New population projections for Queensland and its Statistical Divisions.

Wilson, Tom ; Bell, Martin ; Heyen, Glen 等

The Queensland Government recently released its 2001-based population projections for the State and its Statistical Divisions. This paper reports on the main features of the projections to 2051 at the State level and to 2026 at the statistical division scale, the various projection assumptions made, and the new multiregional projection model written specifically for this round of projections. The projections indicate that Queensland is likely to experience very substantial population growth in coming decades, increasing concentration in the south east, and significant population ageing.

**********

INTRODUCTION

Over the half century from 1951 to 2001 Queensland's population grew from 1.2 to 3.6 million. (1) According to the medium series of the recently released 2001-based Queensland Government population projections, the State's population is likely to increase by a further 2.9 million over the next 50 years, passing the 5 million mark in 2022, and reaching a total of 6.5 million by 2051. (2) The significance of this projected population growth can be appreciated by a comparison with the medium series of the latest United Nations world population projections. (3) For the whole of the 2000-50 period for which the UN publishes its projections Queensland's projected rate of population growth not only outstrips that of the world as a whole, but also exceeds the projected population growth rate of the 'less developed regions' as classified by the UN.

This paper describes the method, assumptions and results of the new projections. It begins with a brief description of the multiregional projection model employed to produce this set of projections. It then goes on to describe the projection assumptions made in each of the low, medium and high series before presenting selected highlights of the future characteristics of Queensland's population. A brief comparison with the new Australian Bureau of Statistics (ABS) projections is made.

MODEL

For this round of population projections the Queensland Centre for Population Research at The University of Queensland developed a new population projection model called Subnational Projections for Australia and Regions of Queensland (SPARQ). SPARQ is a multiregional cohort component model which has the capability to provide population projections by sex and single years of age from 0 to 99 and 100+ over a projection horizon of up to 100 years for Queensland, a variable number of sub-state regions, and for Australia as a whole.

The 'multiregional' description of the model refers to its modelling of internal migration (4) as directional (place to place) flows, rather than as net migration. There are at least three strong arguments for projecting internal migration in this way rather than as net migration numbers.

(1) Perhaps the most important reason is a conceptual one: there is no such thing as a net migrant so the modelling of place to place migration flows is therefore a better representation of demographic reality. (5) Multiregional models treat states and regions as part of a wider interacting demographic system where the flows of migrants between regions depend, amongst other factors, on the size and age-sex composition of each region's population. Since these demographic characteristics change over time it makes little sense to fix the value of net internal migration.

(2) In a similar way, many net migration models incorporate fixed age profiles of net migration. Even in a hypothetical situation where total net internal migration for a region remains fixed over a projection horizon the ageing of the region's population would lead one to expect a change in the age profile of net internal migration.

(3) The third important advantage is a practical one. Multiregional models avoid the potentially embarrassing situation which may arise with a net migration model when it projects more net out-migration in some age-sex groups than there is population, thus giving nonsensical negative populations!

The projection computations in SPARQ proceed as follows. The computer program begins with the 2001 mid-year estimated resident populations and advances in one year steps to produce projected populations for 30th June each year up to 2051. Within each mid-year to mid-year projection interval the calculations are divided into two stages: projections for Queensland and the rest of Australia are calculated first, followed in the second stage by projections for the Statistical Divisions. This ordering allows the statistical division projections to be constrained to the State figures.

