Willem Adema
Jonas Fluchtmann
Willem Adema
Jonas Fluchtmann
Norway has experienced a sharp drop in births since 2009 and total fertility rates fell below the OECD average in 2017. With no immediately obvious individual drivers for the decline, the fall in Norwegian fertility remains a puzzle with incomplete answers. Drawing on the detail in the subsequent chapters, this first chapter outlines change in the economy, society, labour market and family policy environment, which may have affected family formation in Norway. It also assesses which policies might help Norway to sustain fertility rates and how demographic trends may pressurise economic and fiscal sustainability in Norway.
With a strong petrol-based economy and a protective welfare state, Norway has long provided a fertile ground for family formation. Economic security combined with a comprehensive family policy system for families with children throughout childhood means that Norwegians generally are able to combine their work and family responsibilities. Indeed, without pro-natalism being an avowed policy objective, the Total Fertility Rate (TFR) was well above the European and OECD averages throughout much of the 1990s and 2000s, while employment rates for men and women were high (Chapter 2).
However, while the family support system (Chapter 3) and many headline indicators of the Norwegian economy and society remained strong, Norway’s TFR declined by almost 25% over the past decade, falling from 1.98 children per woman in 2009 to 1.48 in 2020. This is the lowest rate since OECD records began in 1960 and well below the current OECD average in 2020 (1.59). After a brief increase to 1.55 in 2021, the Norwegian TFR rate has fallen once more to 1.41 in 2022. These developments have become a major topic of public and academic debate including around issues on downstream implications for population ageing and pressures on the Norwegian fiscal framework. The policy concerns about the falling birth rate were perhaps expressed most explicitly in her 2019 New Year’s speech when the then Prime Minister Erna Solberg urged Norwegians to have more children.
The decline in the fertility rate also caused policy experts, academics and politicians alike to seek explanations. For example, along with other ministries, the Ministry of Children and Families financed a comprehensive research project exploring underlying drivers of the fertility decline, that were found to be complex and multifaceted (Box 1.1). As such, the trend in Norwegian fertility rates remains a puzzle with incomplete answers.
Indeed, it is hard to pinpoint a single reason why the TFR in Norway declined between 2009 and 2020, as the decision to have a(nother) child is affected by a wide range of factors, such as economic and financial security, social norms as well as the overall labour market situation and family policy environment. At first glance, the fallout from the financial crisis after 2008‑09 may pose an obvious explanation for falling fertility rates. However, Norway experienced a light economic slowdown, and its economy quickly rebounded. Most Norwegians benefit from a secure labour market that allows to combine careers with family responsibilities (Chapter 4). This contributes to strong financial stability, which has remained robust in Norwegian households throughout the 2010s. Housing costs have increased in recent years, and attitudes and norms towards family formation appear to be changing, and perceptions of insecurity may also play a role (Chapter 5), but at this time, it is hard to identify objective factors that may have driven a fertility decline in the past decade.
Looking ahead, it is as yet unclear whether Norway will experience a modest rebound of the TFR in future, but fecundity problems at older ages may make a return to TFRs seen during the 2000s (around 1.9 children per woman) less likely. Whether fertility rates fall further or rise again is projected to have only a modest effect on fiscal pressures by 2060. The future of the Norwegian economy is much more susceptible to the anticipated future declines in oil revenues. The government Pension Fund of Norway – sometimes referred to as Norway’s Sovereign Wealth Fund, financed out of oil revenue is expected to continue to finance a major proportion of state expenditure (Chapter 6). However, this is unlikely to provide sufficient support for the transition into an older society.
Aiming for higher fertility rates to avert economic and social pressures – for example, through “baby-bonuses” and similar financial incentives, which often only have a temporary effect at best – would be very expensive. Supporting Norwegians to have the number of children they desire at the time of their choice as well as other responses to demographic change – such as prolonging working lives, increasing long-term investment in private pension savings and improving productivity – might be more promising
In light of the recent decline in Norwegian total fertility rates, the Ministry of Children and Families, the Ministry of Health and Care Services, the Ministry of Labour and Social Affairs, and the Ministry of Education and Research commissioned a comprehensive research agenda under the “Causes of the Fertility Decline” project, implemented in 2019.
