during the COVID-19 period, Jun 2020  
Latest ISSUE Released at 11:30 AM (CANBERRA TIME) 17/06/2020  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All RSS Feed RSS Bookmark and Share Search this Product

This document was added or updated on 08/07/2020.

Measuring the Consumer Price Index during a time of COVID-19


Introduction

Following the March 2020 quarter Note on the impact of COVID-19 on the CPI, this article explains the changes that will be made to the June quarter CPI as a result of the actions taken by governments, businesses and individuals in response to the COVID-19 pandemic (scheduled for release on 29 July). The following measurement approaches have been adopted:

  1. The implementation of free childcare and the changes in rental agreements that reduced the prices paid by tenants, have flowed through to measured CPI.
  2. For unavailable goods and services, the quarterly price movements will be imputed from the headline CPI quarterly movement. This applies to: urban transport fares; domestic holiday travel and accommodation; international holiday travel and accommodation; sports participation; and other recreational, sporting and cultural services.
  3. Seasonal forward factors will be used for three CPI series: child care; medical and hospital services; and pre-school and primary education, while the remaining seasonally adjusted series will maintain their (usual) concurrent seasonal adjustment approach.
  4. CPI expenditure weights have not been updated for changes in household expenditure patterns as a result of COVID-19. Expenditure weights will be updated in the December 2020 quarter, consistent with current practice.

For further information, please email prices.statistics@abs.gov.au


Price collection

The ABS has a range of data sources for the CPI including: direct collection, administrative data, web scraping and transactions 'scanner' data. Direct collection, which contributes slightly more than half the weight of the CPI, is largely collected via on line or over the phone. Less than two per cent of the weight of the CPI is collected by ABS officers in-store.

In late March, the ABS suspended in-store collection to ensure the health and safety of our collectors. This decision remained in place for the June quarter. The ABS continued to directly collect prices online and over the phone, and moved the majority of in-store collections to these modes.


Impact of free child care

On 2 April, the Australian Government announced that child care services would be free for families from 6 April to 28 June. Following the 2 April announcement, free child care was extended until 12 July.

Child care contributes 1.2 per cent of household expenditure in the CPI and is measured through changes in the out-of-pocket expenses for families. Figure 1 summarises changes to various child care support schemes over the years, and the impact these had on the child care price series.

Figure 1 - Child care index in the CPI (2011-12 = 100.0)
Graph showing time series of child care index in the CPI from 2004 to 2020

The latest changes reduced child care out-of-pocket expenses for households to zero for most of the June quarter. Free child care will be taken into account by using the number of days it was in effect (62 out of 65 business days). This results in a 95 per cent price fall in the CPI child care expenditure class for the June quarter, subtracting approximately 1.1 percentage points from the headline CPI.

The scheduled ending of free child care on 12 July will result in a significant rise in the September 2020 quarter childcare price series.

Free child care will also impact these series in the June 2020 quarter:
  • Preschool and primary education, with before and after school care services also being free.
  • Selected Living Cost Indexes, which measure changes in living costs for different household types. Free child care will have a higher impact on employee households, for which the child care series has a weight of 1.5 per cent, compared to a weight of 1.2 per cent in the CPI.


Measuring price change in rents

COVID-19 resulted in significant changes in the rental property market with State and Territory Governments announcing a range of rental support packages.

Rents contribute 6.8 per cent of household expenditure in the CPI, with privately owned dwelling rents contributing 92 per cent of total rents expenditure. Figure 2 provides a summary of the different scenarios by jurisdiction, and how they will be captured in the CPI.

Figure 2 - June quarter scenarios in the CPI rent series.
Diagram showing a summary of the different rent scenarios, which jurisdictions they are applicable, and how they will be captured in the CPI

Permanent or temporary re-negotiations of rent between landlords and tenants will be treated as a price fall in the CPI. Rent reductions due to support packages, such as land tax or rental relief grants, will also result in a price fall in the CPI. This is consistent with the principle that prices collected in the CPI should reflect the out-of-pocket expenditure by the consumer, including any adjustments for Government subsidies or assistance.

The CPI measures price change in the period when the good or service is actually received, regardless of the period in which payment or use occurs. This is known as the 'Acquisitions approach'. Therefore, any deferrals in rental payments will not impact the CPI as rents will be measured as the price including the deferred amount. Eviction moratoriums and rent freezes will also have no impact on prices.


Use of imputation for missing prices

It is common for there to be some missing prices from the CPI each quarter due to goods or services being temporarily unavailable, or businesses being temporarily closed. Missing prices typically represent around 5-6 per cent of the CPI sample each quarter.

For the June 2020 quarter, the proportion of missing prices was 13 per cent, reflecting unavailable goods and services due to COVID-19 restrictions. However, when allowance is made for the goods and services that were unavailable, such as airfares, the proportion of missing prices was closer to 7 per cent. This is only slightly higher than the June 2019 quarter proportion of 5.5 per cent.

The increase in the proportion of missing prices will not have an adverse impact on the quality of the CPI due to the use of imputation methods. Appendix 1 contains a review of the responses from other National Statistical Organisation on this topic.

When discussing imputation in the CPI, a distinction needs to be made between:
  1. Available goods and services to purchase, but with some prices temporarily missing.
  2. Unavailable goods and services, meaning there are no prices to collect for the entire quarter.

Available goods and services

The ABS has well established imputation methods for this scenario, which are supported internationally in the Consumer Price Index Manual: Theory and Practice and used widely by other National Statistical Organisations.

In most cases of temporarily missing prices, a movement is imputed off similar goods or services. This is known as class (or sample) mean imputation. This uses the assumption that the movement of the missing price is the average price change of similar goods and services. The class mean approach is commonly applied to items within the clothing, furniture and household appliance expenditure groups.