In the first stage Australia is divided geographically into two regions: Queensland and Australia minus Queensland. Then for each age-sex group in each region projections are made using the population accounting equation:

Population(t+1) = Population(t) + Births(t,t+1) - Deaths(t,t+1) + Interstate in-migrations(t,t+1) - Interstate out-migrations(t,t+1) + Net international migration(t,t+1)

where 't' and 't+1' refer to mid-year points in time and 't,t+1' a mid-year to mid-year time interval. The projection of births and deaths is fairly straightforward as the approach is the same as that used for single region cohort component models described in the demographic textbooks. The means by which the numbers of projected interstate migrations are obtained are a little more complex. For each age-sex group projected migration flows are found as the product of an occurrence/exposure migration rate and the population at risk in the area of origin. These occurrence/exposure migration rates are defined as the number of migration occurrences in a year divided by the population exposed to the risk of migrating. So, for example, the rate of in-migration to Queensland (which in this two-way division of the country is also the rate of out-migration from Australia minus Queensland) is defined as:

Migration from Aus minus Qld to Qld (t,t+1) 1/2 [Pop. of Aus minus Qld(t) + Pop. of Aus minus Qld(t+1)].

The population at risk of migration is approximated by the mean of the start and end-of-interval populations in the area of origin. The rates were calculated using the average of several years' worth of ABS Medicare-based migration statistics in the numerator and estimated resident populations in the denominator. Due to data limitations at the sub-state scale and the desirability of a common method at both state and sub-state levels, international migration is projected using net figures, rather than as immigration and emigration which would be conceptually more satisfactory. Projections for Australia as a whole are calculated by summing the Queensland and Australia minus Queensland figures.

In the second stage the program computes the statistical division projections, constraining all the projected births, deaths and interstate migrations to the State figures. Three types of migration flow--interstate, intrastate and international--are distinguished in the population accounting equation:

Population(t+1) = Population(t) + Births(t,t+1) - Deaths(t,t+1) + Interstate in-migrations(t,t+1) - Interstate out-migrations(t,t+1) + Intrastate in-migrations(t,t+1) - Intrastate out-migrations(t,t+1) + Net international migration(t,t+1)

Interstate migration is projected as in stage 1 with occurrence/exposure migration rates and origin populations at risk. For interstate migration from each statistical division to Australia minus Queensland the population at risk is the statistical division population; for interstate migration into each statistical division the relevant population at risk covers Australia minus Queensland.

The calculation of intrastate migration is more complex. A fully multiregional migration matrix of 11 origins by 11 destinations by two sexes by 101 age groups would be much too big relative to the number of migrations in the matrix to give stable migration rates. A bi-regional reduction of the full multiregional model is used instead. This has the advantage of requiring much less input data and data smoothing but giving results very close to that of the full multiregional model. (6) In the bi-regional approach the program deals with each statistical division in turn, dividing Queensland into two parts consisting of the statistical division in question and the rest of the State (thus the rest of Queensland residual region varies depending on which statistical division is the focus of interest). Migration exchanges between the statistical division and the rest of the State are then projected as before using occurrence/exposure migration rates and origin populations at risk. Once intrastate migration for all the Statistical Divisions has been calculated a small adjustment is made to ensure that for each age-sex group net intrastate migration sums to zero. Migration data for the numerators of the intrastate migration rates were estimated using Census migration statistics constrained to the State-level Medicare-based migration figures. As before, international migration is included as net flows.

The way in which internal migration is projected is the major distinction between SPARQ and the ABS subnational population projection model. Like SPARQ, the ABS model projects place to place migration flows by age and sex. However, the ABS model then scales these migration figures to pre-defined net internal migration totals for each state and region. (7) The consequence is that the age structure of net migration responds to the changing age structure of the regional populations but not to changes in the population sizes of regions relative to one another. In a series of projection experiments conducted by the first two authors it has been shown that substantial differences in projected populations may arise by constraining migration flows to net totals in this way. (8) This is why the ABS approach was not used.

PROJECTION ASSUMPTIONS

Three projection series--high, medium and low--were formulated for this round of projections. The fertility, mortality, internal migration and international migration assumptions used in each of the series were prepared by the Queensland Government Population Projections Advisory Group whose members were drawn from several government departments and The University of Queensland. The assumptions are discussed in turn.

