METHODS AND PRACTICAL CONSIDERATIONS


INTRODUCTION

2.1 The majority of HFCE data align seamlessly into the CPI weight structure, however some challenges do exist due to conceptual, data source and coverage differences. These challenges arise because the CPI and HFCE estimates are produced for different purposes and use different data sources. This chapter outlines these challenges and the proposed treatments by the ABS. Empirical results are presented in chapter 3.


OVERARCHING CHALLENGES

(a) Classification of expenditure

2.2 Classifications are used by the ABS to group data collected from survey respondents (e.g. businesses, households) in a structured and consistent way. Classifications are developed to support the purpose and outputs of a specific set of statistics. In the case of household expenditure, various classifications are available and used by the ABS, depending on the purpose. As different classifications are used, there is a need to map the data collected under one classification to the other classification to make the data meaningful for each use, so concordance mapping is needed.

2.3 Information collected through the HES is classified according to the HEC, which is part of the Household Expenditure Survey and Survey of Income and Housing, User Guide, Australia, 2009-10 (cat. no. 6503.0). As the Australian CPI is classified according to the CPICC, found in Consumer Price Index Commodity Classification, Australia, 16th Series, 2011 (cat. no. 6401.0.55.004), a concordance mapping between HEC and CPICC is necessary to implement household expenditure data for CPI purposes. In total, there are 700 individual HEC codes that are used to map household expenditure to 87 ECs.

2.4 As HFCE covers a wider range of goods and services, it is further dissected by the purpose for which households engage in the transactions using the Classification of Individual Consumption by Purpose (COICOP). HFCE aligns as far as possible to COICOP, however, there are instances where it is not yet possible for Australia to follow all of COICOP's recommendations.

2.5 The CPI and HFCE collect and publish data using different classifications. The availability of classification concordances allows expenditure to be mapped between HFCE and CPI. As part of the annual Australian System of National Accounts (cat. no. 5204.0), Supply-Use tables enable expenditure that is more detailed than expenditure aggregates defined by COICOP categories. Supply-Use data can be further disaggregated based on the Input-Output Product Classification (IOPC). IOPC to CPICC mapping can take place based on the availability of a classification concordance, found in Australian National Accounts: Input-Output Tables, Product Details (cat. no. 5215.0.55.001).

2.6 Mapping the HFCE data from IOPC to CPICC sometimes requires further disaggregation with expenditure of one IOPC pertaining to multiple CPICCs. In these instances, expenditure is further disaggregated using supplementary data sources. In general, HFCE expenditures can be apportioned as proportions, defined as follows:

if the IOPC does not map to an EC

if the IOPC maps to only one EC

if the IOPC maps to multiple ECs (2.1)

Where;

e_i = representative expenditure of EC i mapped to an IOPC (derived from alternative data sources)
e_j = representative expenditure of ECs j mapped to an IOPC (derived from alternative data sources)_.

(b) Scope - Exclusion of expenditure by Not for Profit Institutions Serving Households (NPISH)

2.7 HFCE estimates include expenditure that is excluded from the CPI. The challenge is to find methods to exclude this expenditure so HFCE can be used for CPI weighting purposes. This sub-chapter details the ABS approach to estimate and then remove the NPISH contribution to HFCE for each CPI EC.

2.8 NPISH "consist of non-market non-profit institutions (NPI) that are not controlled by government" (UNSD 2008, p.74, paragraph 4.93). Unlike other NPIs, which are treated as corporations, NPISH can and do engage in final consumption expenditure in the form of goods and services provided to households free or at prices that are not economically significant. This does not mean that NPISH cannot produce market output but the majority of their costs of production are devoted to the provision of non-market output (UNSD 2008). At present, expenditure estimates for the NPISH units are published together with those for the household sector in HFCE, and cannot be explicitly identified. Examples of NPISH units include (but are not limited to) charities, aid agencies, religious institutions, cultural clubs and relief agencies.

2.9 As NPISH are not within the scope of the Australian CPI (it does not constitute expenditure by households) their expenditure must be removed from HFCE data in order to derive conceptually correct weights for CPI use.

2.10 Expenditure estimates of NPISH are currently explicitly compiled every six years coinciding with the HES. The contribution of NPISH to the CPI basket is small, with expenditure representing approximately 2.0% of the total HFCE that is included in the CPI basket. In order to remove NPISH expenditure on an annual basis, it is suggested NPISH is estimated as a proportion of total HFCE for each IOPC component, where the ratios are based on the closest available NPISH expenditure. This approach can be summarised as follows:

(2.2)

Where;

= NPISH expenditure at period t for IOPC i
= HFCE expenditure at period t for IOPC i
= NPISH expenditure in weight reference year for IOPC i (i.e. HES year)
= HFCE expenditure in weight reference year for IOPC i (i.e. HES year).

