12 U2007 Socio-demographic Composition

socio-demographic composition — assessment results

1 This working paper describes the assessment for socio‑demographic composition (SDC) and the resulting factors. The development of the assessment method in the 2004 Review is discussed in Volume 7 of the 2004 Review Working Papers.

DESCRIPTION OF THE FACTOR

2 Socio-demographic composition disabilities reflect the effect on the cost of providing services of differences between States in the characteristics of their populations. More specifically, this disability factor measures the relative impact that differences between States in the characteristics of State populations (such as age, sex, Indigeniety, income and location) have on costs of State services because of differences in average use patterns and unit costs.

ASSESSMENT METHOD

3 The general approach to the factor is to identify the population groups for which services are provided and, if appropriate, sub-groups that use the services more (or less) than average and/or cost more (or less) than average per unit of service provided.

4 At its simplest, a State has a positive (or negative) disability if the share of its population in the user group exceeds (or is less than) the national share. The factor reflects an assumption of a direct link between the above (or below) average shares of State populations that potentially use the service and the per capita cost of providing it. Thus if the proportion of a State's population aged 65 and over exceeds the national proportion by 10 per cent, it is assumed, all else considered equal, that the State would incur per capita expenses on services for the aged that are 10 per cent above average.

5 The way the factor is calculated varies between categories. In cases where a single user group is identified, the factor is equal to the proportion of a State's population in that group compared with the national proportion. In other cases, where use varies among sub-groups of the relevant population (which is often the total population) or more resources are required to provide the service to some sub-groups, a more detailed approach is applied. That approach derives notional, cost weighted users for each State based on national average use rates and cost weights.

6 Attachment A contains the guidelines the Commission used when deciding whether a socio-demographic composition factor should be assessed. In brief, a factor was assessed if there was:

· a conceptual case;

· evidence to support that case;

· material differences between States; and

· a way of measuring this difference.

7 Where full national data sets on user groups, their use and unit costs were available, factors could be calculated directly using the available data. When less than full data were available, judgment was used to estimate missing data if the conceptual case was strong, the effects were material and sufficient data were available to allow estimates to be made with confidence.

Incorporating Australian average use rates

8 It was often the case that a part of the population eligible to receive the service (which may be the total population) uses the service more often than other population groups. The disability calculation allowed for this by including use weights. The calculation then proceeded in the following basic steps:

· determining the population sub‑groups relevant to the service;

· extracting population data (usually Census) for each sub‑group;

· estimating the number of users within each sub‑group (usually from administrative data sets);

· calculating Australian average use rates for each sub-group in the population (calculated as Australian total by dividing the number of users within each sub‑group by Australian total at population data (usually Census) for each sub‑group);

· applying the Australian average use rates (from above) to the population of each group in a State to calculate weighted populations (or service use population) for each State; and

· deriving disabilities by dividing each State's weighted population by its unweighted population and then comparing this proportion to the Australian proportion.

9 In many cases, there were several causes of differences in the use of services. For example, Indigeneity, income or location, as well as age, might affect use of a service. Where possible, weights for population groups with more than one common characteristic were calculated simultaneously (or jointly) to avoid double counting. For example, if the Commission decided that service use differed by Indigeneity and income then a separate use weight would be calculated for Indigenous low income people, non-Indigenous low income people, Indigenous high income people and non‑Indigenous high income people.

10 Under the 2004 Review method a weight was calculated for each population sub-group with common characteristics.

11 Calculations of Australian average use rates for different sub‑groups were usually well supported by data on service use. Where possible national data collections such as those published by ABS and the Australian Institute of Health and Welfare (AIHW) were used. In the absence of full data, partial information (say from one or more States or from related national collections) and judgment were used. The absence or unreliability of State or Australian Government administrative data on the use of services by different groups of people constrained the range of use effects that could be reflected in the factors.

12 Basing use weights on national averages aimed to ensure that any effects of an individual State's policy on use in that State, or anomalies in data at State level, did not affect the assessment.

13 The process of applying average use rates to disaggregated State population data requires State population to be disaggregated in a similar way. The ability to obtain Census data with many cross-classifications supported the calculations well and enabled them to be done in a way which avoided double counting of effects. Using a single data source (the Census) enabled consistency of population counts across all States and expense categories.

