Although there are several determinants of demographic change in a country – fertility, mortality, immigration and emigration, working in tandem with the age and sex distribution, to shape the future size and composition of the population – in most national policy debates concerning population ageing, only fertility and immigration are discussed. However, both of these “remedies” have drawbacks that warrant consideration.

The decline in fertility to below two children per woman has been met with concern in many advanced economies (Sobotka et al., 2019). In 2015, 66% of European governments and almost 40% of Asian governments had policies in place to raise fertility or impede further decline (United Nations, 2018). However, substantial long-term fertility increases are difficult to achieve. Policies aimed at increasing fertility tend to have a larger effect on the timing of births than on the total number of children born (Bergsvik et al., 2021). Although studies have found some correlation between extensive public support to families and higher fertility (e.g., Wood et al., 2016), the answer to the question of whether the higher fertility is caused by the costly policies or by favourable economic conditions that made such policies possible is not obvious. And although (quasi-)experimental studies have suggested that certain policies may affect fertility, the future effects of these measures may be limited if their coverage is already wide (such as day-care in many Western countries), or these measures may have unwanted side effects, such as reducing female employment (Bergsvik et al., 2021). Moreover, changed fertility is linked to questions about climate and global sustainability. Higher fertility and, hence, a larger world population will make it more challenging to meet global food needs and to reduce global warming and biodiversity loss (Bongaarts, 1992; Bongaarts and O'Neill, 2018; Casey and Galor, 2017; Crist et al., 2017; O'Neill et al., 2010; Reher, 2007; Tamburino et al., 2020; Wilmoth et al., 2022; Wynes and Nicholas, 2017).

Increased immigration can reduce population ageing in the short term, since immigrants often arrive in their twenties or thirties. However, measures to increase immigration may be politically controversial. Furthermore, increased immigration has limited effects on ageing in the long term, because immigrants who remain in the destination countries also get older. Studies on replacement migration have generally concluded that to prevent low-fertility countries from experiencing demographic ageing, the volumes of immigration – or positive net migration – that would be required would have to be far higher than in the past, and would likely be politically unfeasible (Bijak et al., 2008; Blanchet, 1989; Heleniak and Sanchez Gassen, 2016; Paterno, 2011; United Nations, 2001). In the long run, immigration has tended to affect population size far more than its age structure (Alho, 2008; Bujard, 2015; Murphy, 2016); and in the even longer run, stable population theory posits that populations with sustained below-replacement fertility and constant immigration eventually become stationary, with age structures that depend on the distribution of the immigrants’ arrival ages (Arthur and Espenshade, 1988; Espenshade et al., 1982; Schmertmann, 1992).

For completeness, we should also mention the effect of mortality on population ageing. If the long-term trend of mortality decline halts, or if mortality increases, especially in the older age groups in which deaths most commonly occur, population ageing would be counteracted. As shown by Lutz and Scherbov (2003), future old-age dependency ratios are sensitive to different future mortality trends. However, measures to actively prevent further increases in remaining life expectancies are not on the political agenda, and are unlikely to be proposed.

The literature on how changed emigration can affect future ageing in a Western country like Norway is scarce. There are several potential explanations for why this is the case. First, many of today's ageing nations are characterised by more immigration than emigration, and while immigration has received considerable scholarly attention, there is much less research on emigration from Western countries. Second, as data on emigration are often inadequate, and many countries and agencies instead rely on net migration figures when, for instance, projecting future migration (Cappelen et al., 2015), it is difficult to make projection scenarios in which only the rates of emigration change.¹ Third, given that leaving a country is now considered a fundamental human right, reduced emigration may be considered equivalently hard to achieve (and as controversial) as the other demographic components mentioned above. Nevertheless, countries have policies in place that may affect the incentives to emigrate or to stay, as we discuss towards the end of this paper.