Fertility

Over the last 30 years total fertility rates (TFRs) in Queensland and Australia as a whole have followed a general down-wards trend, declining steeply in the 1970s, plateauing to some extent in the 1980s, and then continuing on a gentle decline throughout the 1990s. (9) The recent decline has been due to pronounced falls in the age-specific fertility rates of women in their twenties. These falls are only partly compensated by rising rates for women in their thirties and forties. (10) Overseas evidence shows these changes in fertility are common. In many countries the declines have taken fertility to historically low levels (for example, the TFRs in 2001 were 1.57 in Sweden, 1.41 in Switzerland, 1.24 in Italy and 0.93 in Hong Kong). (11) There is little empirical or conceptual evidence to suggest that these trends are about to change. Space precludes any further justification of the projection assumptions here; a more detailed discussion may be found in a Queensland Government working paper. (12) For the medium series, then, it was assumed that Queensland's TFR would gradually decline to 1.60 by 2020-21, remaining constant thereafter. For the Australia minus Queensland region it was assumed that the Queensland to Australia minus Queensland TFR ratio of recent years would hold constant, giving a TFR of 1.56 from 2020-21 onwards. The low series assumes a faster decline over the next two decades to 1.40 (Queensland) and 1.36 (Australia minus Queensland) while the high series approximately maintains recent fertility levels, setting a constant 1.80 for Queensland and 1.75 for Australia minus Queensland throughout the projection horizon. Projected TFRs for the Statistical Divisions were prepared by multiplying the State TFR by scaling factors calculated from the experience of the last decade. The scaling factors were assumed to remain constant over time.

Mortality

Substantial gains in life expectancy in the first half of the twentieth century resulted from major reductions in death rates at the childhood and young adult ages due to the near eradication of infectious diseases. (13) Because death rates at these youthful ages are now very low, further progress will be limited and will have little impact on life expectancy. Life expectancy gains since the 1970s have come mostly from falling death rates in the later adult and elderly ages. (14) Reductions in death rates at these ages have been substantial over recent decades and have been mostly responsible for raising life expectancy at birth over the 1971-2001 period from 68.2 to 77.7 years for males and from 74.7 years to 82.8 years for females. (15) However, whilst considerable progress continues to be made against mortality there are some signs that further reductions will become gradually more difficult to achieve. (16) It is therefore assumed that Queensland's future life expectancy at birth gains will be smaller than in recent decades. For all three projection series it is assumed that life expectancy will increase to reach 88 years for males and 90 years for females by 2050-51. It is expected to be very slightly higher in Australia minus Queensland (but rounded to the nearest year the values are the same as those for Queensland). These assumptions are the average of the ABS 1999-based projection assumptions for high and standard life expectancy. As with fertility, projected life expectancy values for the Statistical Divisions were obtained by applying scaling factors based on trends over the 1990s.

Internal migration

Queensland.

The internal migration assumptions were formulated in terms of gross migraproduction rates (17) (GMRs). The GMR is a single index summary of age-specific migration rates analogous to the TFR. It can be interpreted as the average number of migrations a person would make in their lifetime from one region to another assuming constant age-specific rates and an absence of mortality. For the medium assumption out-migration GMRs from Queensland to Australia minus Queensland were set to increase by 15 per cent over the course of the projection horizon whilst it was assumed that GMRs for in-migration to the State would rise by 30 per cent. The modest increase in GMRs to 2050-51 is based on the belief that we will live in an increasingly mobile world due to factors such as the continued professionalisation of the workforce, rising incomes, greater experience of places through wider travel, and an increasingly flexible labour market where changes of employment are likely to be more frequent. The 30 per cent rise for Queensland assumes that, as the State's population grows, Queensland will become increasingly attractive to interstate migrants, both for job-related moves and for retirement and lifestyle-related migrations. For the low series it was assumed that the GMRs for migration to and from Queensland would remain constant. The high series has a 30 per cent rise in the out-migration GMR by mid-century and a 60 per cent rise in the in-migration GMR.

