3112.0 - Demography Working Paper 1996/4 - When ERPs aren't Enough, 1996
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 19/05/1999
Page tools:
 Print Page
 Print All
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Demography Working Paper 96/4
COMMENTS SOUGHT In the 1996 Review of Demography Statistics the Australian Bureau of Statistics (ABS) undertook to prepare an issues paper covering service population estimates. This paper discusses demand for population estimates which are not exclusively (or even necessarily) based on 'place of usual residence' and the statistical issues associated with their estimation. Currently ABS uses the former as the conceptual basis for population estimates. Service population estimates may include both usual residents and non-usual residents. ABS welcomes any comments you may wish to make about this paper. ABS is also seeking expressions of interest from organisations prepared to contribute to a pilot study of service population estimates for their own purposes. The study would probably involve use, for the most part, of existing data sources and have as its objective the preparation of service population estimates which meet user specified requirements in relation to data quality, cost and timeliness considerations. Written comments should be addressed to John Paice Director Demography Australian Bureau of Statistics P.O. Box 10 Belconnen ACT 2616 email: john.paice@abs.gov.au Telephone comments and request for further information should be made to Katrina Phelan on (06) 252 6573. 1 EXECUTIVE SUMMARY 1 This paper reports on user needs and the conceptual, methodological and reporting load issues involved in estimating service populations. It has been prepared in response to the Australian Bureau of Statistics (ABS) 1996 Review of Demography Statistics, with the intention that the paper will promote discussion about the issues raised within. 2 The term service populations refers to those persons who demand goods or services from providers of such commodities. Such persons may be permanent or temporary residents of the area from which the service is sought, or they may be daytime visitors (including commuters), overnight or short-term visitors to the area. 3 Nearly 40 percent of agencies which responded to the 1996 Demography Review specified that estimates of resident populations (ERPs), as currently prepared by the ABS, do not meet their population needs since services can also be demanded by persons not resident to an area. 4 It is not surprising that the Review concluded that support for service population estimation is strong and widespread given the expansion of the service industry sector over the past twenty years (as measured by employment growth). Such support is also consistent with informal comments made by a variety of population users, especially over the last decade, about limitations associated with ERPs. 5 Whilst population estimates based on place of 'usual residence' are conceptually sound and are favoured over 'place of enumeration' estimates by many international statistical agencies, the relevance of 'usual residence' based estimates to some users is limited by the level of population mobility hidden within these estimates. Concerned users are therefore seeking a supplementary series of population estimates to ERPs. 6 Arguably the most significant issue associated with service population estimation is the lack of uniformity between service providers and the products they deliver. Such diversity makes the topic of service populations complex. Specifically, considerable variability exists in the demographic and geographic characteristics of the target populations to which different services are intended. 7 In generalised terms, however, service populations can be defined as consisting of resident populations (or parts thereof) and or non-resident populations (or parts there of) for a defined geographic area. This generalised definition illustrates that service providers' needs are likely to be better served if components of resident and non-resident populations can be estimated separately. 8 The types of non-resident populations of interest to service providers includes commuter populations, daytrippers, tourists and other overnight visitors (for one or many nights) and a diverse range of temporary residents (e.g. seasonal employees such as fruit pickers; cyclical employees such as fly-in/fly-out mine workers; winter-time visitors to sunbelt zones and residents of Aboriginal outstations in the dry season). 9 It is anticipated that the service population needs of some service providers will be so specific as to have little or no generic applicability to others (for example, persons demanding particular health care services). Identification of such service populations is likely to be possible only by use of datasets which relate specifically to provision of those services, such as associated administrative datasets. By contrast, the needs of a variety of service providers are expected to be assisted if users can gain access to a range of non-resident population estimates which have been prepared from generic datasets. 10 Access to generic non-resident population estimates is expected to assist in the creation of user specific service population definitions, by allowing individual users to combine none, some, or all of the component non-resident population estimates with resident population estimates (ERPs) (or parts thereof). 11 Thus adoption of a component approach to non-resident population estimation could maximise the possibility of users' diverse needs being met. Furthermore, a component approach would allow estimation of specific types of non-resident populations to commence once data sources, which meet user requirements, are identified without need to wait until data for all types of non-resident populations are identified. In general terms, a component approach offers the advantage of allowing non-resident population types to be estimated in accordance with a range of priorities, including timeliness, data access, user demand and cost considerations. 12 The success of generic estimates will, however, depend on how closely data definitions associated with such estimates can meet individual user requirements. Some user needs, for example, are expected to allow for data definitions which do not strictly observe the statistical classification principle of mutual exclusivity. In practical terms, this could be very important as use of existing data sources may make it difficult to create data definitions for a variety of non-resident population types which are mutually exclusive. 13 A brief description of existing datasets which may prove useful in preparing generic non-resident population estimates is provided in Appendix A. Use of existing data sources to prepare estimates of non-resident populations has a number of advantages over development of new data collections, including immediate access to data (possibly even time series data), lack of duplication of effort, reduced respondent burden and reduced collection costs. But such advantages are tempered by the ability of existing data definitions to meet users needs. 14 Thus the data definitions associated with specific datasets (especially in relation to time, purpose-of-movement, length of stay and geographic criteria) would need to be assessed in accordance with a variety of users' needs to ascertain the value of preparing generic non-resident population estimates. 15 While geographic issues which relate specifically to service population estimates would need to be addressed as part of data definitional considerations, many of the geographic needs of service population users are also expected to be assessed as part of the new standard spatial boundaries being developed in response to the current ABS Review of the Australian Standard Geographical Classification (ASGC). The benefits of the ASGC boundary revisions will not be available to service population estimates before their intended introduction in mid 1998, with their actual implementation expected to be staggered across specific ABS collections. 16 Frequency of estimation issues, in terms of user needs and data quality considerations, also impact on non-resident population estimation. In particular, seasonality is expected to be complicated because mobility patterns are likely to vary for different types of non-resident populations. For example, seasons could range from being eight hourly divisions of a 24 hour period for commuter populations to weekly (or monthly) divisions of a year for itinerant agricultural workers. 17 Costs associated with preparing service population estimates are expected to be considerable even if data from existing data sources are used. In particular, gaining access to a variety of datasets (possibly managed by a number of agencies) and extracting and maintaining data from such datasets, are expected to involve a substantial commitment in terms of human and computing resources. 