Alternative projection series
Nine official series of population projections have been produced by Statistics New Zealand using different combinations of fertility, mortality and migration assumptions (Table 1). A further five series are presented below in the Additional 'what if?' scenarios section and in Table 4. The projections are neither predictions nor forecasts. They provide an indication of possible future changes in the size, growth rate and age-sex structure of the population. The population projections have as a base the estimated resident population of New Zealand at 30 June 2004. These are the second set of national population projections to use data from the 2001 Census, and cover the period to 2051 at one-year intervals.
Among the official series, projection series 9, which assumes high fertility (a long-term average of 2.10 births per woman), low mortality (life expectancy at birth increasing to 86.0 years for males and 89.0 years for females by 2051), and long-term annual net migration of 15,000 people, yields the highest population in 2051 – 5.90 million. This is an increase of 1.83 million or 45 percent from 2004. Series 1, which assumes low fertility (a long-term average of 1.60 births per woman), high mortality (life expectancy at birth increasing to 81.0 years for males and 85.0 years for females by 2051), and long-term annual net migration of 5,000 people, projects the lowest population in 2051 – 4.26 million. Series 1 projects the population to peak at 4.46 million in 2033, before deaths begin to outnumber the combined effect of births and net migration and the population decreases for the remainder of the projection period. Series 2 (low fertility), series 3 (high mortality) and series 4 (low migration) also project the population to start declining before 2051.
Which projection series should I use?
A number of projection series have been produced to illustrate a range of possible scenarios. Users can make their own judgement as to which projection series is/are most suitable for their purposes. However, at the time of release, Statistics New Zealand considers mid-range projection series 5 the most suitable for assessing future population changes. Series 5 assumes:
- Fertility: the total fertility rate will decrease to 1.85 births per woman by 2016 and then remain constant.
- Mortality: life expectancy at birth will increase to 83.5 years for males and 87.0 years for females by 2051.
- Migration: there will be a long-term annual net migration gain of 10,000 people from 2009.
The following analysis is based on series 5, unless otherwise stated. In graphs, a break in data between 1990 and 1991 denotes a change from the de facto population concept to the resident population concept.
What has changed from the 2001-base projections?
The base population at 30 June 2004 is 4.06 million. This is 18,000 or 0.5 percent higher than the 4.04 million projected at 30 June 2004 from the 2001-base projections (series 4 and 5), because observed net migration and births were higher than projected.
The medium-migration variant of the 2004-base projections assumes a net migration gain of 104,000 in the five years to 30 June 2006 and 38,000 in the five years to 30 June 2011. By comparison, the corresponding variant of the 2001-base projections assumed net migration gains of 100,000 and 25,000, respectively.
In the 2004-base projections, long-term annual net migration levels of 5,000, 10,000 and 15,000 are used as the low, medium and high migration variants, respectively. Previously, alternative migration variants of 0, 5,000, 10,000 and 20,000 were used, with 5,000 considered the most suitable long-term annual net migration level. While net migration has averaged about 5,000 during the last 50 and 100 years, a net migration level of 10,000 (the average during the last 15 years) better reflects recent and likely future migration trends. Changes in immigration policy during the last 20 years are a key factor behind the increase in average annual net migration. Sustained long-term net flows of 0 and 20,000 people per year are considered unlikely.
The medium-fertility variants of both the 2001-base and 2004-base projections assume that the total fertility rate will decline to 1.85 births per woman during the projection period, and then remain constant. However, the 2004-base projections assume it will take longer for the total fertility rate to decline to this long-term level. The new assumptions incorporate the recent rise in the total fertility rate, from 1.90 in the year ended June 2002 to 2.01 in the year ended June 2004 (based on births by date of registration). The medium variant of the 2001-base projections assumed the total fertility rate would fall to 1.89 births per woman in 2006, then to 1.85 births per woman by 2011. The medium variant of the 2004-base projections assumes that the total fertility rate will fall to 2.00 births per woman in 2006, 1.92 births per woman in 2011, then to the long-term level of 1.85 births per woman by 2016.