The scarcity of research on emigration from wealthy countries, including on its effect on ageing, is mirrored in a lack of interest from policymakers in Western countries.² It is, however, documented that emigration (or negative net migration) can affect ageing in some typical net emigration countries, such as in Central American countries in close proximity to the United States (García-Guerrero et al., 2019) and in Eastern European countries (Botev, 2012; Fihel et al., 2018; Philipov and Schuster, 2010; Potancoková et al., 2021; Rees et al., 2012). Although emigration usually makes a population older, whether this is actually the case depends on the ages of those who leave and of those who remain (Gavrilov and Heuveline, 2003; Parr, 2021), and on the fertility and mortality of the emigrants and of those who remain (García-Guerrero et al., 2019).

Over the last decades, several studies have presented comprehensive scenarios for Europe's population development and ageing with different sets of assumptions regarding – among other factors – emigration and immigration, or net migration (Bijak et al., 2008; Marois et al., 2020; Potancoková et al., 2021; Rees et al., 2012). These studies often applied different scenarios for intra-EU (or intra-European) migration and migration between EU/Europe and the rest of the world, and they based their emigration scenarios on observed emigration rates (which may be adjusted up or down in different scenarios). In the scenarios, changed intra-EU emigration from one country would affect intra-EU immigration to another EU country, making it hard to estimate how a change in emigration only would affect a country's ageing rate.

This paper assesses whether reduced emigration could mitigate a country's population ageing challenges, using Norway as a case study and applying the official model for population projections employed by Statistics Norway. This model uses high-quality register data covering the entire population, and it allows us to estimate the effects of changed emigration, rather than of changed net migration only. Moreover, it allows us to explore how much each of the other components of demographic change (fertility, mortality and immigration) need to change in order to yield the same effects on ageing as reduced emigration. To our knowledge, this has not been done in the literature before. We show how reduced emigration affects a multitude of population measures, including the share of immigrants in the population and other indicators of ageing, such as the prospective old-age dependency ratio (POADR). Furthermore, we add to the literature by estimating how reduced emigration for certain subgroups only would affect the different measures of ageing and the immigrants’ share, while also taking into account that the reduced emigration of natives would result in lower return migration back to Norway.

The ageing trend in Norway is similar to that of many other Western countries. Of the Norwegian population of about 5.4 million in 2022, elderly aged 65 or older comprised 18%. As shown in Figure 1, the Norwegian population is expected to continue to grow in the future, but at a slower pace than in the previous decades. The share of elderly (aged 65+) is projected to increase to 26% by 2050, and the OADR is projected to increase from 0.31 today to 0.49 in 2050, according to the main alternative in Statistics Norway's 2020 population projections (Syse et al., 2020). This is shown in the lower left panel of Figure 1. The lower right panel of the figure shows that both immigration and emigration have increased considerably over the last decades, and that they are expected to remain at relatively high levels in the future, with more immigration than emigration.

Like in many other countries, policymakers in Norway worry about the consequences of population ageing. In 2019, the prime minister expressed concerns about declining fertility (the Norwegian TFR decreased from almost two in 2009 to around 1.5 in 2019), and encouraged Norwegian couples to have more babies (Solberg, 2019).

Changes in emigration have not been part of the Norwegian discourse about remedies for ageing. However, as Figure 1 shows, about 30,000 individuals emigrate from Norway every year. This corresponds to about five emigrations per 1000 inhabitants. That rate is higher than the crude emigration rates of some other European countries, like Italy, Portugal and France, but it is clearly lower than the rates of other countries, such as Iceland, Switzerland, Ireland and Lithuania (see Figure 2).

Around 70% of emigrants from Norway are persons who have previously immigrated to Norway, while 30% are non-immigrants (natives). Most of the immigrants who emigrate have relatively short durations of stay in Norway, and immigrants from other Western countries have higher emigration rates than refugees and family migrants from less wealthy parts of the world. While most non-native emigrants return to their country of origin, especially if they are Nordic, the share of those who move on to a third country is also relatively high (almost 50%) among immigrants born in a less wealthy country such as Pakistan, Vietnam or Somalia (Pettersen, 2013; Skjerpen et al., 2015).