In some cases, the carry forward imputation approach is used, with available prices from the previous period and a movement of zero recorded. The carry forward approach is used where prices do not change very frequently, such as for renting a dwelling.

Unavailable goods and services (footnote 1)

International guidelines and methods are less well established for imputation where no prices are available or no consumption has taken place (even when prices are available). Examples of this during the June 2020 quarter include domestic and international travel, and attending sporting events.

The ABS developed a CPI imputation framework to decide the imputation method to be used for different scenarios (figure 3).

The first decision is to determine whether the good or service is available or unavailable to purchase, regardless of whether prices can be collected. The next decisions to follow are:
  • in the case of an available good or service where expenditure has declined significantly, it will be treated as unavailable. Urban transport fares are an example where expenditure has declined by over 80 per cent from normal levels.
  • in the case of an unavailable good or service where a significant amount of its typical expenditure has been substituted to another good or service, donor imputation will be used. An example of this is consumers substituting alcohol consumed on premises to takeaway alcohol.

Figure 3 - CPI imputation decision framework
Diagram showing a decision tree framework to determine the imputation approach with the advent of unavailable goods and services


Imputation approach
  • Standard imputation - used as described above in the case of goods and services that are available to purchase but for which prices are temporarily missing. This may be due to the item being out of stock or the business being closed. There is a small increase in the number of temporarily missing prices in the June quarter where this approach will be used.
  • Donor imputation - used for an unavailable good or service where a significant amount of its typical expenditure has been substituted to another good or service. This approach imputes off a 'donor' series located elsewhere in the CPI. For example, takeaway alcohol will be used as the donor for alcohol consumed on premises, which is classified separately within the Alcohol and tobacco group.
  • Imputation off the headline CPI - used for an unavailable good or service where consumption is not being substituted elsewhere. This has the effect of the movement of the imputed series not contributing to the CPI quarterly movement. This reflects the fact that there are no prices being transacted, and no (or very little) consumption of the particular good or service has occurred during the quarter.

Table 1 provides a summary of the imputation methods that will be used for the goods and services particularly affected by COVID-19 actions in the June 2020 quarter.

Table 1 - CPI imputation approach for the June 2020 quarter.
Series
Goods and services included
Weight (%)
Imputation approach
Restaurant mealsEntree, main and dessert meals from restaurants, cafes and clubs
3.3
If available, takeaway prices used or standard imputation if not available
Alcohol consumed on premisesBeer, wine and spirits from restaurants, cafes and clubs
1.5
Donor imputation off takeaway alcohol
Urban transport faresTrain, bus, tram, ferry
0.7
Impute off headline CPI
Domestic holiday travel and accommodation Airfares, train, bus, ferry, hotels, motels, holiday houses, caravan parks, shared accommodation.
3.0
Impute off headline CPI
International holiday travel and accommodationAirfares, tours, overseas accommodation
3.4
Impute off headline CPI
Sports participationGym and sports club membership fees, sports lesson fees
0.9
Impute off headline CPI
Other recreational, sporting and cultural services Attending sports events, cinema, theatre, lessons
1.1
Impute off headline CPI


Seasonal adjustment in the CPI

Of the 87 CPI expenditure class series published by the ABS, 55 are seasonally adjusted. With COVID-19 related events affecting prices for a range of goods and services purchased by households, the ABS has assessed each of the 55 seasonally adjusted series to determine the appropriate treatment.

Consistent with other economic statistics, for the calculation of seasonally adjusted movements, the ABS has decided to move to forward factors (published in Table 14 of the CPI) for any CPI series significantly affected for a potentially prolonged period. In these cases, the use of forward factors will replace the standard concurrent adjustment.

The use of forward factors retains the pre-COVID-19 seasonal factors and allows time for an assessment of the full impact. Forward factors will be used in the June 2020 quarter for these CPI series:
  • Child care
  • Medical and hospital services
  • Preschool and primary education

All other CPI seasonally adjusted series will remain concurrently adjusted.

For the CPI imputed series discussed in this article, a consistent approach will be used for the All groups CPI seasonally adjusted series. This means that for those series shown in table 1 above as being imputed off headline CPI in original terms, these series will be re-imputed off the aggregate seasonally adjusted level.


Footnotes

1. The ABS conducted extensive consultation on this topic with a number of National Statistical Organisations and Professor Kevin Fox, UNSW, and would like to thank them for their contributions.


Appendix 1 - review of literature and response by other National Statistical Organisations

Country/author
Agency
Imputation approach
Reference
EuropeEurostatHeadline CPI for non-seasonal products, and typical seasonal movement for seasonal products.Guidance on the compilation of the CPI in the context of COVID-19 crisis
New ZealandStats NZHeadline CPIImpacts of the COVID-19 lockdown on methodology for food price index April 2020
CanadaStatistics CanadaHeadline CPITechnical Supplement for the April 2020 CPI
USBLSStandard imputationEffects of COVID-19 Pandemic and Response on CPI
BelgiumStatistics BelgiumCarry forward for non-seasonal products and typical seasonal movement for seasonal productsImpact of the COVID-19 on the measurement of inflation
FranceINSEESame as EurostatHow to compute a CPI in the context of the Covid-19 crisis?
ItalyIStatHeadline CPIThe Covid-19 crisis and the compilation of CPIs
NetherlandsCBSSame as EurostatImpact of imputation methods on the CPI
NorwayStatistics NorwaySame as EurostatCorona consequences for CPI
United KingdomONSHeadline CPICoronavirus and the effects on UK prices
Erwin Diewert and Kevin FoxNBER working paperHeadline CPIMeasuring Real Consumption and CPI Bias Under Lockdown Conditions