The medium- and low- mortality variants of the 2004-base projections assume that male life expectancy at birth will reach 83.5 and 86.0 years, respectively, in 2051. The corresponding assumptions for females are 87.0 and 89.0 years, respectively. These are slightly higher life expectancies than assumed for the 2001-base projections, and give a wider range between the low and high variants in 2051: 5 years for males and 4 years for females. Previously, the range between the low and high assumptions was 4 years for males and 3 years for females in 2051.
The combined effect of these changes is that the New Zealand population is expected to reach 4.73 million in 2026 and 5.05 million in 2051 (series 5, 2004-base projections). The five million population mark is projected to be reached in 2041. By comparison, under series 4 of the 2001-base projections the New Zealand population was expected to reach 4.62 million in 2026 and peak at 4.81 million in 2046, before declining slightly to just under 4.81 million in 2051.
Population growth
The New Zealand population grew at an average rate of 1.4 percent a year between 1951 and 2004. Under projection series 5, the population is projected to grow by an average of 0.8 percent a year between 2004 and 2011. The growth rate is expected to slow over the remainder of the projection period, to almost zero by 2051. In the last 10 years of the projection period (2041–2051), population growth is projected to average just 0.1 percent a year.
Births and deaths
The slowing of population growth during the projection period reflects the convergence of births and deaths. Under projection series 5, births are expected to decrease from 58,000 in 2004 to 50,000 in 2051. This is due to the combined effect of the decreasing average number of births per woman and there being fewer women in the childbearing ages. In contrast, deaths are expected to increase dramatically from 28,000 in 2004 to 59,000 in 2051, despite increasing life expectancy. This is because of the increasing number of people born after World War II reaching the older ages. About 73 percent of male deaths and 82 percent of female deaths currently occur at age 65 years and over.
Under the low-fertility projection (series 2), which assumes a long-term average of 1.6 births per woman, births are expected to fall to 38,000 a year in 2051. If low fertility is also combined with low migration and high mortality (series 1), births are projected to fall to just 34,000 a year. In contrast, the high-fertility projection (series 8), which assumes replacement fertility of 2.1 births per woman in the long term, shows births increasing to 63,000 in 2051. Series 9, which assumes high migration and low mortality as well as high fertility, projects that births will increase to 69,000 in 2051. There is much less variation in the number of deaths between projection series, ranging from 56,000 deaths in 2051 under the low-mortality projection (series 7) to 61,000 under the high-mortality projection (series 3).
Natural increase (excess of births over deaths) is projected to decline steadily from 30,000 in 2004 to just above zero in 2041. From 2042, deaths are expected to outnumber births by a growing margin. By 2051, series 5 projects 9,000 more deaths than births. Only in the projection series assuming high fertility (series 8 and 9) do births continue to outnumber deaths by 2051.
Ageing population
Regardless of which projection series is chosen, there will be significant changes in the age structure of the population. All series project more older people and further ageing of the population. All series, except the high-fertility series 8 and 9, also project fewer children.
The median age (half the population is older, and half younger, than this age) of New Zealand's population increased from 26 years in 1971 to 35 years in 2004. According to projection series 5, half of the population will be 40 years and older by 2020, and half the population will be 45 years and older by 2045. This reflects the combined impact of sub-replacement fertility, increasing longevity, and the movement of the large numbers of people born after World War II into the older ages.
Higher migration levels appear unlikely to significantly slow the ageing of the population. The median age of the population in 2051 is projected to be 45 years assuming net migration of 15,000 a year (series 6), 46 years assuming net migration of 10,000 a year (series 5), and 47 years assuming net migration of 5,000 a year (series 4). Fertility has a much more significant impact on the age structure, with the lowest median age in 2051 (43 years) provided by the high fertility series 8, and the highest median age in 2051 (49 years) provided by the low fertility series 2.
Children
The number of children (aged 0–14 years) peaked at 940,000 in 1974, then decreased steadily to 770,000 in 1989, before increasing again to 890,000 in 2004. Under projection series 5, the number of children is expected to gradually decline over the projection period, to 820,000 in 2021 and 790,000 in 2051. This mainly reflects the impact of sustained sub-replacement fertility. As a result, children will account for only 16 percent of the population by 2051, compared with 22 percent in 2004 and about 33 percent during the early 1960s.