Most emigrants from Norway are 20–40 years old or below age 10 when they leave Norway. As summarised in Figure 3, the mean age of people who emigrate from Norway is almost 30 years, which is about five years higher than the mean age of those who immigrate. This age difference also means that women who emigrate have on average fewer remaining years for childbearing than women who immigrate. Appendix Figure A.1 shows emigration by age, in absolute numbers and per 1000, for immigrants and natives (non-immigrants). The figure indicates that the emigration rates are clearly higher for immigrants than for natives, at all ages.

The Norwegian population projections are based on detailed data from the Norwegian population register, which has data on all immigrations and emigrations at the individual level, as well as on other demographic events. The projections, and the data and methods used, are further described by Syse et al. (2020). In short, the model applies deterministic cohort-component methods, and emigration is projected using registered emigration rates (the last 10 years) for different subgroups by age, sex, immigrant background and (for immigrants) area of origin (see Footnote 3) and duration of stay in Norway.³ Future emigration from Norway is calculated from these fixed rates being applied to the projected future population of Norway. Emigration is projected to decrease slightly (cf. Figure 1 and Appendix Table A.1), mainly because of a declining share of immigrants with a short duration of stay. The other components of population change (fertility, mortality/life expectancy and immigration) are projected using a mix between expert judgements and separate models (Lee-Carter/ARIMA for mortality and a separate econometric model for immigration, Cappelen et al., 2015). Future immigration is calculated separately for three different origin areas (Western countries, Eastern EU and the rest of the world; see specification in Footnote 3), as well as for natives returning to Norway. While the total fertility rate (TFR) is projected to increase from 1.5 to a long-term level of about 1.7 children per woman, and life expectancy is projected to increase from around 81 years for men and 85 years for women to 89 and 91 years, respectively, in 2060, future immigration to Norway is projected to decline somewhat, from around 52,000 to a long-term level of around 37,000 annually.

As most emigrants are of childbearing age (cf. Appendix Figure A.1), lower emigration will also affect the number of births, since people who do not emigrate experience age- and sex-specific probabilities of both giving birth and dying. Whereas the age- and sex-specific probabilities of dying are the same for all subgroups in the model, the fertility rates differ between immigrants and the rest of the population, and (for immigrants) by area of origin and duration of stay in Norway.⁴

In this paper, we use the official medium projections (“main alternative”) as a baseline. In our first investigation of how reduced emigration affects ageing, we run the projection model with a hypothetical scenario in which all emigration rates are reduced by 50%, while the baseline assumptions are used for all the other components (i.e., fertility, mortality and immigration). Second, we run the model while changing each of the other components of demographic change (one at a time), with the aim of obtaining the same OADR in 2060 as the 50% reduction in emigration rates. We also explore the (even more) hypothetical scenario in which all emigration from Norway is stopped. Furthermore, we investigate the effect of a reduction in emigration rates for certain subgroups only (immigrants vs. natives, younger vs. older age groups).

The changes in the demographic components employed in these scenarios are large, and are, admittedly, not the most plausible ones. However, the relationships between them may provide useful information about whether a change in emigration could be an anti-ageing formula. To apply somewhat more realistic scenarios, we also make projections in which emigration is reduced for both immigrants and their Norwegian-born children, and one in which 50% lower emigration among natives leads to 50% lower return migration of natives, due to the smaller “pool” of native Norwegians living abroad. In all the other scenarios, the assumptions about future immigration to Norway are the same as in the official population projection (our baseline).

To summarise, we run the population projection models until 2100 with the following scenarios:

In each scenario, we explore the following measures of population change (# indicates number of persons):⁵

Table 1 shows historical figures of the above measures of ageing and the immigrants’ share for 2020 (the baseline year for our projections) and 2022 (the most recent year).