Statistical Divisions

For both interstate and intrastate migration at the Statistical Division scale it was assumed that the out-migration GMRs for all three projection series would increase gradually in line with the all-Queensland out-migration GMR (0 per cent, 15 per cent or 30 per cent) by 2050-51. For in-migration, however, it was anticipated that Brisbane, Moreton and Wide Bay-Burnett would grow in attractiveness to interstate migrants more than other parts of the State and so the in-migration GMRs were set to rise by more than the State amount. Being mindful of land supply constraints in south east Queensland, Moreton's in-migration GMR was assumed to increase faster than Brisbane's.

International migration

The medium series assumes a net gain from international migration of 16,000 per annum for Queensland, except for the early years of the projection which are assumed to trend from the recent high net migration figures. The Australia minus Queensland assumption is an annual net gain of 84,000 through international migration exchanges, giving an all-Australia figure of 100,000. The low net migration scenario for Australia is set at 75,000, with 12,000 of that allocated to Queensland whilst the high scenario assumes 125,000 per year, with Queensland taking 20,000 of that total. These relatively low shares of net international migration assume that Queensland will not encroach on Sydney's role as the country's principal entry port for international migrants. Data from the last 30 years support this assumption, revealing no clear relationship between Queensland's population size and its share of national net international migration. The distribution of net international migration to Statistical Divisions was based on past trends as assessed by a combination of population accounting equation residuals, (18) census data and the judgement of the Population Projections Advisory Group.

RESULTS

Queensland

Figure 1 shows how Queensland's population is projected to increase over the first half of this century according to the high, medium and low series. By 2026 the medium series projection puts the State's population at 5.3 million, with the high and low series about 0.4 million either side of this. The extent of this projected growth between 2001 and 2026, from 3.6 to 5.3 million, is roughly equivalent to the current population of the Brisbane statistical division. This 46 per cent increase compares with a 2001-26 medium series projected growth of 22 per cent for Australia minus Queensland (from 15.8 to 19.2 million).

[FIGURE 1 OMITTED]

Looking further into the future, the medium series projection suggests that Queensland will be home to a total of 6.5 million people by mid-century. Reflecting the greater uncertainty of the demographic future this far out, the high-low range is much wider, varying from 5.3 million in the low series to 7.8 million according to the high series. Despite these very substantial projected increases, however, the annual addition to the State's population according to the medium series is likely to decline from 78,000 in 2001-02 to 59,000 by 2025-26, falling to 38,000 by 2050-51. In terms of the population growth rate this is a decline from an annual average of about two per cent per annum over the 1990s to 0.6 per cent by mid-century, though this lower rate of growth will still be substantially above the 0.25 per cent projected for Australia minus Queensland. An analysis of the components of change--births, deaths and migration--sheds some light on why the State's absolute annual increase and rate of growth are likely to decline in the long run.

The projected components of change from the medium series are plotted in Figure 2. Most of the decline in population growth can be seen to stem from the coming increase in the number of deaths. Although death rates are projected to fall considerably in the future, the growing size of the elderly population, particularly with the addition of the baby boom cohorts, will outweigh these mortality reductions and therefore push up annual deaths figures. A probable increase in the number of births will cushion the impact of this mortality trend to some extent. The rise in the number of births comes despite the long-run projected TFR being well below replacement level. This is because of the sheer magnitude of the growth in Queensland's female population of childbearing age through migration.