The project consisted of five reports exploring the underlying causes of the fertility decline, co‑ordinated by the Centre for Fertility and Health at the Norwegian Institute of Public Health and involved researchers at various other Norwegian research institutes, including Statistics Norway, the Institute for Social Research and the University of Oslo. The findings of the project were synthesized in a final report, published in 2020, failed to identify any singular factor that could explain recent fertility trends. In fact, the project found no clear link to economic hardship following the great financial crisis unfolding in the late 2000s, nor was the decline in fertility especially pronounced among those with relatively poor health. While social value change, particularly declining religiosity, may have been a contributory driver, there is no hard-nosed evidence on this (Hart and Kravdal, 2020[1]).
The Norwegian Research Council also finances two ongoing projects focussing on fertility declines. The first one, “The Puzzle of the Recent Fertility Decline”, aims to further explore underlying drivers that may be able to explain falling birth rates in Norway and beyond. Lars Dommermuth leads the project in collaboration with various researchers at Norwegian and European universities and research institutes, with a foreseen end in 2023 (Statistics Norway, 2022[2]). The second one, “Falling Fertility and Rising Social Inequalities”, focusses on the role of perceived economic uncertainty how unequal fertility patterns emerge in the interplay between new gender roles and social groups. It is a collaboration involving Trude Lappegard, Axel Peter Kristensen, Lars Dommermuth, Tom Kornstad and Statistics Norway, with a foreseen end in 2024 (University of Oslo, 2022[3]).
Because of continued population ageing, Norway will also have to prepare for a substantial increase in long-term care (LTC) needs in the future. To adapt to rapidly rising LTC demand, better use needs to be made of the available health and care workforce as well as technology and digital solutions in the sector. Stylised projections also suggest that in future additional LTC workers will be needed to avert a looming LTC crisis.
Like most OECD countries, Norway has been experiencing record-low levels of their TFR over recent years. Despite relatively steady and high levels of fertility throughout the 1990s and early 2000s – a period over which many other countries saw their fertility rates drop substantially – Norway only experienced a strong decline in birth rates since 2009, when Norway had a TFR of 1.98 children per woman (Figure 1.1). The Norwegian TFR of 2020 (1.48 children per woman) is now well below the OECD average (1.59). Norway recorded a slight uptick in the TFR to 1.55 in 2021 following the onset of the COVID‑19 pandemic, but there was no effect on first births and the TFR subsequently dropped to 1.41 in 2022. It is unclear if longer-term concerns about political, economic and security concerns arising from aftermath of the COVID‑19 crisis and Russia’s war of aggression in Ukraine will affect family formation in Norway.
Total fertility rates are falling in large part because women in Norway – and across the OECD overall – have their children later today than a decade ago, and often have their first child in their thirties. In Norway, the average age at which mothers give birth to their children has increased from 30.0 to 31.4 between 2009 and 2020. Many more 25‑ and 30‑year‑olds are without children today than 10 years ago. Between 2010 and 2021, the proportion of 25‑year‑olds without any children increased by 13 percentage points to 85%, and for 30‑year‑olds by 11 percentage points to 52%. Even so, the two‑child norm remains strong in Norway, and the proportion of women in their late 30s and 40s who have two children remains stable (Chapter 2).
In line with a general delay in first births (as well an increase in women without children in general), the tempo‑adjusted TFR remains relatively high and has not declined as fast as the TFR. This is because the concept of the TFR may underestimate the actual fertility of women over the whole span of their reproductive years as it assumes that the timing of births is the same for the current and previous generations. When adjusting for tempo effects in birth rates by using a model that accounts for the timing and parity of birth (i.e. the tempo‑adjusted fertility rate), the fertility rate in Norway instead fell by 0.32 children per woman, from 2.21 in 2010 to 1.89 in 2019 (Chapter 2). These trends point towards a general postponement of births in Norway, which imply that some rebound in TFRs may occur in future.