The estimates for various IOPC components are then excluded from the HFCE (including NPISH) to obtain HFCE excluding NPISH estimates as follows:

(2.3)


(c) Coverage - Metropolitan households and selected living cost indexes

2.11 In terms of geographic coverage, the Australian CPI is compiled separately for the eight Australian capital cities. Each capital city's index is weighted to produce the index for the eight capital cities, which is considered the equivalent of a 'national' price index. It is estimated that the geographic coverage of the CPI covers approximately 64% of the Australian population (ABS 2011). Ideally, the Australian CPI would encompass all Australian households (like HFCE) however practical implementation is hindered by the additional costs associated with collecting prices outside capital cities.

2.12 The CPI differs in coverage to HFCE data, which is compiled at a National level. The analysis conducted in this paper derives CPI expenditure weights at the National level using HFCE data directly with further work being undertaken to remove non-metropolitan expenditure to align with the CPI. In order to utilise national HFCE data for CPI weighting purposes at the capital city level, a pro-rata approach is suggested to overcome the practical differences in geographic coverage. This proposal makes use of price updated expenditure aggregates for each capital city when distributing the annually re-weighted eight capitals link period expenditure aggregates. Further details on this method are included in Appendix 2. There would also be opportunities to adjust for household demographic changes within each capital city over time however, this paper does not explore this concept.

2.13 The Selected Living Cost Indexes (SLCIs) (cat. no. 6467.0) publish price indexes for five main sub-populations (e.g. Pensioner and Beneficiary, Employee, Age pensioner, Other Government Transfer Recipient and Self-funded Retiree) of Australian household. Expenditure weights for the SLCIs are currently derived from the HES, and categorised based on the principal source of household income. Within HFCE, it is not possible to identify the income characteristics of households necessary to derive sub-population weights. As a result, this paper does not further investigate the frequency of updating weights for SLCIs.


(d) Revisions - Impact of revisions of HFCE data on CPI EC weights

2.14 The headline measure for the CPI is the original series and it is only revised in exceptional circumstances. National Accounts' annual HFCE data undergoes annual cyclical revisions, where preliminary estimates can be revised up to twice after release and periodic historical revisions can take place, spanning the entire Supply-Use series. This occurs when there are significant statistical developments or when more complete benchmark data - such as the HES - becomes available (paragraphs 3.5 and 3.6 below summarise HFCE vintages and normal cyclical revisions).

2.15 In order to quantify the impact of HFCE revisions on CPI weights, analysis has been undertaken to investigate the impact of using HFCE data from preliminary (t-1), revised (t-2) and final (t-3) estimates. The impact of historical revisions is also analysed, with empirical results for all types of revisions provided below in chapter 3.


SPECIFIC EC CHALLENGES

2.16 The vast majority of HFCE estimates are fit for CPI weighting purposes. This sub-chapter outlines the relatively small number of CPI ECs where additional action is needed to address specific EC challenges.


(a) New dwelling purchase by owner-occupiers EC

2.17 In the CPI, New dwelling purchase by owner-occupiers EC refers to the cost of net additions of household sector dwellings as a measure of owner-occupier housing costs, and includes new homes (excluding land) and major improvements. Specifically, expenditure on New dwelling purchase by owner occupiers EC comprises of four components; owner-occupied housing (OOH) costs, first home buyers grants (FHOGs), alterations and additions, and installed appliances.

2.18 While the HES is available as a data source, too few transactions are reported to produce reliable results, and it also includes the land component which is out of scope when using an acquisitions approach(footnote 1) to compiling the CPI. As a result, alternative data sources have typically been used to supplement the HES for this EC.

2.19 In the National Accounts, persons who own the dwellings in which they live are treated as owning unincorporated enterprises that produce housing services that are consumed by the household to which the owner belongs (UNSD 2008, p. 187, paragraph. 9.65). As the National Accounts separate the ownership of dwellings from the household sector they create notional transactions between households as landlords and tenants. The housing services produced are deemed to be equal in value to the rentals that would be paid on the market for accommodation of the same size, quality and type, and they impute values of the housing services (i.e. imputed rents) and record these as household final consumption expenditures of the owners.

2.20 The conceptual basis for the acquisitions approach to OOH applied in the CPI is based on the premise that OOH has a dual nature - part consumable and part asset. As a result of this conceptual difference OOH will continue to be derived using methods outlined in 16th Series CPI review, with changes to the data sources proposed in this study.

2.21 To derive the OOH component in the CPI, expenditure is comprised of a price (p) and quantity (q) dimension, where p represents the average price of dwellings for that period and q the net additions of new owner-occupied dwellings. To derive estimates for p and q, two data sources have been identified to align the derived expenditure with the CPI concepts.

2.22 The price dimension can be estimated through the average value of private dwelling completions by State, published in Building Activity, Australia (cat. no. 8752.0).

2.23 The quantity dimension can be estimated from the change in dwelling stock used in the compilation of HFCE. Dwelling stocks are adjusted to take into account other changes, for example demolitions, net conversions from commercial uses and dwelling completions not within scope (ABS 2014). The dwelling stock estimates are then separated as owner-occupiers and renters using data from the Census of Population and Housing. For non-Census periods, the dwelling stock is moved forward using the number of dwelling unit completions by State from the ABS' Building Activity, Australia (cat. no. 8752.0).