Incorporating Australian average unit cost rates

14 It is often the case that the States as a whole decide to devote more resources to some groups of people (such as people with low income or Indigenous people). It is also the case that providing a service to people with particular characteristics is more difficult and more expensive (such as people with low English fluency or Indigenous people) than providing the service to other people. These cost differences relate to the cost of providing each episode of service and are separate from any cost effects due to differences in the relative use of the service. The socio-demographic composition factors often allowed for the differences in cost per episode of service.

15 The factors were calculated by applying Australian average unit cost weights to State populations already weighted by average use rates (if differences had been observed and measured). These were disaggregated by the different characteristics. The per capita use and cost weighted population of each State relative to the Australian per capita use and cost weighted population gave the factor for each State.

16 The application of Australian average unit cost weights followed a process similar to that for Australian average use weights. The process involved:

· determining the population sub‑groups relevant to the service;

· extracting population data (usually Census) for each sub‑group;

· estimating the number of users within each sub‑group (usually from administrative data sets);

· estimating the amount/proportion of expenditure on each sub‑group;

· obtaining a average unit cost weight for each group (by dividing the amount/proportion of expenses on each sub-group by the number of users within each sub-group (usually from administrative data sets)) and implicitly expressing the average cost of providing a service to the chosen group of people as a proportion of the average cost of providing a service to all people (for example, a cost weight of 1.5 implied that for every $1 spent on the average user $1.50 needed to be spent on the high cost group);

· applying the average unit cost weight for each group to the service use population of each group in each State to create weighted populations for each State; and

· determining disabilities by dividing each State's weighted population by its unweighted population and then comparing this proportion to the Australian proportion.

17 Assessing Australian average unit cost weights thus required data that:

· demonstrated that particular sub‑groups of people had an effect on the cost of providing a service; and

· allowed the size of that sub‑group to be measured in each State.

18 There was a greater use of judgment by the Commission in setting Australian average unit cost weights. States often did not keep data on the extra resources they devoted to servicing some groups of people[1]. Thus, the Commission used cost data from only a few States, or based the Australian average weights on judgment informed by workplace visits, Commonwealth policy, academic studies, and other information such as the findings of the Indigenous Funding Inquiry. (For example, the extra costs in servicing remote Indigenous people were often based on data or evidence from States that had a significant number of remote Indigenous people in their population. The Australian average unit cost weight for humanitarian refugees was suggested by one State.)

19 Use and unit cost weights, or an overall cost weight incorporating use and unit cost impacts, have been applied in the calculation of the SDC disability factor.

The range of population characteristics reflected in disabilities

20 The major population characteristics the Commission recognised under the 2004 Review method as having an impact on the use and unit costs of providing services were:

· age and sex;

· socio-economic status, measured by:

(i) income: persons were considered to have low socio-economic status if:

- the income of the household in which they lived was less than $31 200; or

- their individual income was less than $20 800[2]; or

- they were on a social security benefit or pension;

(ii) concession holder status: Australian Government health care or pensioner card holder numbers were used in the concessions categories;

(iii) unemployment status; or

(iv) other indicators, such as people over 60 years who lived alone (this was used in the Aged care services category);

· location/region of residence — which recognised that the use of services varied depending on where people live. In some categories cost differences by location have been recognised in a socio‑demographic composition factor. Where this occurred the Commission endeavoured to make sure there was no double counting with other factors recognising cost differences, such as dispersion and input costs. For the 2004 Review method, the locational dimension was generally implemented using an Accessibility/Remoteness Index of Australia (ARIA)[3] type classification to define similar locations or regions within each State;

· cultural and linguistic background — four impacts were recognised for the 2004 Review method:

(i) differences in service use rates by people who have a non-English speaking background, measured using information from administrative data sets;

(ii) costs of translation services and extra servicing time, recognised by applying cost weights to people with low English fluency;

(iii) costs relating to cultural background, recognised by applying cost weights to people with culturally or linguistically different background[4]; and

(iv) costs relating to humanitarian migrants; and

· Indigeneity — for the 2005 Update method, four impacts were recognised:

(i) differences in the extent to which Indigenous people used particular services, usually based on administrative data and ABS experimental estimates of the Indigenous population;

(ii) cost weights to recognise that a unit of service cost more to provide to an Indigenous person for reasons associated with their Indigeneity;

(iii) costs relating to translation services and additional servicing time, recognised by applying cost weights to all people (including Indigenous people) who have low fluency in English; and

(iv) costs arising because services must be provided to people who live predominantly in accordance with traditional Indigenous cultures, which was proxied by applying an additional weight to remote Indigenous people.