18 Preparation of service population estimates which meet specific user needs (either for an individual or a variety of population users) require that factors such as purpose of use, quality, feasibility and cost be discussed and agreed on by the estimation agency and the client. 19 This research paper has been prepared to assist such discussions. 2 INTRODUCTION 20 The objective of this paper is to report on user needs and the conceptual, methodological and reporting load issues involved in estimating service populations. This paper has been prepared in response to the 1996 Review of Demography Statistics (ABS 1996a, 12-13) and with the intention that it will promote discussion about the issues raised within. 21 The term service populations refers to those persons who demand goods or services from providers of such products. Such persons may be permanent or temporary residents of the area from which the service is sought, or they may be daytime visitors (including commuters), overnight or short-term visitors. 22 Unless otherwise specified, for convenience considerations the terms usual resident and resident are used throughout this paper to denote the same concept, namely that of a person who has lived or intends to live at a specified residential address for a period of six months or more. This is the usual residence definition used by the Australian Bureau of Statistics (ABS) in the Australian Censuses of Population and Housing (ABS 1996c, 134). Similarly, the terms non-usual resident and non-resident are used interchangeably in this paper. 23 Demand for service population estimates was canvassed as part of the ABS's 1996 Review of Demographic Statistics. Nearly 40 percent of agencies which responded to this review specified that estimates of resident populations (ERPs), as currently prepared by the ABS, do not meet their population needs since services can also be demanded by persons not usually resident to an area (ABS 1996a, 1 & 13). 24 User consultation was the principal research source used in preparing this paper. This was necessary because although research into the effect on service provision for specific types of non-resident populations is quite extensive (especially for tourists, commuters and retirees with regard to sunbelt migration), the topic of service population estimation for a variety of non-resident population types is not well documented in the demographic literature. Many government agencies are now devoting considerable resources to researching and reporting on their specific service populations. Two notable examples are the Australian Institute of Health Welfare (Madden et. al. 1996) and the Western Australian Local Government Grants Commission (WALGGC 1995). Similarly, a recent paper from the Centre for Aboriginal Economic Policy Research (CAEPR) on methods of estimating Indigenous service populations is expected to assist in fine-tuning service delivery provided by the Aboriginal and Torres Strait Islander Commission (ATSIC) as well as by Commonwealth, State and Territory agencies (Taylor 1996). 25 In an effort to consolidate users' comments received as part of the 1996 Review of Demography Statistics, additional comments about the service population needs of each of the State and Northern Territory Local Government Grants Commissions were sought. User comments reveal the topic of service population to be complex because of variations between service providers and the products they deliver. In population terms, considerable diversity exists in the demographic and geographic characteristics of the service populations to which different services are intended. 26 The impact of such differences on users' needs and the conceptual and practical considerations associated with preparing service population estimates are discussed in Sections 4 and 5 of this paper. This paper also provides some background details about service population estimation in relation to the evolution of population estimation in Australia (refer Section 3). A brief description of existing datasets which may prove useful in preparing service population estimates is provided in Appendix A while Appendix B contains an extract copy of respondent and ABS comments about service population estimates, from the Review of Demography Statistics 1996 report. 27 In preparing this paper, the knowledge and co-operation of many ABS staff and users of ABS population data have been sought. In particular, the assistance of officers from the States and Northern Territory Local Government Grant Commissions is gratefully acknowledged. The views expressed in this paper, however, are principally the author's and do not necessarily reflect those of the ABS or other agencies contacted. 3 BACKGROUND 28 During the late 1970s and early 1980s, the ABS undertook considerable work to change Australia's official population estimates from the historical base of 'place of enumeration' to a 'place of usual residence' basis. This change was undertaken to allow Estimated Resident Populations (ERPs) to be produced for both census times and intercensally. User consultations undertaken by the ABS in the late 1970's confirmed that ERPs would better serve the statistical needs of the majority of ABS population users. In addition, preparation of ERPs would allow the ABS to discontinue the practice (since 1966), of producing postcensal estimates on a hybrid of 'place of enumeration' and 'place of usual residence' base. From the census date 30 June 1981, ABS population estimates have been prepared on a 'place of usual resident' base (ABS 1983, 1-6 ). Revised 'place of usual resident' estimates have been calculated back to 1971. 29 Research into service populations can, in one respect, be viewed as something of a retrograde step in the evolution of population estimation. Specifically, since service population estimates incorporate a combination of resident and non-resident populations, it has been necessary to revisit many of the issues addressed as part of the research undertaken in the late 1970's (ABS (Choi) 1979). Furthermore, it is possible that population estimates based on 'place of enumeration' rather than 'place of usual residence' may better meet some service providers' population estimation needs. 30 It must be emphasised that population estimates based on place of 'usual residence' are conceptually sound and are favoured over 'place of enumeration' estimates by many international statistical agencies (Shryock and Siegel 1971, 93-94). The relevance of 'usual residence' estimates for specific applications, however, is limited by the level of population mobility hidden within these estimates. For example, respondents to the 1996 Review of Demography Statistics considered ERPs to be particularly deficient as estimates of service populations for the high mobility areas of tourist regions, mining towns, seasonal employment areas and central business districts within major urban centres (ABS 1996a, 10). 31 Demand for service population estimates can, therefore, be considered to reflect demand for estimates of population mobility not currently met by censal and intercensal estimates. As explained in paragraph 28, since 1981, censal and intercensal estimates of population mobility have been designed to principally estimate population movement between places of usual residence. Specifically, census estimates of population mobility reflect changes in place of usual residence as recorded at the census date and one year and five years ago. Similarly, intercensal estimates (prepared quarterly for the States and Territories and effectively prepared annually for sub-State levels) relate mostly to usual residence mobility, although present data limitations may result in the capture of some non- resident migration (refer paragraph 56 for more details). 32 Censal and intercensal estimates understate usual residence mobility for persons who undertake multiple migrations during the reference periods, especially those who return to their previous address within one reference period. In addition, census estimates understate usual residence mobility for persons who are absent from the census (either because they were overseas at the time of the census or had died sometime during the intervening five year period) and for persons who provide incomplete usual residence address details (either for past or present addresses). 33 In addition to these usual residence mobility limitations, intercensal estimates of population migration do not provide comprehensive estimates of mobility to or between non- usual residence locations (such as work sites, education facilities, creches, amusement parks and other recreational facilities, hospitals and other short-term or seasonal accommodation sites such as motels and other types of commercial accommodation, private holiday houses, shearers' quarters and Aboriginal community outstations). 34 By comparison, it is anticipated that existing census data, especially 'place of enumeration' data and 'place of employment' details may assist with estimation of some specific types of non-usual residence mobility. Further details about the census and other possible data sources for service population estimation are provided in Appendix A. A detailed discussion of user' needs with respect to service population estimates is provided in Section 4. 4 USER NEEDS 35 The table below shows employment changes in selected industries between 1971 and 1991. These figures highlight the significant shift toward employment in service related sectors and by comparison, the significant shift away from employment in the agriculture and manufacturing sectors, during this twenty year period. TABLE 1 PERCENTAGE SHARE OF EMPLOYMENT IN SELECTED INDUSTRIES AUSTRALIA
