The rise in birth numbers in 2006–08 has raised questions about the accuracy of official population projections and the ability of demographers to anticipate population trends. This article addresses those questions, but also encourages users of population projections to acknowledge uncertainty and consider alternative scenarios to the mid-range projection.
New Zealand’s population is continually changing: the number of births is currently at levels last seen in the early 1960s; life expectancy has never been as high; and the net outflow of New Zealand residents to Australia is at record levels. These are just some of the demographic data released by Statistics New Zealand in recent months (see 'more information' link above).
Statistics NZ also produces population projections, and these have come under recent spotlight as an upsurge in births has surprised many. Births increased by 11 percent from 57,700 in 2005 to 64,300 in 2008.
Underlying this rise in births is increasing fertility rates for women at all childbearing ages. Notably, fertility rates have increased among women aged 15–29 years, whereas rates for this age group had previously been decreasing (see figure 1). Also, the increase since 2005 is not just confined to New Zealand; fertility rates are increasing in several Organization for Economic Co-operation and Development (OECD) countries, including Australia and the United Kingdom.
In truth, demographers are uncertain as to the drivers of short-term changes in fertility rates. Moreover, there is no early warning system to advise when births are about to swing up or down. Clearly, births reflect decisions of couples and mothers, and individual and family circumstances. But underlying these factors are various social and economic conditions. If couples are more likely to have children when their social, economic and political environment is most stable, then financial security and job security might be conducive to childbearing.
New Zealand’s fertility rates have been falling over the long-term. The upswing in births in 2007–09 probably reflects changes in the timing of births rather than a profound shift to larger family sizes. Women born in the 1930s averaged 3.5 births each during their lifetime; women born in the early 1950s about 2.5 births each; women born in the early 1960s about 2.3 births each; and women born in the early 1970s will average about 2.2 births each (cohort fertility tables from Demographic Trends). Similarly, one in twelve women born in the 1930s remained childless, compared with one in nine women born in the early 1950s, and one in seven women born in the early 1960s (2006 Census of Population and Dwellings). The drivers for these trends are complex, but include increasing participation of women in tertiary education and the workforce.
International fertility levels vary widely. Among the 30 OECD countries, total fertility rates (the average number of births per woman based on the age-specific fertility rates prevailing in a given year) ranged from 1.1 (Korea) to 2.2 (Mexico and Turkey) in 2005 (OECD). New Zealand had the fifth highest total fertility rate in 2005 (at 2.0), with the United States and Iceland at 2.1. The OECD average was 1.6 in 2005. On that basis alone, there is ample scope for New Zealand’s total fertility rate to decrease well below the 2008 level of 2.2 births per woman.
A discussion of current and future New Zealand fertility levels would be remiss without mentioning ethnic differentials. New Zealand’s above average fertility levels, by OECD standards, partly reflects the higher fertility of Māori and Pacific women whose total fertility rates were 2.8 and 3.0, respectively, in 2005–07. By comparison, women without a Māori or Pacific ethnicity had a total fertility rate of 1.8. Currently about two in five births have a Māori or Pacific parent.
Did Statistics NZ's population projections anticipate the rise in births?
The 2006-base national population projections were derived in 2007 when birth numbers and birth rates had only just begun their most recent upswing. In 2007, mid-range projection series 5 was considered to provide the best indication of future population trends – over both the short term and long term.
A cursory comparison suggests the 64,100 births registered in the year ended June 2008 was well above the 62,100 projected under series 5 of the 2006-base national population projections (see figure 2). However, Statistics NZ also projected 64,000 births under series 9, and 60,200 under series 1. The alternative projections allowed for the fact that fertility rates after 2007 were uncertain: the short-term increase in births could be brief or continue for several years; in the long term, women could average 1.7–2.1 births each over their lifetime. Recorded birth numbers in 2007–09 are therefore following the high trend, as indicated by projection series with a high fertility assumption (series 9).
Statistics NZ always produces and publishes more than one projection. The future is inherently uncertain, so no one should expect a single projection to predict the future. Indeed, it would be irresponsible of Statistics NZ to produce just one projection, thereby implying a certain future.
Furthermore, these alternative projections are not extreme scenarios. Rather, they are plausible scenarios, and collectively they provide an indication of future population trends as well as the general uncertainty around those trends. Estimating and presenting such uncertainty poses challenges. In reality, there is a continuum of possible population and birth outcomes. Although a discussion of stochastic or probabilistic projections is beyond the scope of this article, Statistics NZ has been developing a stochastic approach to its national population projections and welcomes discussion on this issue (email: email@example.com).