Series 1 projects the fewest children in 2051, with a decline to 580,000. In contrast, series 9 projects almost twice as many children in 2051 as series 1, with the number increasing to 1.04 million. Alternative fertility assumptions have the largest impact on the number of children, with 630,000 projected in 2051 assuming low fertility (series 2), and 960,000 projected assuming high fertility (series 8). In comparison, the range in the number of children given by the alternative migration assumptions is from 730,000 (series 4) to 850,000 (series 6). Differences in mortality assumptions have a negligible effect on the number of children.
Working-age population
The working-age population (those aged 15–64 years) more than doubled from 1.20 million in 1951 to 2.69 million in 2004. Over the next 20 years, it is projected to grow by 11 percent to peak at 2.98 million in 2024. It will then decline slowly to 2.93 million by 2051. The working-age population will make up 58 percent of the total population in 2051, compared with 66 percent in 2004.
Most of the increase in the working-age population will be in the older half of the age bracket, and thus it will take on an older profile. The population aged 40–64 years has been increasing rapidly since the early 1980s as the baby-boomers move into this age group. This will continue for the next two decades, with the number of people in this age group expected to increase from 1.26 million in 2004 to 1.51 million in 2023. In 2051, 1.55 million people (31 percent of the population) will be aged 40–64 years. This is the same percentage as in 2004, although the proportion will exceed 33 percent between 2010 and 2020.
The number of people aged 15–39 years is projected to fluctuate only slightly over the projection period, between a peak of 1.47 million in 2026 and a low of 1.38 million in 2051. This age group accounted for about 41 percent of the population in the mid-1980s and 35 percent of the population in 2004, but is expected to account for only 31 percent in 2026 and 27 percent in 2051. The ratio of the 40–64 year age group to the 15–39 year age group is projected to rise from 0.88 in 2004 to 1.12 in 2051, with the 40–64 year age group surpassing the 15–39 year age group in size by 2011. The ratio was 0.57 in 1984.
Older people
The number of people aged 65 years and over has doubled since 1970, to 490,000 in 2004. Projection series 5 indicates that there will be 1.33 million people aged 65 years and over by 2051, 2.7 times the 2004 total. The largest growth will occur between 2011 and 2037 as the baby-boomers move into this age group. From 2039, the 65 years and over age group will make up about one-quarter of all New Zealanders, compared with 12 percent in 2004.
Within the 65 years and over age group, the number of people aged 85 years and over is expected to increase from 54,000 in 2004 to about 320,000 in 2051. They will make up 24 percent of the 65 years and over population, compared with 11 percent in 2004. The number of people aged 85 years and over is projected to increase substantially, even under the high-mortality projection (series 3), to 260,000 in 2051. Under the low-mortality projection (series 7), there would be 390,000 people aged 85 years and over in 2051.
Although more males are born than females (at a ratio of about 105 to 100), the higher mortality of males at all ages means that females outnumber males by a growing margin in the older ages. In 2004, there were 79 males for every 100 females in the 65 years and over age group. At ages 85 years and over there were only 44 males for every 100 females. The projections assume some catch-up in male life expectancy relative to that of females. By 2051 there will be 85 males for every 100 females aged 65 years and over. For ages 85 years and over, the sex ratio is expected to increase to 70 per 100 by 2051.
Dependency ratios
Dependency ratios relate the number of people in the 'dependent' age groups (defined here as 0–14 years and 65 years and over) to the working-age population (15–64 years). Although commonly used, they are crude measures because they do not allow for the fact that some people aged 15–64 years may not be in the workforce, while some people aged 65 years and over may still be in the workforce.
The 0–14 and 65+ dependency ratios are projected to follow contrasting paths over the projection period. The 0–14 dependency ratio (the number of people aged 0–14 years per 100 people aged 15–64 years) decreased from a peak of 57 per 100 in 1961 to 33 per 100 in 2004. This downward trend is expected to continue, with the ratio dropping to 27 per 100 in 2023, then remaining about this level for the rest of the projection period.
The 65+ dependency ratio (the number of people aged 65 years and over per 100 people aged 15–64 years) increased from 15 per 100 in 1951 to 18 per 100 in 2004. It is expected to increase substantially after 2011 to reach 45 per 100 in 2051. This means that for every person aged 65 years and over, there will be 2.2 people in the working-age group in 2051, compared with 5.5 people in 2004. The 65+ dependency ratio is projected to overtake the 0–14 dependency ratio by 2022.