36 Although some employment declines can be expected to have occurred from technological substitution, changes in employment patterns are generally considered good indicators of variations in the relative importance of industries to the economy as a whole. Thus, it is not surprising that the 1996 Review of Demographic Statistics concluded that support for service population estimates is strong and widespread (ABS 1996a, 12). Of the 73 organisations which responded to this Review, 27 (i.e. nearly 40 percent) expressed support for access to service population estimates. Estimates of resident populations were described as being insufficient for their population needs because non-residents were also known to demand their services. In particular, respondents considered ERPs to be especially poor indicators of service demand for employment related services, health care services, child care, transportation, water and sewerage (ABS 1996a, 10). 37 Support for service population estimation was received from a diverse range of organisations but broadly speaking, each of these organisations can be described as being either directly or indirectly responsible for provision of services. For example, government agencies such as the NSW Department of Bush Fire Services, Tourism NSW, the NSW Environmental Protection Agency, Sydney Water, the Victorian Department of Sport and Recreation, the South Australian Police and most State Treasuries and Health and Community Services authorities, advocated estimation of service populations (ABS 1996a, 12). The diversity of comments received as part of the Demography Review provided a rich source for identifying user needs as well as many of the conceptual issues associated with service population estimation. 38 In general, government agencies are seeking service population estimates to assist with policy and planning activities (including location of service centres) and the estimation of costs (resource and environmental) associated with service provision. Users' articulated a need for estimates of the number persons who seek services within a geographic area, especially those who are not usually resident in the area. While employment was cited as the principal perceived reason for service usage by non-residents, a variety of non-resident populations are of interest to these organisations and include commuter populations, daytrippers, tourists and other overnight visitors and a diverse range of temporary residents (e.g. seasonal employees such as fruit pickers; cyclical employees such as fly-in/fly-out mine workers; winter-time visitors to sunbelt zones and residents of Aboriginal outstations in the dry season). Few respondents specified a requirement for details such as age and sex but it is anticipated average length of stay data would assist many to estimate the impact of non- resident populations on demand for their services. With regard to geographic areas of interest, users' needs were especially varied. 39 Comments additional to those received from the Demography Review were sought from each of the State and Northern Territory Local Government Grants Commissions. The Commissions are responsible for recommending the distribution of financial assistance grants between local councils. Funds are allocated on service provision equalisation criterion (WALGGC 1995, 1-4) and therefore knowledge of the Commissions' service population needs was expected to assist identification of service population needs in general. 40 In actuality, the needs of each Commission proved to be as diverse as those articulated by the variety of service providers who responded to the Demography Review. The Commissions' preferences can be summarised as a need for estimates of non-resident populations to augment resident population estimates (ERPs) for Local Government Areas and Community Government Councils (split between community centres and outstations). Access to both resident and non-resident population estimates is sought to assist with existing funding allocation models. The Commissions are seeking access to independently prepared non-resident population estimates (for commuters and other daytrippers, tourists and an array of temporary or seasonal residents as specified in paragraph 38) with the objective of enabling average annual population estimates to be calculated. As with the Demography Review responses, although the Commissions specified a need for population numbers only, average length of stay details are also expected to be necessary to allow annual averages to be estimated (this issue is discussed further in the Data Definitions section, refer paragraph 54). 41 When combined, users' comments confirm that the topic of service populations is made complex by the diversity of service population definitions sought by service providers. Typically, disparate definitions are sought because service providers can be responsible for dissimilar geographic regions and for different population subgroups. As a consequence, it is unlikely that a single series of service population estimates will be able to be defined which will meet the needs of all service providers. 42 In generalised terms, service populations can be defined as consisting of resident populations (or parts thereof) and or non-resident populations (or parts there of) for a defined geographic area. This generalised definition illustrates that service providers' needs are likely to be better served if components of resident and non-resident populations can be estimated separately. Furthermore, specific issues associated with user requirements also suggest a component approach to service population estimation is preferable. 43 While most issues will be reserved for discussion in Section 5, one data quality issue is mentioned briefly below to illustrate a specific advantage of preparing separate resident and non-resident population estimates. 44 As explained in paragraph 40, access to non-resident population estimates is sought by the State and Northern Territory Local Government Grants Commissions for use as supplements to ERPs in funding allocation models. Depending on how such estimates are used, amalgamation of ERP and supplementary data can result in overcounts of the total State (or Territory) population. For example, the population base currently used by the WA Grants Commission to allocate funds for sanitation and refuse services, is an urban population estimate (based on ERPs) plus a net estimate for the number of additional non- resident persons employed in the area. In cases where the net estimate is negative, the net additional employment estimate is set to zero. Consequently, when aggregated across the State, the combined estimates will sum to a value greater than the State population. This consequence is considered less of an issue by the WA Commission (and the other Commissions which use an effectively similar principle) than the potential of under servicing utilities such as sanitation (WALGGC 1995, 28-30). Specifically, this service population estimation method recognises that sanitation is a permanent provision even when the population is not static throughout a 24 hour period. This example also illustrates that persons may need to be counted more than once (i.e. as both residents and non-residents) in the calculation of specific service population totals. 45 Finally, user comments also reveal that different service providers are interested in different types of non-resident populations (e.g. commuters, itinerant workers, tourists, etc.). In the case of the States and Northern Territory Grants Commissions, such variability is internalised since they use different population bases to determine funding allocations for different services. For example, as specified in paragraph 44, the WA Grants Commission uses a combination of resident and commuter population estimates as the base population to allocate funds for sanitation and refuse services. By contrast, the base population used by the WA Grants Commission to allocate funds for law, order and public safety is the resident population alone, although adjustments are made for tourists and geographic considerations (WALGGC 1995, 17). 46 As a consequence of the diversity of service population estimates sought by users, estimation of separate components of the non-resident population is expected to maximise the possibility of these various needs being met. Adoption of a component approach to service population estimation would allow user specific service populations to be defined by augmenting none, some, or all of components of the non-resident population to resident population estimates (ERPs) currently prepared by the ABS. 5 THE ISSUES 47 Broadly speaking the issues associated with estimating service populations can be split between conceptual issues (including definitional considerations) and production issues (such as data sources and frequency of estimation). Such issues are discussed in the sections below. Many of these issues were identified from user comments while others reflect statistical technicalities. 