21 Table 1 shows the characteristics considered in each category. The section on socio‑demographic composition in Volume 7 of the 2004 Review Working Papers contains a list of the cost and use weights applied in each category for each of the population characteristics. The table indicated whether the weight was based on data or judgment and the evidence on which the judgment was based.

Table 1 Socio-demographic characteristics for which disabilities were assessed — 2004 Review method

Age

Sex

Income

CALD

Indigeneity

Remote general

Remote Indigenous

Other

Hospital patient fees

Schools education

Pre-School

Government primary

a
b

Non-government primary

Government secondary

a
b

Non-government secondary

VET — Institutes

ü
d

Transport of Rural School Children Schools

Table 1 Socio-demographic characteristics for which disabilities were assessed — 2004 Review method (continued)

Age

Sex

Income

CALD

Indigeneity

Remote general

Remote Indigenous

Other

Inpatient services

Acute inpatient services

Non-acute inpatient services

Non inpatient services and Community Health

Emergency departments

Outpatients

Community Health

Population and Preventative Health

Breast Cancer Screening

Communicable Disease Control

Organised Immunisation

Other Public Health

Family and Child Services

Child and Youth Support Services

Child Care

Aged Care Services

Aged services

Disability services

Municipal Rate Concessions

Table 1 Socio-demographic characteristics for which disabilities were assessed — 2004 Review method (continued)

Age

Sex

Income

CALD

Indigeneity

Remote general

Remote Indigenous

Other

Homelessness and General Welfare

Other Welfare

SAAP

Housing

Private Rental and Home Purchase Assistance

Public Housing Maintenance

ü
g

Public Housing Management

ü g

User Charges

ü g

Services to Indigenous Communities

Police

Administration of Justice

Corrective Services

Community Based Corrections

Juvenile Detention

Prisons

National Parks and Wildlife Services

Electricity and Gas

Concessions Component

General Subsidies

Table 1 Socio-demographic characteristics for which disabilities were assessed — 2004 Review method (continued)

Age

Sex

Low income

CALD

Indigeneity

Remote general

Remote Indigenous

Other

Water, Sanitation and Protection of the Environment

Concessions Component

General Subsidies

Non-urban Transport

Concessions Component

General Subsidies

Urban Transit

Concessions Component

Pricing

General Public Services

Regulation and Planning Sevices

Ethnic Affairs Component

Depreciation

u = use weight c = cost weight uc = combined use cost weight

a Differential cost weights for cost weights for capital cities.

b Differential cost weight for humanitarian refugees.

c Differential cost weights based on State's assessed Government expenses.

d Differential use weight for labour force status.

e Differential use weight for students living within 60kms (Government primary students and 80kms (Government secondary students) from the nearest school.

f Differential use weight for single people.

g These factors are weighted by SARIA for urbanisation effects as well as remoteness.

h A weight of 1.1 was applied to the population of Sydney and Melbourne.

i Relevant population based on health care card and pensioner card holders.

j Cost weight based on the cost of delivering services to different geographical regions of the State.

k Cost weight based on size, location, water availability and water quality.

l Differential use weights for non-metro residents and concession travellers.

m Relevant population based on recipients of unemployment benefits (including CDEP).

n In the Depreciation assessment socio-demographic factors were assessed using a global approach. For the buildings and plant and equipment components of Depreciation a socio-demographic factor was assessed using broad functional groups — Education (27 per cent), Health (29 per cent), Law and Order (9 per cent) and other (35 per cent). The socio-demographic composition factor for the welfare housing component of Depreciation was assessed using the socio-demographic composition factor for housing depreciation, which is explained in the Housing working paper.

Results for 2005-06

22 Table 2 shows the categories in which a socio-demographic composition factor was assessed in the 2007 Update and the factor assessed for 2005-06.

Table 2 Socio-demographic composition factors, 2007 Update, 2005-06

Table 2 Socio-demographic composition factors, 2007 Update, 2005-06 (continued)

CONTRIBUTION TO GST REVENUE DISTRIBUTION

Effect of assessment on grants

23 Table 3 shows the category's contribution to the distribution of GST revenue and Health Care Grants (hereafter described as GST revenue) implied by the 2006 Update and the 2007 Update.

Table 3 Socio-demographic composition, effect of assessment on GST revenue distribution, 2006 Update and 2007 Update

(a) Assuming same pool and a constant population.

(b) This figure shows the change in the amount redistributed among the States between the 2006 Update and the 2007 Update. It does not necessary equal the difference in the total redistribution from EPC between the two inquires.