[FIGURE 2 OMITTED]

However, the gradual drop in natural change is not responsible for the entire projected decline in population growth. Interstate migration also plays a role. Although both in- and out-migration numbers are projected to rise significantly in the coming decades the net balance of interstate migration declines slightly, falling from 29,000 in 2001-02 to 25,000 by 2050-51. This may seem counter-intuitive for Australia's growth State. But in fact the medium series projection assumes that Queensland's attractiveness to interstate migrants will grow throughout the projection period: the migration rates from Australia minus Queensland to Queensland have been set to increase over the course of the projection horizon. Coupled with a growing population in the rest of the country a large rise in in-migration flows is expected. But out-migration from Queensland is projected to grow slightly faster, and this is simply due to the fact that Queensland's more rapid growth generates a proportionally larger population at risk of migration. This is one of the key feedback loops in population dynamics that net migration models fail to capture. Despite this slowdown, under the medium assumption Queensland's 18.7 per cent share of the national population in 2001 is set to grow to 21.6 per cent by 2026 and 23.4 per cent by 2051.

As well as being home to a much larger population in the next few decades, Queensland will also have a much older population. The population pyramids in Figure 3 depict the evolving age-sex structure of the State's population at 25 year intervals: 1976, 2001, 2026 and 2051. It is clear that the largest absolute and percentage increases are at the oldest ages. The population aged 90+, for example, is projected to grow from just 14,000 in 2001 to 182,000 by 2051 according to the medium series. The median age of 35.0 years in 2001 is likely to rise to 47.3 years by this time, compared to a shift from 35.5 years to 46.6 years for Australia minus Queensland. The slightly faster ageing to be experienced by Queensland is driven to a large extent by a significant retirement element in the State's age profile of in-migration.

The ageing of the population will undoubtedly have a great variety of consequences for the social and economic life of Queensland in the future--changing patterns of consumer demand, changing health care requirements and a different workforce composition to name but three examples. But ageing also has implications for several demographic indicators. One of these is the sex ratio. Because of differential male and female mortality and the very substantial projected growth of the elderly age groups, the overall sex ratio of Queensland's population will fall slightly, from 99.1 males per 100 females in 2001 to 98.1 by 2051.

[FIGURE 3 OMITTED]

Further details of the State's population projections are available online. (19) We turn now to briefly examine some of the spatial aspects of Queensland's projected population change.

Statistical Divisions

Table 1 summarises the projections of total population size at the statistical division scale. In terms of total population growth three groupings of Statistical Divisions may be distinguished:

(i) the very high growth regions of Moreton and Wide Bay-Burnett;

(ii) the western Queensland regions with little projected change in population size (North West, Central West and South West);

(iii) the moderately high growth of the other coastal regions (all other Statistical Divisions).

The fastest growing region, Moreton, is projected to grow by 81 per cent over the 25 years to 2026 according to the medium series projections (and to nearly double its current size under the high series). The high growth of Moreton and Brisbane is expected to give south east Queensland a population in 2026 equivalent to that of the whole of Queensland today. The differential projected rates of population growth amongst the Statistical Divisions result in a shifting geographical distribution of Queensland's population towards the east coast and the south east, further challenging the view that Queensland's is Australia's most decentralised state. Figure 4 shows how each region's share of the State's population is expected to change over the 2001-26 period. Details of projected statistical division age structure changes are not discussed here. They are available at the Queensland Government website mentioned above.

What are the demographic drivers of these projected regional population changes? Figure 5 (20) provides a preliminary answer by displaying for each statistical division the 2001-26 annual averages for the Crude Rate of Natural Increase (CRNI) and the Crude Net Migration Rate (CNMR). The CNMR is defined as net intrastate, interstate and international migration combined, divided by the population in the statistical division. (21) The sum of the CRNI and the CNMR gives the overall growth rate--represented in the graph by the distance from the diagonal dashed line. It can be seen that Figure 5 largely supports the tri-partite classification of the Statistical Divisions suggested above. The two very high growth regions are projected to grow mostly from migration gains whilst the healthy natural increase of the three inland regions prevents major population decline from migration losses. The other regions will grow strongly from both the CRNI and the CNMR.