5.1 LACK OF CONFORMITY IN USER NEEDS 48 The issue of lack of conformity in user requirements for service population estimates has been discussed in detail in Section 4. As specified in paragraph 46, preparation of separate estimates of resident populations (ERPs) and a variety of non-resident populations is expected to maximise the possibility of users' diverse needs being met. A component approach to estimation would give users discretion to combine separate estimates in accordance with their preferences. Furthermore, a component approach would allow estimation of specific types of non-resident populations to commence once data sources, which meet user requirements, are identified. In general terms, a component approach offers the advantage of allowing non-resident population types to be estimated in accordance with a range of priorities including timeliness, data access, user demand and cost considerations. 49 The variety of non-resident populations sought by users include commuter populations, daytrippers, tourists and other overnight visitors (for one or many nights) and a diverse range of temporary residents (e.g. seasonal employees such as fruit pickers; cyclical employees such as fly-in/fly-out mine workers; winter-time visitors to sunbelt zones and residents of Aboriginal outstations in the dry season). 5.2 DEFINITIONAL CONSIDERATIONS 50 To meet user requirements, a variety of non-resident population definitions will be needed which, when combined with ERPs, will be comprehensive of users' service population needs, while singularly, will be mutually exclusive of the other component definitions. Development of such definitions would involve considerable resources and would require further user consultation. 51 Assistance may be available if English language translations of two international papers on mobility definitions can be obtained (Chermak 1991 and Husa 1991). Specifically, Chermak's paper purports to contain definitions of population mobility relating to commuting, temporary migration, labour migration and permanent migration while Husa's paper is described as discussing concepts and methods for defining and measuring forms of non- permanent mobility. The former is written in Bulgarian while the latter is in German. 52 Despite considerable research only one English language reference was found which relates to a variety of non-resident population types. Stan Smith's paper entitled Estimating Temporary Populations discusses the method devised by Robert Schmitt to estimate commuter populations and tourist populations in Hawaii (Smith 1994, 4-7). Schmitt used time to split these two types of non-resident populations, specifically between daytime visitors and overnight visitors (involving one or many nights). 5.2.1 Time and purpose-of-movement criteria 53 In general, additional time criteria to those specified by Schmitt as well as purpose-of-movement criteria are expected to be necessary to define the greater variety of non-resident populations (refer Section 5.1) sought by service population users. For example, separate estimates of tourists and seasonal employees might not be able to be differentiated by time criteria alone. 54 By contrast, the need to define additional time criteria may become less critical if length of stay details are available. Length of stay data are desirable not only for definitional considerations but also to enable the calculation of the average number of non-residents in an area for a defined period (e.g. annually, which would specifically benefit the Grants Commissions). Such averages are expected to assist estimation of the impact of non-resident populations on demand for local services. In addition, length of stay data are expected to be particularly important for specific types of non-resident populations, such as seasonal employees. 55 Reliance on existing data sources to estimate types of non-residents may prevent mutually exclusive time and purpose-of-movement criteria from being created for each type of non-resident population. For example, a recent newspaper article quotes the NSW Farmers Federation as estimating that one-third of fruit and vegetable pickers are backpackers (Travellers snap up jobs 1996, 27). It is possible such backpackers will be classified as tourists in some existing data collections (such as Overseas Arrivals and Departures) while simultaneously being classified as seasonal employees in other existing data collections (e.g. CES Job Search data). If combined, such data sources would produce overcounts of these two types of non-resident populations. Such overcounts may be needed to define specific service populations (refer example in paragraph 44). Otherwise, such overcounts may violate the statistical classification principle of mutual exclusivity, which requires that aggregated statistics not exceed a defined total. 56 Data currently used to estimate ERPs intercensally does not fully support the ERP usual residence criterion (for domestic residents only). Specifically, Medicare address change data has been used since 1986 to estimate interstate migration and while most data can be expected to reflect movements from one place of usual residence to another place of usual residence, some address changes may reflect temporary residential relocations or altered mailing arrangements (possibly involving non-residential addresses). Medicare data limitations make it impossible to separate usual residence migration from other change of address data. Likewise, a variety of administrative datasets are used to distribute population estimates within each State and Territory on an annual basis. This distribution process effectively produces intrastate migration estimates without explicit reference to usual residence criterion. Consequently, intercensal ERP estimates (for both interstate or intrastate migration) may inadvertently include estimates of the non-resident population (i.e. persons who reside or intend to reside at an address for less than a total of 6 months). 57 For similar considerations, a present Census definitional inconsistency which results in some non-residents from overseas (i.e. those who intend staying in Australia for less than twelve months) being included in censal ERP estimates, is to be looked at as part of development of the 2001 Census of Population and Housing (ABS 1996 a, 10). 58 As described in para 55, lack of definitional exclusivity between datasets creates the potential for overcounts to occur when such data are combined. While it is anticipated that persons may need to be counted more than once for specific service populations (refer example in paragraph 44), where possible, observance of the principle of mutual exclusivity to define components of the non-resident population can be expected to enhance the quality of estimates produced. 5.2.2 Geographic criteria 59 Another way of viewing service populations is to consider them as the destination for services which have originated from service providers. Such an analogy, however, conceals a key issue associated with service populations, namely geography. 60 Responses to the Demography Review indicate a lack of uniformity in the spatial boundaries used by agencies who are seeking service population estimates. Such diversity in spatial boundaries gives rise to issues relating to service population estimation in particular and statistical output in general. 61 The ABS is currently reviewing the geographic needs of statistical users in general, in its Review of the Australian Standard Geographical Classification (ASGC). User consultations conducted in Phase One of the ASGC Review reveal that users have a myriad of geographic requirements and that these are unlikely to converge in the future. By contrast, user needs for greater geographic variety are expected to continue to grow, especially with the increased use of GIS technology (ABS 1996c, paragraphs 51, 52 & 53). Examples of the geographic needs of users found from the ASGC Review include:
62 In defining new standard spatial boundaries which meet the geographic needs of ABS data users in general, the ASGC Review Team is looking to find solutions which will encompass the geographic needs of service population users in particular. Preliminary conclusions from the ASGC Review suggest that commuter populations and service provision, especially in non-urban areas of Australia, may be used as criteria to define new standard spatial boundaries (ABS 1996c, paragraphs 61-70). Consequently, adoption of such criteria can be expected to produce standard spatial units which maximise the likelihood of service providers' diverse geographic needs being met. The benefits of the ASGC boundary revisions will not be available to service population estimates before their proposed introduction in mid 1998, with adoption of the new boundaries likely to be phased in across the ABS collections. 63 Variability in the spatial boundary requirements of service providers may also reflect differences in the criterion used by service providers to define their boundaries. For example, service providers may define their boundaries in accordance with pre-defined administrative arrangements or in accordance with the residential addresses of their customers. The geographic regions resulting from these two criterion can be expected to be quite different for some services. For example, lack of cardio-vascular facilities in Canberra hospitals require Canberra residents to be treated in Sydney hospitals. Consequently, the geographic boundaries of Sydney hospitals offering cardio-vascular care may cover a large section of NSW and the ACT if based on the residential address of patients, but may be significantly reduced if defined according to hospital administration arrangements. Similarly, functional boundaries may differ from those defined administratively because persons living (or working or visiting) at the boundary extremes may use the services of providers in neighbouring areas. 64 Both these examples illustrate a geographic issue specific to service population estimation, namely identification of appropriate criteria to define the geographic needs of service providers. 65 Due to the diversity of geographic boundaries used by service providers, use of geography to define non-resident populations is expected to limit the applicability of consequent estimates for some users. For example, the present definition used to estimate daytrippers from the Domestic Tourism Monitor, requires that eligible persons are away from their usual residence for at least four hours and the round trip distance is at least 50 kilometres (Bureau of Tourism Research 1995, 15). This distance definition is known to prevent the WA Local Government Grants Commission from obtaining estimates of the number of daytrippers between Perth and Fremantle. 66 Thus, minimised use of geographic criteria in defining data can be expected to produce estimates of non-resident populations which maximise the geographic needs of a variety of service population providers. In practice, however, avoidance of geographic criteria may be difficult especially if pre-defined data sources are to be used in estimating types of non-resident populations (as illustrated by the examples above). 5.3 DATA SOURCES 67 Details about a variety of non-resident population types are currently collected as part of administrative data associated with service provision (e.g. hospital patient records) and from specific statistical surveys (e.g. International Visitor Survey). A number of existing data sources which may prove helpful in estimating non-resident populations are cited in Appendix A. Where apparent, possible advantages and disadvantages associated with these datasets are also mentioned. A thorough assessment of the datasets mentioned in Appendix A (and possibly others) would be required before any could be used with confidence to prepare non-resident population estimates (either for individuals or a variety of populations users). 68 Use of existing data sources to prepare estimates of service populations has a number of advantages over development of new data collections, including immediate access to data (possibly even time series data), lack of duplication of effort, reduced respondent burden and reduced collection costs. By contrast, the value of using existing data sources to estimate non-resident populations is limited by the ability of the existing data definitions to meet the data requirements sought (as illustrated in paragraphs 55 and 65). While modification of existing data definitions is possible, alterations may compromise the original intention of the collection and will be at the expense of time series comparability. 69 It is possible that the service population needs of some service providers will be so specific as to have little or no generic applicability to others (for example, persons demanding particular health care services) and that identification of such service populations will only be likely by referencing datasets which relate specifically to provision of those services. 70 It is also anticipated, however, that access to a range of generic non-resident population estimates would assist the service population needs of a variety of service providers. Obviously, the success of such generic estimates would depend on how closely data definitions associated with these estimates can meet individual user requirements. The data definitions associated with specific datasets (especially in relation to time, purpose-of-movement, length of stay and geographic criteria) would need to be assessed in accordance with a variety of users' needs to ascertain the value of preparing generic non-resident population estimates. 5.4 FREQUENCY OF ESTIMATION CONSIDERATIONS 71 With the exception of Sports and Recreation Victoria and the States and Northern Territory Local Government Grants Commissions, respondents to the 1996 Review of Demography were not specific in their frequency requirements for service population estimates. Consequently it is anticipated additional consultation will be needed to ascertain user needs with regard to this matter. 72 Data quality considerations, as much as user preferences, are expected to dictate estimation frequency. For example, data quality considerations are reflected in the Sports and Recreation Victoria request for estimates of commuter populations to be produced at least every two years because of continual changes in employment patterns (as illustrated in Table 1, Section 4). Similarly, the issue of seasonality for some types of non-resident population mobility (e.g. tourism) may require data estimation to be undertaken on a more frequent basis than for other types of non-resident populations. Seasonality considerations are expected to be especially pertinent to the Grants Commissions preference for average annual estimates of non-resident populations (refer paragraph 40). In practice, data seasonality is expected to be a complicated issue since seasons will vary for different types of non-resident populations. For example, seasons may be considered divisions of a 24 hour period in the case of commuter populations, or weekly or monthly divisions of a year for agricultural workers. In addition, mobility patterns (including timetables) are likely to differ between types of itinerant agricultural workers (e.g. fruit pickers and shearers). 73 Use of existing data sources to estimate non-resident populations may impose additional frequency of estimation constraints. For example, reliance on Census of Population and Housing data would restrict estimates to a five yearly basis. Frequency of estimation considerations would therefore need to be considered as part of dataset assessments. 5.5 ESTIMATION RESPONSIBILITIES 74 Various government agencies which responded to the 1996 Demography Review consider the ABS to be the most suitable agency to prepare service population estimates. In particular, the ABS is seen as having a comparative advantage in gaining access to datasets needed to prepare service population estimations which will be comprehensive and consistent for all Australia. In addition, the ABS is considered to have a comparative advantage in terms of statistical integrity. Specifically, as an independent agency charged with preparing Australia's official statistics, service population estimates prepared by the ABS are expected to be of a high quality and to be impartial. 75 Whilst these sentiments are true in general, it is expected the ABS will not a have comparative advantage over all agencies in gaining access to some datasets. For example, the State and Territory Health Departments and the Australian Institute of Heath and Welfare (AIHW) are expected to have more ready access to Health related datasets than the ABS. 76 Responsibility for preparation of specific service population estimates which have little or no generic appeal, is considered best accepted and undertaken by the relevant service providers (or agencies responsible for overseeing the provision of such services such as ATSIC, AIHW etc). Furthermore, even if preparation of generic estimates of service population components is undertaken by an independent agency, it is anticipated responsibility for compilation of such components into service population estimates is best assumed by individual data users. Preparation of generic estimates will only be possible if agreement can be reached between a range of users about the data definitions to be used. 77 Costs associated with service population estimation are expected to be considerable even if data from existing data sources are used. In particular, gaining access to a variety of datasets (possibly managed by a number of agencies) and extracting and maintaining data from such datasets can be expected to involve a substantial commitment in terms of staff and computing resources. 