UTILISING HFCE DATA FOR ANNUAL RE-WEIGHTING
2.1 HES data provides weighting information for CPI and SLCI sub-populations, and will continue to be the principal data source for the CPI and SLCIs weights when they are updated for the reference period of the survey. The CPI population subgroups are the eight capital cities; the SLCI subgroups are Employee, Aged pensioner, Other government transfer recipients, and Self-funded retiree Households, as a weighted average of the eight capital cities. The next HES is scheduled to be conducted in 2021-22. Until then, the ABS will use HFCE data to annually re-weight the CPI and SLCIs. ABS research shows that annually re-weighting reduces the upward substitution bias that can emerge when expenditure weights are held fixed across a number of years.
2.2 Methods to align HFCE data to the CPI scope and concepts are outlined in ABS (2016), along with the approach to produce an annually re-weighted CPI. Since then, the ABS has further investigated two related topics to:
PRODUCING ANNUAL WEIGHTS FOR THE SLCIs
|(a) annually re-weighting the SLCIs; and |
(b) deriving capital city expenditure weights in inter-HES years.
2.3 The current expenditure weights for the SLCIs are derived from the HES. Weights for four household types are produced, based on the principal source of household income.
2.4 The SLCIs sub-populations are:
|(a) Employee households - those households whose principal source of income is from wages and salaries; |
(b) Age pensioner households - those households whose principal source of income is the age pension or veterans affairs pension;
(c) Other government transfer recipient households - those households whose principal source of income is a government pension or benefit other than the age pension or veterans affairs pension; and
(d) Self-funded retiree households - those households whose principal source of income is superannuation or property income and where the HES defined reference person is 'retired' (not in the labour force and over 55 years of age).
2.5 As is the case for the CPI, the SLCIs will be enhanced by more frequent weight updates. The key benefits from annually re-weighting the SLCIs include:
|(a) maintaining coherence between the CPI and the SLCIs; and |
(b) reducing substitution bias.
2.6 Annual re-weighting of the SLCIs utilising HFCE data requires a number of challenges to be resolved. The most significant challenge is that HFCE data for the sub-population is not available.
2.7 Following ABS investigations, the approach to annually re-weighting the SLCIs can be summarised as follows:
|(a) Align HFCE data with the scope and classifications of the SLCIs at a detailed level. |
This requires removal of some components of HFCE (e.g. expenditure by Non Profit Institutions Serving Households (NPISH)) to align with the SLCIs, and concordance of HFCE data to the SLCIs and to the CPI EC classification (ABS, 2016). This provides HFCE data for 2015-16 and 2016-17 for each EC of the SLCIs and CPI (with the exception of the mortgage interest EC in the SLCIs sub-populations, which is discussed in 2.9).
(b) Apply the movements in HFCE aligned data between 2015-16 and 2016-17 for each EC to update the expenditure weights (footnote 1) for the SLCIs sub-populations.
2.8 This approach assumes that annual movements in HFCE aligned data capture annual changes in expenditure in response to relative price changes or preferences across the four sub-populations. For example, if the relative price of beef increases, it is assumed that the four household sub-populations groups adjust their level of consumption and substitute to, for example, chicken or lamb, in a similar manner.
Special case - Housing (New dwelling purchase and Mortgage interest charges ECs)
2.9 As part of these investigations, the ABS addressed unique conceptual differences between the CPI and SLCIs. Specifically, within the Housing group, the New dwelling purchase by owner-occupiers EC measured in the CPI is based on the acquisitions approach, whereas the mortgage interest charges EC measured in the SLCIs is based on the outlays approach, see paragraph 3.5 (ABS, 2017b). The ABS has investigated this conceptual difference and determined that the scope and coverage of the dwelling interest charges from the household interest data in the Australian National Accounts (ABS, 2017c) aligns with the SLCIs mortgage interest charges EC. Therefore, National Accounts data will also be used to re-weight Mortgage interest charges in the SLCIs.
2.10 Empirical results of this methodology are summarised in section 3 of this paper.
PRODUCING ANNUAL WEIGHTS FOR THE CPI CAPITAL CITY INDEXES
2.11 National Accounts HFCE data is compiled at the national level (ABS, 2015a). This poses a challenge to derive expenditure weights for the individual capital cities. The ABS will address this challenge by:
|(a) Applying the movements in HFCE aligned data for each EC to update the weighted eight capital cities CPI. |
(b) Deriving HES based proportions for each capital city EC, using HES data of expenditure in each capital city to the weighted eight capital cities index for each EC.
(c) Applying these capital city proportions for each EC in (b), to the updated weighted eight capital cities index in (a) to produce capital city expenditure weights for each EC (footnote 2).
2.12 This approach assumes the following:
|(a) no relative change in the expenditure proportions of capital city households relative to the rest of state households; this is implicit in applying National HFCE aligned data to the weighted eight capital cities; and |
(b) for each capital city, ratios (in terms of underlying quantities) by EC are fixed for 6 years. This fixes the relative proportion of expenditure on, for example, milk by Sydney households relative to total milk expenditure by all capital cities.
The ABS will continue to monitor and test these assumptions each year. Where required, adjustments to capital city weights will occur by utilising alternative data and qualitative information to incorporate significant changes that apply to specific capital cities (e.g. a significant downturn in the housing market in one capital city).
1 Expenditure weights derived from the HES (in this case, HES 2015–16). <back
2 A detailed numerical example demonstrating this approach is presented in Appendix 2. <back