In addition to the measures listed above, we assess how the future numbers of births, deaths, emigrations and immigrations differ between the scenarios (statistics for 2020 and 2022 are displayed in Table 1, whereas the results from the respective scenarios are shown in Appendix Table A.1).

The results from our first analyses, in which the baseline (the main alternative from the official population projections) is compared with a scenario with 50% lower emigration rates, show that this reduction decreases the old-age dependency ratio (OADR) in 2060 from 0.54 to 0.52 (Table 2). This is still clearly higher than today's 0.31. Hence, with 50% lower emigration, the increase in the OADR from 2020 to 2060 would be 10.4% lower than in the baseline alternative.⁶ Other ageing measures (median age, share aged 65+ years and aged 20–64 years and the TDR, see Appendix Table A.2) show a similar pattern, with 50% lower emigration leading to only a slim reduction in ageing. Moreover, the prospective old-age dependency ratio (POADR), which is designed to account for changes in longevity so that the threshold between “working age” and “old” changes as people live longer (Sanderson and Scherbov, 2010), is only slightly affected: i.e., it goes down from 0.22 to 0.21, compared with 0.16 in 2020. In 2100 (lower part of Table 2), when many of the “emigrants” who never left will have grown old, the OADR in the scenario with 50% lower emigration (0.62) is even closer to the baseline OADR (0.63). Even a total stop of all emigration from Norway (right column in Table 2) only reduces the OADR from 0.54 to 0.49 in 2060 and from 0.63 to 0.60 in 2100.

Table 2 further shows that reducing emigration rates by 50% would have the same effect on ageing in 2060 (OADR of 0.52) as increasing the annual immigration by 25%, increasing all mortality rates by 20%, or increasing the total fertility rate (TFR) by 15% – which corresponds to one-quarter child per woman for natives (from 1.70 to 1.95).

While the effects of increased immigration or increased mortality are relatively similar to those of decreased emigration on most of the different ageing measures, increased fertility stands out as having a more rejuvenating long-run effect. This can also be seen in Appendix Figure A.2 and in the age profiles in Appendix Figure A.5. The effect of increased fertility on the OADR occurs later (when the additional children start reaching their twenties) and lasts longer than in the other scenarios. Hence, whereas reduced emigration may postpone the rise in the OADR in the first decades, increased fertility reduces the ageing challenges more in the long run.

Moreover, reduced emigration has a stronger effect on population growth than the other scenarios. In the −50% emigration scenario, population growth from 2020 to 2060 is 80% higher than in the baseline scenario, which is more than in the scenarios with higher fertility or immigration (and is clearly more than in the scenario with increased mortality).

Different scenarios give different population compositions by immigrant background. The share of immigrants in the population is highest in the scenario with reduced emigration, and the composition of immigrants also changes; the share originating from Western countries (Country Group 1) increases the most (in relative terms), whereas the share originating from the rest of the world (Country Group 3) is higher in the scenario with higher immigration than in the scenario with 50% lower emigration. As Western immigrants have the highest emigration rates, reduced emigration rates would keep a larger share of the relatively mobile Western immigrants from leaving Norway.

The results from the second part of our analyses – in which we investigate the effects of reduced emigration for certain subgroups only – show that the strongest effects on both ageing and population size can be observed when we limit the reduction in emigration rates to those below age 50 (see Table 3, Figure 4 and Appendix Figures A.3–A.5). The effects of the other subgroup analyses are less pronounced for both the ageing measures and the population counts. The effect of the reduced emigration of immigrants (particularly if we include their Norwegian-born children) on population size is larger than that of the reduced emigration of natives, whereas this is not the case for population ageing.

However, if the emigration of natives is lower, the number of natives abroad who can potentially return “home” will shrink. In the scenario in which 50% lower emigration among natives is coupled with 50% lower return migration among natives, the effects on population ageing as well as on population size and composition are minuscule.⁷

The different scenarios presented above also result in different projected numbers of births, deaths, immigrations and emigrations. These are shown in Appendix Table A.1.