78 In documentation relating to the 1996 Demography Review, the ABS specified it was willing to prepare service population estimates on a user pays basis (ABS 1996a, 10). Responses to this statement, indicate a user pays approach is considered sub-optimal by a variety of population data users. Principally, concerned users consider anything other than a comprehensive and consistent approach to service population estimation which encompasses all of Australia (as opposed to the ad hoc approach expected from a user pays service), underestimates the value of service population estimates to population users in general. 79 While the ABS is sympathetic to this call for estimation consistency, the results of this research suggest it is unlikely that users' diverse needs can be met by a single set of non-resident population estimates. Preparation of service population estimates which meet specific user requirements (either for an individual or a variety of population users) would require factors such as purpose of use, quality, feasibility and cost of estimation to be discussed and agreed on by the estimation agency and the client. This research paper is expected to assist such discussions. BIBLIOGRAPHY Australian Bureau of Statistics (C.Y. Choi) [1979] - Population estimates in Australia - a discussion paper. Australian Bureau of Statistics: Demography and Social Branch, R-Series, Demog-1. Australian Bureau of Statistics [1983] - Methods and procedures in the compilation of estimated resident population 1981 and in the construction of the 1971-81 time series. Australian Bureau of Statistics: Catalogue No. 3103.0 Australian Bureau of Statistics [1991] - 1991 Census dictionary, census of population and housing. Australian Bureau of Statistics: Catalogue No. 2901.0 Australian Bureau of Statistics [1995] - Population estimates: concepts, sources and methods. Australian Bureau of Statistics: Catalogue No. 3228.0 Australian Bureau of Statistics [1996a] - Review of demography statistics 1996. Australian Bureau of Statistics: Demography Working Paper 96/2. Australian Bureau of Statistics [1996b] - Australian Standard Geographical Classification (ASGC) Review, Phase One Report. ASGC Review Team, Classifications and Data Standards Section, Australian Bureau of Statistics. Unpublished working paper, September 1996. Australian Bureau of Statistics [1996c] - 1996 Census dictionary. Australian Bureau of Statistics: Catalogue No. 2901.0 Australian Bureau of Statistics [1996d] - Tourism indicators Australia, March quarter 1996. Australian Bureau of Statistics: Catalogue No. 8634.0 Australian Bureau of Statistics (Tricia Cook) [1996] - Evaluation of administrative data sources for use in quarterly estimation of interstate migration between 1996 and 2001. Demography Working Paper 96/1. August 1996. Bell, Martin [1995] - Internal migration in Australia 1986-1991: Overview Report. Bureau of Immigration, Multicultural and Population Research. Canberra: Australian Government Publishing Service. Bureau of Tourism Research [1995a] - Domestic Tourism Monitor, 1993-94. Canberra: Bureau of Tourism Research. Bureau of Tourism Research [1995b] - International Visitor Survey, 1994. Canberra: Bureau of Tourism Research. Chermak, Zdenek [1991] - Developing the forms of spatial mobility of the population. Naselenie, Vol. 9, No. 3-4 Sofia, Bulgaria. Husa, Karl [1991] - Who is a migrant ? Problems of documenting and defining spatial mobility in developing countries. Demographische Informationen 1990/91. Vienna, Austria. Jacquot, Alain [1993] - Spatial friction between population and employment. Espace, Populations, Societies, No. 2. 1993. Villeneuve d'Ascq, France. McHugh, Kevin E. and Mings, Robert C. [1991] - On the road again: seasonal migration to a sunbelt metropolis. Urban Geography, Vol. 12, No. 1, Jan-Feb 1991. Silver Spring, Maryland. McKenzie, Fiona [1994] - Regional population decline in Australia: impacts and policy implications. Bureau of Immigration and Population Research. Canberra: Australian Government Publishing Service. Madden R., Wen X., Black K., Malam K. & Mallise S. [1996] - The demand for disability support services in Australia : a study to inform the Commonwealth/State Disability Agreement evaluation. Australian Institute of Health and Welfare. Canberra: Australian Government Publishing Service. Maher, Chris A and Stimson, Robert J. [1994] - Regional population growth in Australia: nature, impacts and implication. Bureau of Immigration and Population Research. Canberra: Australian Government Publishing Service. National Office of Local Government [1995] - Local Government National Report. 1995-1996 Report on the Operation of the Local Government (Financial Assistance) Act 1995. Canberra: Australian Government Publishing Service. Office of National Tourism [1996] - Forecast. The fifth report of the tourism forecasting council. Vol. 3 No.1, November 1996. Commonwealth Department of Industry, Science and Tourism. Canberra. Paris, Chris [1994] - New patterns of urban and regional development in Australia: demographic restructuring and economic change. International Journal of Urban and Regional Research, Vol. 18, No. 4, Dec 1994 (pp 555 - 72). Cambridge, Massachusetts/Oxford, England. Rose, Leslie S. and Kingma, Hildy L. [1989] - Seasonal migration of retired persons: estimating its extent and implications for the State of Florida. Journal of Economic and Social Measurement, Vol. 15, No. 1, 1989. Springfield, Virginia. Shryock, Henry S. and Siegel, Jacob S. [1971] - The methods and materials of demography. U.S. Bureau of The Census. Washington: U.S. Government Printing Office Schmitt, Robert C. and Zane, Lynne Y.S. [1995] - Military personnel and dependents in population analysis and projection. Applied Demography, Vol. 9, No. 3 Winter 1995. Washington. Smith, Stanley K. [1994] - Estimating temporary populations. Applied Demography, Vol. 9, No. 1, Summer 1994. Washington. Tayman, Jeff [1994] - Estimating population, housing and employment for micro-geographic areas. In: Studies in applied demography, edited by K. Vaninadha Rao and Jerry W. Wicks (pp 101-7). Bowling Green State University, Department of Sociology, Population and Society Research Center: Bowling Green, Ohio. Taylor, J [1992] - Population mobility: policy relevance, survey methods and recommendations for data collection. In: A national survey of indigenous Australians: options and implications. Research Monograph No 3, 1992, edited by J.C. Altman. Canberra: Centre for Aboriginal Economic Policy Research, Australian National University. Taylor, J. [1996] - Short-term Indigenous population mobility and service delivery. No. 118/1996 Canberra : Centre for Aboriginal Economic Policy Research, The Australian National University. Travellers snap up jobs. [1996, 10 November] by Fia Cumming, Political Correspondent. The Sun-Herald Sydney: John Fairfax Publications Pty. Ltd. Western Australian Local Government Grants Commission (WALGGC) [1995] - Local Government Grants: Principles and methods for the distribution of Commonwealth Financial Assistance in Western Australia. 2nd ed. Perth: Western Australian Local Government Grants Commission. APPENDIX A: EXISTING DATA SOURCES AND SOME ISSUES ASSOCIATED WITH THEIR USE Details about a number of existing datasets which may prove helpful in estimating non- resident populations are discussed below. As mentioned in the body of this paper, there are several advantages to using existing data sources over developing new data collections to estimate non-resident populations. These include gaining immediate access to data (possibly even time series data), lack of duplication of effort, reduced respondent burden and reduced collection costs. By contrast, the value of using existing data sources is limited by the ability of existing data definitions to meet the data requirements sought. While modification of existing data definitions is possible, alterations may compromise the original intention of the collection and will be at the expense of time series comparability. The types of non-resident populations of interest to service providers includes commuter populations, daytrippers, tourists and other overnight visitors (for one or many nights) and a diverse range of temporary residents (e.g. seasonal employees such as fruit pickers; cyclical employees such as fly-in/fly-out mine workers; winter-time visitors to sunbelt zones and residents of Aboriginal outstations in the dry season). Adoption of a component approach to non-resident population estimation is expected to maximise the possibility of users' diverse needs being met. Furthermore, a component approach would allow estimation of specific types of non-resident populations to commence once data sources, which meet user requirements, are identified without the need to wait until data for all types of non-resident populations are identified or developed. In general terms, a component approach offers the advantage of allowing non-resident population types to be estimated in accordance with a range of priorities, including timeliness, data