[FIGURE 4 OMITTED]

Comparison with ABS projections

We now consider how the Queensland Government projections compare with those of the recently released ABS 2002-based projections. Comparison is only possible at the State, capital city and rest of state scale as this is the spatial disaggregation provided by ABS, and we limit the comparison here to the State level figures. Table 2 provides a comparison of the three Queensland Government series with the ABS A, B and C series. It can be seen that the medium series projections are in remarkably close agreement with the ABS series B figures (in fact too close for them to show up on a graph). Whilst it might be expected that the Queensland Government and ABS projections for the State would be very roughly similar, it should be noted that the very close correspondence revealed here has arisen simply by chance. No attempt was made to constrain the Queensland Government projections to those of ABS (nor was it possible given that the ABS projections were published after the Queensland Government projections).

[FIGURE 5 OMITTED]

The projected natural change and net migration (internal and international combined) are very similar for the first half of the projection horizon. According to the Queensland Government medium series, natural change is unlikely to fall as far into negative territory as projected by the ABS. Detailed projection results (not shown in the table) reveal that the difference between the two series is mostly due to variation in the numbers of deaths--caused by the diverging life expectancy at birth trajectories of the two sets of projections. The projected natural change differences in the latter half of the projection horizon are partly balanced out by different net migration numbers, resulting in similar total populations. The projected age structures are very similar too, with the median age for both sets of figures at about 42 years by 2026 and 47 by 2051. The higher life expectancy and slightly lower fertility in the Queensland Government projections result in a mid-century population which is very slightly older than that projected by the ABS series B, with age groups 60 and over being marginally larger than in the ABS series B projections, and the younger age groups slightly smaller.

CONCLUSIONS

Queensland is likely to experience very significant population change in coming decades. Although population futures cannot be predicted with a high degree of precision, particularly more than a generation ahead, the broad features of Queensland's demography in the first half of the 21st century are certain: substantial population increase, substantial ageing, and increasing concentration in the south east and coastal areas. The prospect of a mid-century population equivalent to that of New South Wales today, and south east Queensland housing the current population of Sydney will provide today's and tomorrow's policy makers with some tough challenges.

Table 1: The past and projected populations of Queensland's Statistical
Divisions, 1976, 2001 and 2026, ('000s)

 Projections
 Low Medium
 Estimates series series High series
 1976 2001 2026 2026 2026
 thousands

Brisbane 1,020 1,650 2,140 2,292 2,451
Moreton 198 724 1,221 1,313 1,408
Wide Bay-Burnett 144 236 323 358 389
Darling Downs 163 210 239 257 276
South West 29 27 25 28 30
Fitzroy 130 182 216 235 257
Central West 14 12 11 12 13
Mackay 90 138 167 181 195
Northern 138 190 237 258 279
Far North 125 224 295 321 348
North West 42 34 31 34 38
South East Queensland 1,219 2,375 3,361 3,605 3,859
Queensland 2,092 3,629 4,906 5,289 5,685
Australia minus
 Queensland 11,941 15,756 18,262 19,162 20,068

Source: Queensland Government population projections 2003 and ABS
Estimated Resident Populations adjusted for boundary changes by
Queensland Department of Local Government and Planning

Table 2: Summary statistics comparing the ABS and Queensland Government
population projections for Queensland

 Low Medium
 Old Govt ABS C Old Govt ABS B

Assumptions
Long-run TFR 1.40 1.43 1.60 1.64

 Years
Male [e.sub.0], 2050-51 87.8 84.0 87.8 84.0
Female [e.sub.0], 2050-51 90.0 87.7 90.0 87.7

 Years
Net internal, 2005-06 27.7 16.0 29.3 26.0
Net internal, 2025-26 17.5 16.0 26.5 26.0
Net internal, 2050-51 11.7 16.0 24.5 26.0
Long-run net 12.0 13.7 16.0 19.5
international

Results

Total populations Thousands
Population, 2006 3,986 3,936 4,016 4,000
Population, 2016 4,521 4,374 4,678 4,667
Population, 2026 4,906 4,757 5,289 5,305
Population, 2051 5,275 5,173 6,470 6,430

