Urban Place Analysis

Data to 2013 are drawn from the Royal Society Marsden project The ebbing of the human tide. What will it mean for the people? (also known as 'Tai timu tangata. Taihoa e?''). The project was undertaken by researchers from the University of Waikato, Massey University and Motu between 2014 and 2016. Data for the period 2013-2018 were generated by Natalie Jackson as part of the CaDDANZ project (funded by MBIE), using the same methodology.

These maps and graphs provide a long-term overview of the components (drivers) of population change across New Zealand's urban places as defined by Statistics New Zealand prior to their recent shift to SA2 geographic classification; the data thus provide a unique and comprehensive historical picture of urban area change across the period 1976-2018. Where relevant, the data have been annualised. The methodological notes below summarise how this data set was constructed. The CaDDANZ project explores how these data can be visualised and made more accessible and useful to end users.

Urban Places (Maps and Graphs)

The ‘change by component’ and ‘component comparison’ maps represent simultaneously both annual (average) percentage change, and change in absolute size, by using a mix of symbols – colour and size. Maps for net migration, natural change, and total population change are provided either separately (e.g., net migration across the entire period) or collectively (e.g., net migration, natural change and total change for each single period) by choosing the relevant category from the list. (Natural Change refers to both the difference between births and deaths, which is conventionally referred to as 'natural increase' or 'natural decrease', and to change in the size of each cohort as younger cohorts move into the next age group.)

Most of the spatial variation in population change is explained by net migration, because it differs markedly for each urban place, while natural change is relatively even across New Zealand. Although some towns are already experiencing natural decrease (where deaths exceed births), at this stage the decrease is relatively small and plays a minor role in overall size.

The spatial variation patterns of net migration are more complex to understand and are driven by economic and employment factors as well as lifestyle choices, such as climate and topography, and access to essential services – hospitals, universities and airports. These spatial patterns in net migration vary with different age groups as can be seen in the maps. Younger people are moving away from small agricultural towns to larger cities with universities and employment options, as well as tourist towns. People approaching retirement age are moving out of large cities to lifestyle towns but still want to be close to large population centres and international airports. See the readings below for more information.

The ‘change by component’ graphs use the same data to show how the components of change interact to determine total population change. For most urban places, overall natural increase (births minus deaths) is tracking downwards as birth numbers fall and death numbers rise (the result of population ageing); however, this situation differs by age group. For most towns and rural centres, natural change below age 54 has been diminishing, as smaller, later-born (Gen X, Y and millennial) cohorts replace larger earlier-born ones (the baby boomers), while that above 55 years of age is increasing, as baby boom cohorts replace—and in many cases join—their smaller cohort parents. By contrast, the net migration trends for each age group vary considerably by urban area, although net loss at 15-24 years is historically widespread, while areas gaining people aged 25-39 and 40-54 years invariably see an accompanying net migration gain at 0-14 years, their children. This information can be used to determine what age groups have been/are attracted to a town or conversely what age groups have moved/are moving away. Town planners can use this information to plan and strategize accordingly

The ‘summary’ graphs focus on the impact of structural ageing (showing how numbers by age are ‘rippling’ across the age structure) and net migration by age (to produce ‘net migration age profiles’). Where population numbers are small the graphs tend to be ‘messy’ and viewers may not see any clear picture emerging. Where numbers are sufficient, the top panel provides a clear illustration of the process of population ageing, where numbers at the youngest ages were highest (or near-highest) in 1976 and have fallen since, and numbers at older ages are successively higher at each successive census. The picture is somewhat different where areas have grown significantly across all ages, such as Rolleston. The middle and lower panels showing net migration numbers and rates by age tend to show fairly consistent patterns over time, at least where numbers are not too small. Most graphs show migration concentrated among the younger age groups—see for example Dunedin, Palmerston North and Wellington, which have among the most consistent migration age profiles of this type. Fewer have migration concentrated to the right, where there is a lot of migration around retirement age—see for example Paeroa, Pauaunui Beach, Thames, Waihi Beach, Warkworth and Winton.

The ‘with/without migration’ graphs use the above data to show the impact of migration on population change per se. The methodology (see also below) applied fertility and survivorship rates to each base population (separately by age and sex) for each census year (time t) to derive an ‘expected’ population at t+5 years. The expected population was then compared with the actual population at t+5 years to derive a residual measure for migration (net migration = net change – natural increase). For ‘size with migration’, each successive base population included the compounding effects of fertility, survivorship and migration. For ‘size without migration’, each successive base population included the compounding effects of fertility and survivorship only.