access, user demand and cost considerations. CENSUS OF POPULATION AND HOUSING DATA For convenience considerations, unless otherwise specified the term Census is used to denote the Census of Population and Housing. In summary, data from the Census are expected to assist with quinquennial estimation of a variety of non-resident populations including commuters to Australia's metropolitan centres, Aboriginals in outstations at the time of the census and a broad but not necessarily distinctly identifiable range of other temporary residents or visitors to a region. Census data may also be able to be used as a base from which intercensal estimates of types of non-resident mobility can be prepared. But judicious care will be needed with intercensal estimation of non-resident populations to ensure for example, that seasonal biases which may be inherent to Census data are excluded from estimates. These issues will be discussed in more detail in the following paragraphs. Estimation of persons involved in non-resident migration (i.e. persons who reside or intend to reside at an address for less than 6 months in one year) will be possible from Census data by cross tabulation of 'place of enumeration' and 'place of usual residence' address data. This has been employed to produce estimates of short-term mobility within ATSIC regional councils from 1991 Census data (Taylor 1996, 2-7). Changes in data collection practices for the 1996 Census will allow persons who have no usual residence (according to the Census definition) to be identified. When cross referenced with 'place of enumeration' and 'employment' details, it may be possible to deduce estimates of seasonal employees and itinerant short-term residents from the 'no-usual residence' data. Similarly, when 'employment' and 'age' details are cross referenced with 'place of enumeration' and 'place of usual residence' data, estimates of winter-time migrants to the sunbelt zones may be able to be estimated from Census data. In general, however, lack of purpose-of-movement details are expected to limit identification of specific types of non- residents from Census data. Furthermore, census estimates of specific types of non-residents may not be typical as a consequence of the Census reference date. This is especially likely for non-resident movements which are subject to seasonal variability. The Census reference date is not randomly selected but chosen to reflect a time in which a particular type of short-term migration is likely to be at a minimum. Specifically, the Census date is selected to prevent it coinciding with school vacations. This policy required the Census date (of 30 June since 1933) to be changed (to August 6 in 1996), thereby preventing it from coinciding with a school vacation period which was created when the State and Territory Education Departments converted to a four-term school year in the 1980s. Conducting a Census during school vacations is undesirable because of the increased likelihood of collection difficulties and data quality considerations associated with people being away from their usual residences (ABS 1991, 34). Consequently, estimates of non-resident migration prepared from cross tabulations of 'usual residence' and 'place of enumeration' data can be expected to understate a specific type of seasonal migration, namely school vacation migration. Conversely, it is possible that other types of seasonal migration may be prevalent at or around the Census date. For example, this date may correspond with shearing or fruit picking activities in certain parts of Australia. Extreme care will therefore be needed if seasonal estimates (other than for the season represented by the Census date) or average annual estimates of non-resident migration are to be prepared using census data either directly or as a base to prorata other data. One advantage of the Census reference date coinciding with a period in which domestic tourism is likely to be at a minimum, is that non-censal estimates of tourists, with the possible exception of business travellers, can be combined with censal estimates of non-residents without undue concern about overcounting (refer paragraph 55 for further details about this issue). With regard to commuter populations, census counts can be prepared by cross tabulating 'place of usual residence' and 'place of employment' data for those Australian metropolitan centres in which 'place of employment' data are coded. Extensions to usual coding arrangements in WA are expected to also allow fly-in and fly-out employees for the Pilbara mining region to be estimated from 1996 Census data. OTHER DATA SOURCES RELATING TO COMMUTER POPULATIONS Intercensal estimates of commuter populations have been prepared for all local government areas in WA (i.e. not just metropolitan regions) by combining the latest population census counts of numbers of people employed in a local government area with counts developed from ABS Business Register data (WALGCC 1995, 28-30). However, care is needed in using ABS Business Register data between the Economic censuses because data maintenance activity has become less regular in recent years due to cost saving measures. The NSW and Qld Local Government Grants Commissions find customer residential address details from Library Services to be a good source for estimating commuter populations. Similarly, Sports and Recreation Victoria have used data from customer surveys at municipal leisure centres to determine the residence and employment locations of users. The Transportation surveys conducted by most of the State and Territory Transport Departments could be used in commuter population estimation. Estimates of employee numbers working to fly-in/fly-out arrangements, are expected to be available from mining company records. To date, discussion of commuter populations has been restricted to persons involved in daytime migration for employment purposes. Whilst employment related activity is expected to dominate commuter populations, daytime migration can occur for reasons other than employment, e.g. for education arrangements. Previous research into school enrolment data suggests data relating to education based commuting is not likely to be readily available (ABS (Cook) 1996, 25). Service providers with a pressing need for a broad range of commuter population estimates, such as transport companies or child-care providers, are expected to be able to use their administrative records and possibly service specific surveys (e.g. transportation studies) to assist them to prepare estimates of their specific service populations. DEMAND FOR COMMUTER POPULATION ESTIMATES A variety of respondents to the 1996 Review of Demographic Statistics specified a need for commuter population estimates. In general, the relative importance of commuter populations to non-resident population mobility has been acknowledged by researchers for many years. For example, estimates of commuter populations were prepared by the US Federal Civil Defence Administration during the Cold War years of the 1950s, to assist in planning the location and size of bomb shelters (cited in Smith 1994, 4). In contemporary terms, the significance of commuter populations to non-resident population mobility is attributed to the occurrence of counter-urbanisation at a time when employment growth has remained concentrated to metropolitan regions (Jacquot 1993 and Knight cited in Maher and Stimson 1994, 12). With regard to future expectations, while migration in general can be expected to decline in Australia because of population ageing (Bell 1995, xviii), declines in counter-urbanisation (and therefore commuting) may not occur at the same rate since counter-urbanisation is generally favoured by relatively older migrants, specifically persons in their thirties and older (Bell 1995, p 82). A contradictory theory, however, suggests that increased reliance on computer technology in some industries has reduced the need for commuting to centralised worksites (Maher and Stimson 1994, 23), which may foreshadow a future decline in the relative importance of commuting. The importance of commuting to non-resident migration in the future will therefore depend, in part, on whether the industries which experience employment growth can readily adapt to decentralised employment practices. Since non-centralised employment practices will not, however, be possible for many types of industries (e.g. mining), the relative importance of commuter population estimation appears to be assured for some time to come. OTHER DATA SOURCES RELATING TO OTHER DAYTRIPPER POPULATIONS Details about Australian residents (although not necessarily in accordance with the ABS definition of usual residence) who undertake daytrips (other than to their normal place of work) which require them to be away from their usual residence for at least four hours and the round trip distance is at least 50 kilometres, are available from the Domestic Tourism Monitor (DTM). Purpose-of-trip information is also captured as part of daytripper details. As explained in paragraph 65, this distance definition is known to limit the value