Figure 4 illustrates how the different scenarios vary across the two main dimensions in our analyses, population size and ageing (the OADR being our primary measure – for results on the POADR, cf. Appendix Figure A.4). The two upper panels show the projected results for 2060, while the lower panels display the corresponding results for 2100. The left panels compare a 50% reduction in emigration with the corresponding changes in the fertility, immigration and mortality assumptions, while the right panels present a comparison of the reduction in emigration rates across various subgroups. Thus, the figure summarises the main results from our analyses. First, although reducing emigration by 50% can mitigate ageing somewhat (reducing OADRs by around 10%), it will by no means stop it. Second, the anti-ageing effect (in 2060) of 50% less emigration corresponds to that of 25% more immigration or one-quarter child more per woman. Moreover, reduced emigration has a stronger effect on population size than the corresponding changes in fertility, immigration and mortality. Finally, whereas reduced emigration among young people has the strongest anti-ageing effect, reduced emigration among natives has a negligeable impact if lower return migration is also taken into account.

Our conclusions are built on several assumptions that may be questioned. Furthermore, our analyses have other limitations as well. First, our measures of population ageing, dependency and support are relatively simple, relying to a large extent on chronologic age, and not on, for instance, information about whether people are in the labour force, or about their productivity, as this information is not included in the Norwegian population projections. However, ageing and dependencies are not solely driven by demographic factors. When considering the future challenges of ageing societies, the actual “dependency” and “support” of older and younger age groups will be of crucial importance. Other studies (e.g., (Bijak et al., 2008; Marois et al., 2020; Potancoková et al., 2021; Rees et al., 2012) have made projections that also take labour force participation and education into account, and most have concluded that the challenges associated with population ageing are less overwhelming when we consider the likely future development of these factors. For instance, Marois et al. (2020) showed that when labour force participation and education are included as additional variables, the projected increase in the dependency ratios is markedly lower. The authors thus concluded that the fears associated with the coming economic burdens of ageing have been unduly exaggerated by the use of conventional age dependency ratios. Other studies have, however, emphasised that the burdens will be large in terms of health and long-term care (LTC) costs (see, e.g., Marino et al., 2017 for LTC costs and Lorenzoni et al., 2019 for health costs), and in terms of costs relating to old-age pensions, driven primarily by the demographic changes associated with ageing (OECD, 2021). While Lorenzoni et al. (2019) suggested that the demographic effects account for just over a quarter of the projected growth in health expenditures, Marino et al. (2017) pointed out that there may be a stronger relationship between LTC spending and demographic change, given that a high share of LTC patients are elderly.

The OADR – our main ageing measure – appears to reflect the current situation in terms of the number of pensioners relative to the (potential) labour supply fairly well: across the OECD countries, the average effective age of labour market exit was 63.8 years for men and 62.4 years for women, albeit with considerable variation (range: 58.1–68.2) (ibid). Moreover, although the remaining life expectancy at the average age of labour market exit increased sharply between the 1970s and 2000s, it has stabilised over the past two decades as the life expectancy gains in old age have been offset by increases in the age of labour market exit. However, these trends might change in the future, and if we compare our results for the OADR vs. the POADR for the different scenarios, we see that although they largely mirror each other in relative terms (i.e., in both measures, reduced ageing is observed most markedly for the “no emigration” scenario in the first half of the period, before the 15% increase in fertility scenario takes precedence), there are some differences worth noting. For one, the 20% increased mortality scenario limits ageing considerably more when measured by the POADR rather than by the OADR. On the one hand, a benefit of the OADR is that it estimates fairly accurately the future labour supply and number of pensioners, whereas it might reflect health and LTC costs less well. The POADR, on the other hand, is likely to overestimate the future labour supply and to underestimate the future number of pensioners, but it may better reflect expenditures and resource needs relating to health and LTC. The projected life expectancy in Norway is quite high, and by the end of the century, the age cut-off for the numerator in the POADR is 80 years. Even in scenarios with healthy ageing and prolonged working lives, such a high age is unlikely to be relevant for pension and labour supply estimations.