Urban Places ('Sea Change'1) Methodology

The data were generated using three main sets of data originally sourced from Statistics New Zealand:

  • Census Usually Resident Population counts by age, sex and meshblock for the census years 1976, 1981, 1986, 1991, 1996, 2001, 2006, 2013 and 2018;
  • Age-Specific Fertility Rates by age and territorial authority area of mother, for the years 1997, 2001, and 2006-2018 inclusive; and
  • Survivorship Rates by age, sex, and territorial authority area for the life table periods 2005-2007 and 2012-2014.

All data were re-based to 2013 geographic boundaries. In particular it should be noted that the data for the 2018 census were generated to match the urban area geographic classification existing for the 1976-2013 period, prior to its reclassification in 2018.

Population data: Mesh-block level counts of the usually resident population by sex and 5-year age group (to 80+ years) for all census years 1976-2013 were sourced from Statistics New Zealand and aggregated to the 2013 geographic area boundaries at urban area (UA) level, based on their urban area and rural centre status in 1976. The allocation to 2013 geographic areas was based on a ‘user-derived correspondence’ for the purposes of the Tai Timu Tangata. Taihoa e? project. The counts are not official statistics but should be thought of experimental estimates intended for use in research.2 This exercise resulted in data for 143 urban areas and 132 rural centres.

Data for the 2018 Census Usually Resident Population (by age and sex) was purchased from Statistics New Zealand (2020a) pre-aggregated to the above urban area classifications.

Birth and survivorship rates for all years for which these data were required (i.e., back to 1976) are not available at urban area or rural centre level and were instead constructed using indirect standardisation. In order to construct birth rates, two customised datasets of births by 5-year age group and territorial authority area of mother were purchased from Statistics New Zealand (2016, 2020b), covering the periods 1997-2013 and 2013-2018 (June years) at 2013 geographic boundaries. Survivorship (Lx)3 rates by age and sex for each territorial authority area were accessed for the years 2005-07 and 2012-14 (the only years for which these data are available) (Statistics New Zealand 2015a).

Calculating birth rates for missing years via indirect standardisationwas done in two main steps. First, age-specific fertility rates were constructed for each of New Zealand’s 67 territorial authority areas (TA) for the June years 1996-97, 2001-02, and 2006-2018 inclusive, using number of births by age of mother as sourced above, and female estimated resident population counts for corresponding 5-year age groups 15-49 years sourced from Statistics New Zealand (various years). The age-specific fertility rates for 1996 and 2001 were then summed and averaged (for each age group and each TA), and their ratio to the equivalent rates for total New Zealand constructed. These relative age-specific fertility ratios for each TA were then held constant and multiplied by the equivalent rates for total New Zealand for the missing years, 1976, 1981, 1986, and 1991. That is, the national values were retrospectively inflated or deflated by the relevant ratio, for each of the four observations 1976-1991, to generate approximate TA level age-specific rates for those years.

The second step involved constructing age-specific fertility rates for each town and rural centre, by applying the age-specific rates for the TA in which each is located to the number of women in each 5-year age group 15-49 years, in each town and rural centre (from Database 1).

The resulting birth rates and numbers at TA level differ slightly from those published by Statistics New Zealand because they are constructed experimentally. As with the underlying population counts, the fertility rates and resulting birth numbers should be thought of as best approximations extracted for these research purposes.

Calculating missing survivorship rates via indirect standardisation was similarly done in two steps. First, Lx values (the average number alive in each age group) by 5-year age group and sex for each TA for two Life-Table periods, 2005-07 and 2012-14, were compared to the average number alive in the preceding 5-year age group. This process produced sex- and age-specific survivorship ratios for each five-year age group to 95+ years, for these two observations (for the purposes of this exercise, considered to be 2006 and 2013). The 2006 ratios were then compared with their national equivalents, to generate relative survivorship ratios for each TA for the missing years: 1976, 1981, 1986, 1991, 1996, and 2001. That is, for each of those observations, the national values were retrospectively inflated or deflated by the relevant sex- and age-specific survivorship ratios for each TA in 2005-2007, to generate approximate TA level rates.

The second step involved constructing sex- and age-specific survivorship rates for each town and rural centre, by applying the rates for the TA in which each is located, to the number of males and females in each five year age group, in each town and rural centre (Database 1). In order to survive age groups above 80 years, the 80+ year age group from Database 1 was prorated to 80-84, 85-89, 90-94 and 95+ years according to the New Zealand distribution (by sex) at those ages.

Again, the resulting data are ‘best approximations’ based on calendar year survivorship ratios and census usually resident population counts.

When the resulting data are compared with published birth and death numbers for each TA, which are available for all years 1992-2018, there is strong correspondence, and the model is thus considered sufficiently robust to use for our purpose of calculating the components of change for towns and rural centres. This is done using a standard cohort component methodology and the ‘residual’ method for separating net migration from net change (e.g., Rowland 2003, Chapter 12).