The total dependency ratio (sum of the 0–14 and 65+ dependency ratios) is projected to increase from 51 per 100 in 2004 to 72 per 100 in 2051 under series 5. All projection series show a similar increase in the total dependency ratio. By 2051, the 65+ dependency ratio will contribute 63 percent of the total dependency ratio. A similar total dependency ratio to that projected for 2051 was experienced around 1960, but then the 0–14 dependency ratio contributed about 80 percent of the total dependency ratio.
Additional 'what if?' scenarios
While the nine official series have been formulated to cover the likely range of each component (mortality, fertility, net migration), five additional projection series have been derived to explore other scenarios of interest.
Under projection series 5, the population increases by about one million people between 2004 and 2051, to just over five million. For the population to reach six million by 2051, either a total fertility rate (TFR) of 2.5 births per woman would be required ('very high' fertility series, also assuming medium mortality and medium migration) or net migration would need to average 25,000 a year ('very high' migration series, also assuming medium fertility and medium mortality). While these scenarios produce a similar population size, the higher fertility assumption would produce a much younger age structure.
With a total fertility rate of 2.5 births per woman, births would continue to outnumber deaths by 29,000 or more throughout the projection period. There would be 88,000 births in 2051 under this scenario, compared with 50,000 under projection series 5. The increased number of births would see the number of children increase by 44 percent over the projection period, compared with a decrease of 11 percent under series 5. Population ageing would continue but at a much slower rate, with the median age increasing from 35 years in 2004 to 39 years in 2051. Series 5 projects the median age to increase to 46 years in 2051.
Net migration of 25,000 a year would also slow the ageing of the population, but much less than the very high fertility level. The median age is projected to increase to 44 years in 2051 with very high net migration – just two years below that for series 5. Because of the higher population, there would be 26 percent more births and 7 percent more deaths in 2051 than projected under series 5. As a result, the final year of the projection period (2051) would be the first year that the number of deaths would exceed the number of births, compared with 2042 for series 5.
The nine official projection series all assume that the gain in life expectancy will slow over the projection period. If gains seen over the last two decades continue, life expectancy at birth would reach 91.3 years for males and 92.9 years for females by 2051. This is a 71 and 65 percent reduction in male and female mortality rates, respectively, over the projection period. Under this mortality assumption, the population would reach 5.31 million in 2051. This is 260,000 more than under series 5, which uses the medium mortality assumption (life expectancy at birth increasing to 83.5 years for males and 87.0 years for females in 2051). Almost 240,000 of this difference would be in the 65 years and over age group, which would more than triple in size, from 490,000 in 2004 to 1.56 million in 2051. The 85 years and over age group would increase, from 54,000 in 2004 to about 480,000 in 2051 – about 160,000 more than under projection series 5. With more people in the older ages, the population would age even faster than projected by series 5 – the median age of the population reaching 48 years in 2051. The 65+ dependency ratio would also be higher, reaching 53 per 100 in 2051 compared with 45 per 100 under series 5.
An interesting projection series for comparative purposes is to assume no migration. This shows the effect on the population solely from births and deaths. With no migration, the population would peak at 4.68 million in 2041 then decline as deaths outnumber births. The population would decrease to 4.63 million in 2051, still 570,000 higher than the 2004 population. Compared with series 5, the population would be lower for all three broad age groups (0–14 years, 15–64 years and 65 years and over) and as a result the median age and dependency ratios would be very similar.
In the long-term, the low, medium and high migration assumptions assume a constant migration level. However, actual net migration tends to fluctuate significantly year to year. A projection series was constructed assuming cyclic migration levels in the long-term, fluctuating between -10,000 and +30,000 on a 10-year cycle. This gives exactly the same net migration gain over the projection period as series 5. The resulting population size in 2051 is little different under the cyclic migration series than under projection series 5, being just 13,000 higher. However, during the projection period the population varies between 5,000 below and 67,000 above the series 5 population in the same year, because of the short-term differences in net migration. Other characteristics of the population (eg age distribution, dependency ratios, births, deaths) are very similar between the two series. A constant level of migration in the long-term is therefore a sufficient assumption for most purposes.
For technical information contact:
Kim Dunstan or Richard Speirs
Christchurch 03 964 8700
Email: demography@stats.govt.nz