Although our results show that lower emigration cannot stop ageing in Norway, measures aimed at reducing emigration could help to address the challenges of population ageing. Hence, a relevant question is whether it is actually possible for policymakers to reduce emigration in a liberal and democratic society like Norway.

Historically, European states have long traditions of regulating the exit of their people. Until the post-World War II period, emigration was not perceived (socially and legally) as a right, and the states were in full control over the international mobility of their subjects (Weinar, 2019). In the post-World War II period, the approach to individual rights changed, and liberal Western democracies made emigration a fundamental human right. Hence, prohibiting emigration is now considered highly problematic. However, that does not mean that countries do not have policies in place that either encourage or discourage emigration. According to a United Nations survey, 20% of governments (26% in Europe and North America) reported having policy measures that sought to lower the emigration of citizens (United Nations, 2020b). Some of the policies are targeted directly at the retention of particular groups, such as the highly skilled (McLeod et al., 2010; Toma and Villares-Varela, 2019; Wickramaarachchi and Butt, 2014).

Moreover, a wide range of policies in other fields could also impact emigration trends. Most notably, immigration policies regulating conditions for immigrants’ residence permits are closely linked to emigration, since individuals who lack permission to stay are expected (or forced) to leave the country. Furthermore, any measures that make staying in a country more attractive will also reduce the incentives to emigrate. For immigrants, whether they experience a welcoming or an unfriendly culture can incentivise them to stay or to leave. Policies aimed at improving the labour market integration of young people and of immigrants – including measures that facilitate the recognition of immigrants’ skills, measures that help immigrants avoid mismatch between their skills and the jobs they get and improve their educational attainment, and general economic policies aimed at increasing employment (Marois et al., 2020; Potancoková et al., 2021) – can increase labour participation while also reducing the incentives to emigrate.

Hence, policymakers who want to reduce emigration do have some tools available. Some of these tools are linked to international agreements on mobility and to policies regarding residence permits and the integration of immigrants, and to general economic policies.

As has been shown in this paper, no single demographic remedy is likely to stop ageing in a country like Norway. However, reduced emigration, higher immigration and higher fertility can contribute to a somewhat slower process of population ageing, and policymakers could consider measures encouraging all of these trends in order to meet future population challenges – in addition to important changes in labour market participation, productivity, etc.

Furthermore, policymakers should assess whether the greater challenge is population ageing or population decline. As we have shown, changes in emigration have a more pronounced impact on population size than on ageing in the Norwegian context.

Although some tools are available for policymakers who want to reduce emigration, achieving a reduction of 50% – the main scenario we explored – may prove just as hard or as controversial for policymakers as increasing fertility by one-quarter child per woman or increasing immigration by 25%. As in all policy-related questions, when comparing the different demographic remedies, the expected benefits of reducing ageing (and increasing population growth) should be weighed against the costs of the proposed policies. Reduced emigration may have several advantages compared to other demographic remedies; for example, it has little effect on environmental sustainability since does not add extra people to the earth as increased fertility does, and it may be less politically controversial than increased immigration. Some of the political measures that could reduce emigration may also have other beneficial effects, such as the better integration of immigrants, reduced labour market mismatches and higher employment rates, which could, in turn, place the society in a better position to handle population ageing.

Cost-benefit considerations should also take into account the costs and the benefits for other countries, and for the migrants themselves. The migration patterns that are most beneficial for one country may not be optimal for other countries. For instance, less emigration by young people would reduce ageing in Norway, but it would have the opposite effect in the emigrants’ destination countries. These cost-benefit considerations should also include costs and benefits at the individual level for the (potential) emigrant, as well as for her/his relatives and friends.

Hence, policy measures aimed at reducing emigration should be carefully chosen, and preference should be given to policies that have low costs for individuals and other countries, and that – independent of their effects on emigration – are likely to improve the society's ability to meet the challenges associated with population ageing.