Calculating components of change by the residual method: The resulting fertility and survivorship rates were used in a conventional cohort component analysis to separate out the contributing effects of net migration and natural increase. First, survivorship rates for each age group were applied to the baseline Census Usually Resident Population count for each individual observation (separately by sex), and fertility rates applied to the survived women aged 15-49 years. The resulting births were summed and apportioned male/female according to the standard sex ratio for New Zealand (105 males per 100 females). Births were entered at age 0-4 years, and all other age groups ‘aged’ by five years. The resulting ‘expected’ population by age and sex was then compared to the actual population at the relevant observation (for example, the survived and ‘reproduced’ population from 1976 was compared to the actual population for 1981), and the difference at each age (five-year age group) taken to be a residual measure of net migration by age across the five-year period. Subtracting total estimated migration from net change in population size between the two observations in turn generates the natural increase component, which in turn is disaggregated into its births and deaths components by summing each individual component generated at each step.

  • 1. The ‘Sea Change’ methodology was developed by Natalie Jackson at the University of Tasmania in 2001, as a way of separating the effects of change in cohort size and net migration at older ages, for Tasmania’s then-29 local government areas.
  • 2. [Disclaimer: Access to these data was provided by Statistics New Zealand under conditions designed to give effect to the security and confidentiality provisions of the Statistics Act 1975. The results presented in these tables are the work of the author/s, not Statistics New Zealand.]
  • 3. [Lx values are a statistical function of the Life Table, via which life expectancy is calculated.]

  • Sea Change Methodology (Summary)
  • 1. Numbers by sex and five year age group for each Urban Place are assembled, for each census year (time t)
  • 2. Survivorship rates by age and sex are applied to base numbers by age and sex
  • 3. Fertility rates for women aged 15-49 years are applied to survived numbers of women aged 15-49 years, and the resulting number of births split male: female according to New Zealand's sex ratio at birth (105:100)
  • 4. Births are entered at age 0-4 years and the population aged by 5 years
  • 5. The 'new' (survived and reproduced) population = the 'expected' population at time t+5
  • 6. The expected population at time t+5 (by age and sex) is compared with the ‘actual’ population at time t+5 (by age and sex) and the difference taken to approximate net migration
  • 7. The process is repeated for each year of the projection

Further reading

Brabyn L (2017) Declining Towns and Rapidly Growing Cities in New Zealand- developing an empirically-based model that can inform policy. Policy Quarterly. 13, 37-46. http://igps.victoria.ac.nz/publications/PQ/2017/PQ-Vol-13-Supplementary-2017.pdf.

Jackson NO, L Brabyn, D Maré, MP Cameron and I Pool (2019) From ageing-driven growth towards the ending of growth. Subnational population trends in New Zealand, in J Anson, W Bartl, A Kulczycki (Hg.) Studies in the Sociology of Population. International Perspectives. Switzerland: Springer Nature. pp 161-193.https://doi.org/10.1007/978-3-319-94869-0

Jackson NO and L Brabyn (2017) The mechanisms of subnational population growth and decline in New Zealand, 1976-2013’ Policy Quarterly Supplement 13: 22-36. http://igps.victoria.ac.nz/publications/PQ/2017/PQ-Vol-13-Supplementary-2017.pdf.

Jackson, NO, L Brabyn and D Maré (2016) New Zealand’s towns and rural centres 1976-2013 – experimental components of growth’. Working Paper No. 7. National Institute of Demographics and Economic Analysis, University of Waikato, Hamilton, New Zealand. http://www.waikato.ac.nz/__data/assets/pdf_file/0017/324134/No7-New-Zealands-towns-and-rural-centres-1976-2013.pdf

References and Data Sources

Rowland DT (2003) Demographic Methods and Concepts Oxford: Oxford University Press.

Statistics New Zealand (2020a) Customised database. Census Usually Resident Population Counts by Age, Sex, and Urban Area (User-Specified).

Statistics New Zealand (2020b) Customised database. Estimated birth occurrence, at 30 June 2013-2018 by Age of Mother and Territorial Authority Area of Usual Residence. JOB 10119.

Statistics New Zealand (2016) Customised database. Estimated birth occurrence, at 30 June 1997-2013 by Age of Mother and Territorial Authority Area of Usual Residence.

Statistics New Zealand (2015) Life Table Functions 2005-07 and 2012-14, by Territorial Authority Area, age and sex.

Statistics New Zealand (various years) Subnational population by age and sex, June 1996, 2001, 2006, 2013, 2018.