Components of change Thousands
Natural change, 2005-06 20.3 19.7 22.4 23.1
Natural change, 2015-16 12.4 11.6 20.2 20.4
Natural change, 2025-26 3.2 5.1 16.2 15.3
Natural change, 2050-51 -23.3 -26.0 -2.2 -12.1
Net migration, 2005-06 45.4 29.7 48.6 45.5
Net migration, 2015-16 34 29.7 43.7 45.5
Net migration, 2025-26 29.5 29.7 42.5 45.5
Net migration, 2050-51 23.7 29.7 40.5 45.5

Age structure Years
Median age 2026 42.6 43.9 41.7 42.2
Median age 2051 49.1 50.3 47.3 46.8

Population Percentages
Population aged 65+, 20.3 22.2 20 20.7
2026
Population aged 65+, 28.3 29.8 27.7 26.8
2051

 High
 Old Govt ABS A

Assumptions
Long-run TFR 1.80 1.84

 Years
Male [e.sub.0], 2050-51 87.8 92.0
Female [e.sub.0], 2050-51 90.0 95.0

 Years
Net internal, 2005-06 30.7 36.0
Net internal, 2025-26 35.9 36.0
Net internal, 2050-51 38.4 36.0
Long-run net 20.0 24.4
international

Results

Total populations Thousands
Population, 2006 4,048 4,059
Population, 2016 4,838 4,955
Population, 2026 5,685 5,878
Population, 2051 7,767 8,094

Components of change Thousands
Natural change, 2005-06 24.5 26.4
Natural change, 2015-16 28.3 31.1
Natural change, 2025-26 30.1 31.7
Natural change, 2050-51 24.3 25.3
Net migration, 2005-06 51.9 60.4
Net migration, 2015-16 53.5 60.4
Net migration, 2025-26 55.9 60.4
Net migration, 2050-51 58.4 60.4

Age structure Years
Median age 2026 40.8 40.9
Median age 2051 45.6 45.7

Population Percentages
Population aged 65+, 19.7 19.9
2026
Population aged 65+, 27.1 27.6
2051

Sources: Queensland Government population projections 2003 and ABS,
Population Projections Australia: 2002-2101, Cat. no 3222.0, Canberra,
2003.
Notes: [e.sub.0] = Life expectancy at birth. ABS projections assume
fixed net internal migration levels. The Queensland Government
projections exhibit gradually changing net internal migration numbers.
Those listed in the table are for selected years.

Acknowledgement

The Queensland Government 2003 population projections were produced through a collaborative research agreement between the Queensland Government Office of Economic and Statistical Research (OESR) and the Queensland Centre for Population Research (QCPR). Funding was provided by OESR and The University of Queensland.

The views expressed in this article are those of the authors and do not necessarily represent the official position of the Queensland Government.

References

(1) Australian Bureau of Statistics (ABS), Australian Historical Population Statistics, Cat. no. 3105.0.65.001, Canberra, 2003; ABS, Population by Age and Sex Australian States and Territories, Cat. no. 3201.0, Canberra, 2003

(2) Queensland Government, Queensland Government Population Projections 2003, Brisbane, 2003. Details are available from http://www.oesr.qld.gov.au/releases/populationprojections

(3) United Nations, World Population Prospects: the 2002 Revision Highlights, United Nations, New York, 2003

(4) Following standard demographic practice we use the term 'internal migration' to refer to migration within Australia and 'international migration' to describe migration exchanges with other countries. A distinction between 'interstate' and 'intrastate' internal migration is made. The terms 'in-migration' and 'out-migration' are used respectively to describe inward and outward internal migration, whilst inward international migration flows are 'immigration' and outward flows 'emigration'.