24 Compared with an equal per capita assessment, the 2007 Update assessment of socio‑demographic composition redistributed $2 040.2 million to Queensland, Western Australia, South Australia, Tasmania and the Northern Territory.

25 Table 4 shows the contribution of each socio‑demographic composition assessment to the distribution of GST Revenue for the 2007 Update.

Table 4 Socio-demographic composition, effect of assessment on GST revenue distribution by category, 2007 Update^(a)

Table 4 Socio-demographic composition, effect of assessment on GST revenue distribution by category, 2007 Update^(a)(continued)

(a) Excludes the following: Debt Charges (ACT Phase-in Factor and Quantity of Borrowing Factor), Superannuation (Accrued Expenses Factor and Historical Costs Factor), Schools (Non-government Cost Factor and Grade Cost Factor).

26 Socio‑demographic composition factors had a large impact on the distribution of GST revenue. This was because there were large differences between States in the proportion of high need and high cost population groups. The proportion of State populations for a number of the population characteristics taken into consideration by the Commission was shown in Table 5. Attachment B of the Relative Fiscal Capacity of the States 2007 contained further information on population location and composition.

27 The greatest difference between States was in regard to the proportion of Indigenous people and the proportion of people outside major cities. As a result, these two characteristics were the main socio-demographic drivers of differences in expense needs.

28 The Northern Territory received higher per capita GST revenue than other States as a result of socio‑demographic factors. It had a demographic composition that was vastly different to other States. The main differences were the high proportion of Indigenous people (27 per cent higher than the average) and people living in remote areas (16 per cent higher than the average). These two disabilities compounded as this high needs group was predominantly located in remote areas where costs per unit of service were generally higher. The Northern Territory also had the highest proportion of people with low English proficiency (high cost group), those aged 0-4 (higher use group for health services), people aged 5-14 (school aged population).

29 Queensland and Western Australia received a greater than average per capita grant for similar reasons to the Northern Territory. They too had an above average proportion of Indigenous people, those aged 0-4, and those aged 5‑14. The proportion that these groups made up of the total population was smaller in Queensland and Western Australia than in the Northern Territory, which was why the per capita GST revenue distribution to these two States is smaller. Queensland had a higher than average number of low income earners (high service use and cost group) and Western Australia a close to average proportion.

30 Tasmania and South Australia were assessed as having higher expense needs as a result of socio-demographic composition mainly as they had a higher than average number of low income earners and a higher than average proportion of people aged 65 and over (high use group for health services, aged and disability services, concessions). In addition, Tasmania had a higher than average proportion of its population living outside major cities, and a higher than average proportion of its population that was Indigenous.

31 Victoria, New South Wales and the ACT had a lower than average proportion of people with the socio-demographic characteristics for which expense needs were assessed. New South Wales and Victoria both had a higher than average proportion of people with low English proficiency and people aged 65 and over. They received more GST revenue as a result of these two characteristics. Victoria needed less than New South Wales mainly as Victoria had a lower proportion of Indigenous people and people outside major cities. There were small differences between the two in regard to other population characteristics with most of these differences meaning that New South Wales had greater needs relative to Victoria. The ACT was the most highly urbanised State and it had a lower than average proportion of Indigenous, low English proficiency, and low income people.

Table 5 Proportion of State populations by selected socio-demographic characteristic — 2001 Census

CHANGES SINCE THE 2006 UPDATE

Effect of assessment on the distribution of GST revenue

32 Table 6 shows the redistribution of GST revenue resulting from the assessments in the 2006 Update and the 2007 Update. It also shows the sources of change.

33 Changes in the distribution of GST revenue between the 2004 Review and the 2007 Update were brought about because the Commission:

· used revised financial data in the category standards and other revised data in factor calculations for the years 2001-02 to 2004-05 used in the 2006 Update; and

· replaced 2000-2001 category standards and factors with those of 2005-06 to move forward the five-year period on which GST revenue distribution were based. Moving the five-year period forward in this way ensures the assessments reflect recent trends in State demographic and economic circumstances on the relative costs of those services.