It is also important to note that Statistics NZ applies assumptions about future age-specific fertility rates (such as those in figure 1) to projections of the population to project birth numbers. This internationally adopted ‘cohort component’ method accounts for changes in the number of women of childbearing age, which, on its own, could cause changes in birth numbers even if fertility rates remained stable.
Deaths and migration
In contrast to births, New Zealand death numbers have varied little from year to year, as life expectancy has increased steadily. Barring major epidemics, wars or catastrophes, mortality and deaths can be projected with some confidence in the short term. In the long term, however, there is little consensus among demographers, epidemiologists, or others on future life expectancy. This reflects that mortality is the result of a complex interplay of biological factors, lifestyle factors (eg diet, exercise), environmental factors and technological factors (eg health, medical science) at an individual and societal level. Trends in male-female and ethnic mortality differentials are also uncertain.
In terms of New Zealand’s external migration, two features relevant to projections are the very high levels of gross migration (arrivals and departures), and the volatility of net migration (arrivals minus departures). In 2008, there were over 4.5 million arrivals into New Zealand, and over 4.5 million departures from New Zealand. For the purposes of population estimation and projection, however, typically only 2 percent of these travellers are moving for the purposes of living in a different country. These ‘permanent and long-term’ migrants provide an indication of the contribution of migration to changes in our population – but only an indication, as the information is largely based on passenger’s stated intentions on arrival and departure cards, and actual behaviour does not always match stated intentions.
Net migration can fluctuate significantly over short periods of time. Since 1951, New Zealand has experienced net permanent and long-term migration ranging between a net outflow of 43,300 in the year ended June 1979 and a net inflow of 42,500 in the year ended June 2003.
Recorded deaths and net migration in 2007–09 are similar to the mid-range projection (series 5), and well within the range provided by alternative projections (eg series 1 and 9).
How does one make sense of multiple projection scenarios?
It is tempting for analysts and policymakers to rely on one number. Rarely, however, can one number measure the complexities of the real world, let alone the real world of tomorrow. The challenge for the user is to interpret the alternative projections, while the challenge for producers of population projections is to facilitate that interpretation.
Some of the more important demographic trends will be realised under virtually all projection scenarios: an ageing population, and related to this, lower population and labour force growth in future; growing ethnic diversity; and smaller average household size. Other trends are more uncertain.
Given this general uncertainty, Statistics NZ releases updated projections every two to three years to incorporate the latest demographic trends. For example, the recent increase in births will affect the number of school entrants within five years. New classrooms or even new schools will be needed if fertility rates remain at their current levels. Whether the short-term increase in fertility rates affects average family size in the long term, however, is still uncertain.
For users of projections, a clear implication of this uncertainty is to not rely on a single projection as a forecast. Alternative projections are produced for conveying uncertainty and alternative population (and birth) outcomes. Users ignore alternative projections at their peril. But how alternative projections are used depends on the specifics of use, such as timeframe (how far ahead one needs to plan), and the relative costs of an under-projection and over-projection. For example, for infrastructure developments it may be better to plan for a higher growth scenario than a mid-range scenario, if the costs of subsequent expansion exceed the costs of over-capacity.
Recorded birth, death and migration numbers for 2007–09 are in the broad range indicated by official population projections released in 2007. However, birth numbers have certainly been at the high end of the projected range. The projections indicated an increase in births beyond 2007, although the extent of the increase varied between projection series.
The future is inherently uncertain. For population projections, the uncertainty increases the further one projects into the future. Uncertainty is also greater for projections of New Zealand subgroups, such as ethnic and local populations. For the latter, this is partly because of the greater volatility, both actual and potential, of migration patterns.
Given the uncertainty, is there any value in trying to anticipate future changes? Planning for the future remains important. In the case of population, change is fundamentally driven by three factors: fertility (births), mortality (deaths), and migration. In addition, about three-quarters of the projected population in 2031 and half of the projected population in 2061 are already alive. In the case of the older population, all those who will be aged 65 and over in 2061 are already alive. The uncertainty of future fertility patterns (after 2006) has no impact on the numbers aged 65 and over until after 2071.
Population projections might therefore be considered with greater confidence than other types of projections (eg economic). This makes population projections a useful information source for planners and policymakers in a diversity of areas including health, education, superannuation, and labour force. Nevertheless, users of projections are reminded that the future is not only uncertain, but non-demographic factors can have important impacts on the demand and use of services.
Births and deaths
Demographic Trends publication
Further information: email firstname.lastname@example.org