(5) A. Rogers, 'Requiem for the net migrant', Geographical Analysis, vol. 22, 1990, pp. 283-300

(6) Given its many advantages it is surprising that the bi-regional reduction of the multiregional model is accorded so little attention in the academic literature. The few exceptions include: A. Rogers, 'Shrinking large-scale population projection models by aggregation and decomposition', Environment and Planning A, vol. 8, 1976, pp. 515-541; A.M. Isserman, 'The right people, the right rates: making population estimates and forecasts with an interregional cohort component model', Journal of the American Planning Association, vol. 59, 1993, pp. 45-64; S.K. Smith, J. Tayman and D.A. Swanson, State and Local Population Projections: Methodology and Analysis, Kluwer Academic, New York, pp. 142-151; and T. Wilson and M. Bell, 'Experiments with different internal migration models in subnational population projections', Queensland Centre for Population Research Discussion Paper, School of Geography, Planning and Architecture, The University of Queensland, 2002.

(7) 'Population projections', Demographic Estimates and Projections: Concepts, Sources and Methods, Canberra, ABS, chapter 6, 1995

(8) Wilson and Bell, op. cit.

(9) A. Taylor, 'Fertility and mortality', Queensland Government Population Projections Background Research paper no. 2, Office of Economic and Statistical Research, Queensland Treasury, 2003

(10) G.A. Carmichael and P. McDonald, 'Fertility trends and differentials', in S. Khoo and P. McDonald (Eds) The Transformation of Australia's Population 1970-2030, Sydney, University of New South Wales Press, 2003, chapter 3, pp. 40-76

(11) Data from T. Wilson and P. Rees, 'Why Scotland needs more than just a new migration policy', Scottish Geographical Journal, 2003, forthcoming

(12) G. Heyen, 'Assumptions used in the Queensland Government population projections to 2051', Queensland Government Population Projections Background Research paper no. 1, Office of Economic and Statistical Research, Queensland Treasury, 2003

(13) Australian Institute of Health and Welfare (AIHW), 'Changes in Australia's disease profile: a view of the twentieth century', Australia's Health 2000, AIHW, Canberra, 2000, chapter 8, pp. 340-364; A. Taylor, op. cit.

(14) H. Booth, 'The changing dimensions of mortality', in S. Khoo and P. McDonald (Eds) The Transformation of Australia's Population 1970-2030, Sydney, University of New South Wales Press, 2003, chapter 5, pp. 104-128

(15) Authors' calculations based on ABS mortality counts and estimated resident populations

(16) V. Kannisto, 'Mode and dispersion of the length of life', Population: An English Selection, vol. 13, 2001, pp. 159-172. Using a mortality database for 13 industrialised countries judged to have high quality data, the author shows that the increasing modal age of death as measured by the life table function [.sub.n][d.sub.x] has been accompanied by a gradual compression of deaths around the mode. Kannisto states that 'this may be interpreted to mean that the prolongation of life is meeting increasing resistance' (p. 169). Life tables calculated for 1971-2001 indicate that this modal age at death--compression relationship also holds for Australia.

(17) A. Rogers, R. Raquillet and L.J. Castro, 'Model migration schedules and their applications', Environment and Planning A, vol. 10, 1978, pp. 475-502

(18) Net international migration calculated as a population accounting equation residual is the number remaining when natural change and net internal migration are subtracted from total population growth.

(19) See http://www.oesr.qld.gov.au/releases/populationprojections/.

(20) The decomposition of the rate of population growth into the crude rates of natural increase and net migration is described in S.H. Preston, P. Heuveline and M. Guillot, Demography: Measuring and Modeling Population Processes, Oxford, Blackwell, 2001, p. 8.

(21) The Crude Net Migration Rate (CNMR) defined in this way does, of course, contain a problematic population 'at risk', because for in-migration the statistical division's population is precisely the population not at risk of in-migration. Whilst conceptually impure and inappropriate for projections, the CNMR is nonetheless a useful measure for describing the rate at which a population is changing size due to migration.