of DTM data to some data users. Specifically, it prevents the WA Local Government Grants Commission from obtaining estimates of the number of daytrippers between Perth and Fremantle. The Bureau of Tourism Research will be assessing whether this criterion needs to be maintained in a new survey (to be called the National Visitor Survey or NVS), which is currently being developed to replace the DTM. Although daytripper data are not presently captured from the International Visitor Survey (IVS), data relating to visitation of selected places and attractions by international visitors may be of some assistance in estimating daytrippers for service population estimates. Collection of daytripper information from international travellers is being examined by the Bureau of Tourism Research as part of their present review of the IVS, happening concurrently with development of the NVS. The benefit of any daytripper definitional amendments, including spatial boundary changes, will not be available for use in service population estimates until after their implementation in the revised IVS and the new NVS, intended by January 1997 and October 1997 respectively (Office of National Tourism 1996, 21). OTHER DATA SOURCES RELATING TO OVERNIGHT VISITORS AND TEMPORARY RESIDENTS Estimates of tourist populations is considered a priority by a variety of respondents to the 1996 Review of Demographic Statistics. By comparison, whilst estimation of a range of temporary residents (such as seasonal employees) is important to specific respondents, demand for such estimates was less well represented in overall responses to the 1996 Demography review. A number of data sources other than the Census of Population and Housing are expected to assist with estimation of a broad range of overnight visitors (for one or many nights) and temporary residents to a region. In particular, estimates of persons who stay overnight (for one or many nights) by purpose-of-visit and length of stay, are available from the Bureau of Tourism Research's IVS and DTM (or subsequently NVS). Data from these Bureau of Tourism Research surveys may better meet user needs for tourist estimates than data from the ABS Survey of Tourist Accommodation (STA). Estimates of visitors who stay in non-commercial accommodation (such as with friends and relatives) or in commercial accommodation establishments with fewer than five rooms (such as Guest Houses) are outside the scope of the STA. By contrast, estimates of visitor numbers for a range of commercial and non-commercial accommodation types, are available from the IVS and DTM surveys (or subsequently the NVS). Furthermore, details about origin-of-visitors are collected from these Bureau of Tourism Research surveys. Origin-of-visitor data are expected to be important in service population estimation and may avoid the problem of overcounts occurring if tourist data are combined with other population estimates. As explained in paragraph 55, overcounting may not be a significant issue for some users. All the same, origin-of-visitor details can be expected to enhance the quality of non-resident population estimates. Data about overseas visitors are also available from the ABS Overseas and Arrivals (OAD) series. OAD data provide statistics on international visitors, including main purpose-of-visit and intended length of stay details. Data relating to visitor destinations below State level within Australia are not, however, available from this data series. Consequently, OAD data are expected to be less valuable to service population estimation than IVS and DTM data. Estimates of persons involved in temporary residential migration (for seasonal and or employment considerations) could be possible from DTM data without the concept of residence presently used in this survey being altered. It is considered unlikely, however, that the Bureau of Tourism Research will wish to adopt modifications which may compromise the original intention of the survey. Data from a variety of Department of Social Security (DSS) and Department of Employment, Education, Training and Youth Affairs (DEETYA) administrative datasets could assist with intercensal estimation of temporary residents. In particular, the Commonwealth Employment Service (CES) datasets are expected to aid seasonal employment estimates. The impending privatisation of this Service, however, could impact future access to such administrative datasets. Estimates of seasonal employees are also expected to be kept by variety of agricultural co-operatives, employer associations and unions. A recent evaluation of the Monthly Population Survey (MPS) and associated household based surveys conducted by the ABS (such as the National Aboriginal and Torres Strait Islander Survey), concluded that special sample design and estimation procedures would be needed before estimates of population mobility from this survey vehicle can be relied on (ABS (Cook) 1996, 12-15). Until a separate review of existing MPS estimation practices with regard to interstate migration estimation has concluded, the potential for the MPS to assist in measuring temporary residential population migration is limited. Finally, a plethora of subject specific survey and administrative data are expected to be available to meet the data needs of specific service providers. Knowledge of the existence of such datasets is expected to be held by the relevant service providers (or agencies responsible for overseeing service provision). For example, a variety of data sources are discussed in the CAEPR paper on Short-term Indigenous population mobility and service delivery (Taylor 1996, 3-13). Similarly, a significant portion of the AIHW paper on The demand for disability support services in Australia is devoted to data sources and limitations (Madden et. al. 1996, 29-36). APPENDIX B: EXTRACT FROM REVIEW OF DEMOGRAPHY STATISTICS 1996 RELATING TO SERVICE POPULATION ESTIMATES Comments received There is strong and widespread support for service population estimates. The agencies that support service population estimates include most State Treasuries and Health and Community Services authorities as well as organisations such as the NSW Department of Bush Fire Services, Tourism NSW, the NSW Environmental Protection Agency, Sydney Water, the Victorian Department of Sport and Recreation and South Australia Police. Service population estimates are perceived as invaluable for the purposes of strategic planning. Estimates based on resident population do not meet the needs of service providers in tourist areas, mining towns and seasonal employment. Service population estimates are relevant to State and local government planning and resource allocation. They are recognised as important by States Grants Commissions in taking service needs into account for the distribution of Commonwealth funds for local governments. The WA Local Government Grants Commission noted that "... a significant number of councils have expressed dissatisfaction with the ERPs. In part, the development of service populations would address these concerns." Users claimed that the ABS is in a position of comparative advantage to produce estimates and has ready access to data for employment and tourist accommodation. It is stressed that service population estimates are required for LGAs or even smaller scales. Many agencies argued that the user pays approach to service estimates is no longer appropriate. Rather, estimates should be produced as a matter of course and the cost be absorbed by the ABS. ANU cautioned that the issue of service population estimates is a complex one. The definition of population will vary according to which service is being considered. For example, commuter populations being present at a particular locality for parts of the day and seasonal peaks of tourists are very different forms of service populations and reflect the issues that need to be addressed. ANU suggested that ABS produce an issues paper which explores the problems involved in service population estimates. ABS response The issues raised by the concept of service population are complex. By their very nature, service populations will vary according to the service identified and the time of day and year. As such a single estimate or even a consensus about the range of estimates of the population to be serviced is unlikely to be achieved. ABS does not see its role as putting out a whole variety of service estimates to meet these needs. ABS will prepare an issues paper in 1997 which will examine user needs, the conceptual and methodological issues involved and the reporting load. For the 1996 Census of Population and Housing the ABS will also code SLA of workplace for persons employed in the Pilbara and possibly other areas to gain a further insight into the issue. In the meantime ABS is willing to assist with the production of service population estimates where the user is prepared to take responsibility for the wide range of explicit and implicit assumptions and cost involved. Document Selection These documents will be presented in a new window.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