2.24 In line with standard practice in the CPI relating to the inclusion of subsidies such as the FHOG, they are treated as negative expenditure and subtracted from the cost of the new dwelling purchase.

2.25 Expenditure on other items that are included in new dwelling purchase by owner-occupiers includes alterations and additions and installed appliances. HFCE data does not include expenditure on alterations and additions, as it is considered a fixed asset rather than consumption by households (ABS 2014). This expenditure data is available as part of the National Accounts alterations and additions component of private gross fixed capital formation (GFCF). This component uses building activity survey data adjusted for under coverage as the survey data only covers approved work valued above $10,000.

2.26 Expenditure on installed appliances is included as part of HFCE and is directly included in new dwelling purchases by owner-occupiers.

Hence, total expenditure on new dwelling purchase of owner occupiers will be derived as:

(2.4)

Where;

annual average price and quantity of completed private residential dwellings at period t
value of First Home Owners Grants at period t
private GFCF for alterations and additions at period t
HFCE expenditure on installed appliances at period t.

(b) Insurance EC

2.27 The Insurance EC covers comprehensive insurance for dwellings and motor vehicles, and compulsory third party motor vehicle insurance services in the CPI. Of particular interest is the insurance for dwellings component of the EC, which is excluded from HFCE as it is considered intermediate consumption in the National Accounts. This difference in scope between HFCE and CPI means an alternative expenditure data source is required, this being the intermediate consumption component of National Accounts, which aligns with CPI concepts (i.e. net insurance charge(footnote 2) ).


(c) Maintenance and repair of the dwelling EC

2.28 Within the CPI, the Maintenance and repair of the dwelling EC measures price change on goods (e.g. paint) and services (e.g. electrician) associated with the maintenance and repair of an owner occupied dwelling. The equivalent expenditure component is predominately excluded from HFCE as it is considered intermediate consumption in the National Accounts ownership of dwellings industry. There is an exception for certain goods (e.g. nails, bolts), which are included in HFCE, because they are considered to be associated with minor maintenance rather than capital improvements.

2.29 Expenditure estimates by households on maintenance and repair of the dwelling will be derived as the total expenditure recorded in both the HFCE and intermediate consumption component for repairs and maintenance. To summarise, Maintenance and repair of the dwelling EC can be expressed as follows:

(2.5)

Where;

expenditure on HFCE data component i mapped to the maintenance and repair of the dwelling in period t.
maintenance and repair component of Intermediate consumption of ownership of dwellings (OOD) for period t.

(d) Property rates and charges EC

2.30 Within the CPI, the Property rates and charges EC measures price change of general rates and garbage collection fees. This EC includes both state and local council property based rates and charges except water and sewerage, taking into account concessional and standard rates (ABS 2011). The equivalent expenditure component is predominantly excluded from HFCE, as municipal rates are considered as a tax on production. There is an exception for certain government regulatory services which are included in HFCE.

2.31 Expenditure estimates by households for the Property rates and charges EC will be derived as the total expenditure recorded in both the HFCE and municipal rates component derived from other taxes on production. To summarise, Property rates and charges EC can be expressed as follows:

(2.6)

Where;

expenditure on HFCE data component i mapped to the Property rates and charges EC in period t
Municipal rates household expenditure for period t.

(e) Other services in respect of motor vehicles EC

2.32 In the CPI, the Other services in respect of motor vehicles EC measures charges associated with the right to own and operate a vehicle on public roads. This includes items such as motor vehicle registration, roadworthiness tests, driver licence fees, parking fees, driving lessons and tollway charges. The equivalent expenditure component is predominantly excluded from HFCE, as vehicle registration and licences are considered as other current taxes on income, wealth etc.

2.33 Expenditure estimates by households on Other services in respect of motor vehicles EC will be derived as the total expenditure recorded in HFCE and other sources (i.e. for vehicles registration and licences). The vehicles registration and licences component will be derived from data supplied by the ABS Government Finance Statistics (GFS) section. To summarise, Other services in respect of motor vehicles EC can be expressed as follows:

(2.7)

Where;

expenditure on HFCE data component i mapped to the Other services in respect of motor vehicles EC in period t
household expenditure on vehicle registrations/licences derived from GFS data for period t.

(f) Other financial services EC

2.34 The Other financial services EC in the CPI covers the cost of those services acquired by households in selling or buying major assets, such as real estate fees and stamp duty fees. Expenditure weights for this EC are currently derived from various data sources (including HFCE).

2.35 The current data sources can be estimated on an annual basis and are maintained for all components of the Other financial services EC except for real estate services. The current CPI expenditure data for real estate services is derived using property transaction data and unpublished ABS survey data on real estate fees. It is proposed to derive expenditure directly from National Accounts' private GFCF ownership transfer cost real estate fees component. This data source uses a similar price and quantity method to estimate expenditure compared to methods used in previous CPI Series reviews. It is also preferred for this research as it increases the coherence in the methods and data sources used in macroeconomic statistics compilation.

1 The acquisitions approach defines the basket of goods and services as consisting of all those consumer goods and services actually acquired by households during the base period (ABS 2011). <back
2 Gross premiums less claims. <back