Table 6 Socio-Demographic Composition, effect of assessment on GST revenue distribution, 2006 Update and 2007 Update^(a)

				NSW	Vic		Qld	WA	SA	Tas	ACT	NT	Total redist'd
				$m	$m	$m		$m	$m	$m	$m	$m	$m
Redistribution from EPC resulting from the 2006 Update assessment^(b)
			Indigenous influences	-339.6	-742.9		261.6	211.9	-109.2	10.4	-30.2	738.0	1 222.0
			Other socio-demographic composition influences^(c)	-404.7	-458.8		474.5	82.6	169.9	141.0	-105.3	100.9	968.8
			Total	-744.3	-1 201.7		736.1	294.5	60.7	151.4	-135.5	838.9	2 190.8

Redistribution from EPC resulting from the 2007 Update assessment^(b)
		Indigenous influences		-364.5	-782.8		276.0	225.4	-105.4	9.8	-35.7	777.2	1 288.4
		Other socio-demographic composition influences^(c)		-340.7	-440.5		450.7	34.9	156.1	141.2	-99.3	97.5	880.4
		Total		-705.1	-1 223.4		726.7	260.3	50.7	151.0	-134.9	874.7	2 168.8

Total effect of revisions and updating
	Indigenous influences			-24.8	-39.9		14.3	13.5	3.8	-0.6	-5.5	39.2	70.8
	Other socio-demographic composition influences^(c)			64.0	18.3		-23.8	-47.6	-13.8	0.1	6.1	-3.3	88.5
	Total			39.2	-21.6		-9.4	-34.2	-9.9	-0.5	0.6	35.9	159.4

(b) Assuming same pool and a constant population.

(d) This figure shows the change in the amount redistributed among the States between the 2006 Update and the 2007 Update. It does not necessarily equal the difference in the total redistribution from EPC between the two inquiries.

Changes due to revising average expenses and factors for years 2000-01 to 2004-05

34 Revising average expenses. Downward revisions were made to average expenses, producing a decrease in the redistribution of GST revenue between the States ($159.4 million). This change in socio-demographic composition has the largest effect on education (particularly the costs of running government primary and secondary schools) and health services (particularly hospitals).

35 Revising factors. Compared with an equal per capita assessment, the 2007 Update redistributed $2 168.8 million away from New South Wales, Victoria and the ACT. The changes to socio-demographic composition factors were small and were noted for various categories.

Changes in State circumstances — replacing 2000-01 with 2005-06 data

36 The key drivers of this change were:

· changes in the pattern of service use and cost in Inpatients and schools; and

· changes in the category standard for Services to Indigenous communities.

37 The socio-demographic composition of the population is the major disability in various categories assessment. It redistributed GST revenue to Queensland, Western Australia, South Australia, Tasmania and the Northern Territory.

This chapter was prepared by the Law and Order section of the Commonwealth Grants Commission. If you have any questions about its content please contact Marc Boisseau on (02) 6229 8889 or marc.boisseau@cgc.gov.au.

39 Date:17/2/07

ATTACHMENT A

GUIDELINES FOR ASSESSING A SOCIO-DEMOGRAPHIC COMPOSITION FACTOR

40 A disability was assessed when:

· a conceptual basis for the existence of an SDC disability is established; and

· either

(i) there is empirical evidence that differences in the SDC of States' populations have impacts on the costs of provision of State government services; or

(ii) there is incomplete evidence that additional use and/or costs are incurred due to SDC influences, but the 'in-principle' case is strong;

· the SDC disability was material; and

· the margin of error associated with the assessment of the disability was acceptably small.

41 Ideally, assessing SDC factors requires data on:

· the number of people in particular socio-demographic groups;

· the utilisation of a service by those population groups; and

· the unit costs of providing the service to those groups.

Unfortunately, such complete data sets rarely exist. However, it is not intended that factors be assessed only where data are complete.

42 The Commission made judgments about the quality of the information available, how representative of the situation for a socio-demographic group across all States it might be, and whether its use would move States' grant shares in the expected direction. Where judgments were made on these issues, the Commission has explained why (or how) it reached its conclusions.

[1] It is also sometimes the case that a plausible argument can be made that servicing some people would cost more but States cannot demonstrate that they provide those services.

[2] These are the upper limits of income band in which the second income quintile fell in the 2001 Census.

[3] The National Key Centre for Social Applications of Geographical Information Systems (GISCA) at the University of Adelaide has developed a range of such indexes. The Department of Health and Aged Care sponsored the development of the original ARIA index. The ABS uses ARIA+ (a national index) in its Australian Standard Geographical Classification. GISCA has also developed a State-based version for the Commission (SARIA).

[4] The number of people to fit this description can be defined in a number of different ways, including: (i) people born in non-English speaking countries; (ii) people with low-English fluency; or (iii) humanitarian migrants.

[return to top]