Notes: The total fertility rate in a specific year is defined as the total number of children that would be born to each woman if she were to live to the end of her child-bearing years and give birth to children in alignment with the prevailing age‑specific fertility rates. It is calculated by totalling the age‑specific fertility rates as defined over five‑year intervals.
Source: OECD Family Database.
For most people, it is an important prerequisite to live together with a partner before having a child. In fact, 71% of 25‑44 year‑old respondents to the Generations and Gender Survey (GGS) who live together as formal cohabitants and/or as legal spouses are parents, while the corresponding figure for those who are single or not living with their partner is 26% (GGS, 2020[4]).
However, fewer Norwegians cohabit with a partner today than a decade ago – particularly among young people, a trend that mirrors the decline in fertility rates among 25‑ to 29‑year‑olds. Between 2017 and 2019, only 51% of Norwegians in this age group lived with a partner on average, down by 8 percentage points from the average level between 2008 and 2010 (Chapter 2). Indeed, despite high house and rental prices, Norway has the highest share of 20‑29 year‑olds that live on their own (44%) across the OECD (Figure 1.2), a proportion that even increased during the period of soaring house prices between 2010 and 2020.
Note: Instead of 2020, data refer to 2019 for Germany, Italy, and Lithuania; and to 2018 for Canada and Iceland. Data for 2010 are missing for Denmark.
Source: Special calculation from the OECD Affordable Housing Database Indicator HM1.4.
It has become more and more difficult for young people in Norway to buy a house in recent years with house prices increasing faster than incomes. In a country like Norway, where homeownership is a strong norm, buying a house may be an important prerequisite for family formation. Even if buying a house is possible, it might not satisfy the size and quality young couples require for raising a child. Those who are unable or unwilling to buy a home face high cost of rental accommodation, with higher and increasing housing cost overburden rates for tenants than for homeowners with mortgages (Chapter 2).
The high cost of housing has an unequal impact on households in Norway, with those at the bottom of the income distribution and/or the youngest hit the hardest. The rent price pressures particularly affect the group of Norwegians among which homeownership has been infeasible, including those in larger cities, those with lower incomes and those on more precarious work contracts. Indeed, Norwegians in the bottom quintile of the income distribution tend to spend more of their disposable income on housing costs than across the OECD on average (Chapter 2). At the same time, young Norwegians aged15-29 years have one of the highest housing cost overburden rates (20%) in Europe, only topped by Greece, Iceland and Denmark (Figure 1.3). The latter also has one the highest shares of young people living on their own (see Figure 1.2), which may explain high housing expenditures among this group.
Note: The data exclude households living in dwellings that are owned outright.
Source: OECD, Table B.1. Housing overburden rates, by age.
Norway provides a comprehensive family policy environment that strongly supports families over the early life‑course of their child(ren). The overarching goal of the Norwegian family policy framework is to offer safe economic and social conditions that ensure the healthy development of children and support the overall well-being of families in which both parents can reconcile work and family responsibilities (Chapter 3).
The aim of full labour market participation by mothers and fathers in Norway is facilitated through universal childcare and paid parental leave with reserved periods for care by both parents. By providing such comprehensive supports over the early years of a child’s life – policies which the population expects to be of a permanent and sufficiently generous nature – the Norwegian family policy framework supports family formation and well-being. Across OECD countries, increases in public expenditure on family benefits, including on parental leave and Early Childhood Education and Care (ECEC), are associated with an increase of the total fertility rate (Box 1.2 in Section 1.3).
Such a comprehensive family policy framework is expensive. Together with France and Hungary (both countries with reflections of pro-natalist notions in their family policies), as well as its neighbours Denmark and Sweden, Norway is among the OECD countries with the highest level of public expenditures on family benefits, about 3.3% of annual GDP in 2019. Most of this spending – about 60% of all public spending on family benefits – goes towards benefits in kind, such as Norway’s universal childcare system, while another substantial part – about 37% of all family support expenditure – goes towards cash benefits, such as parental leave allowances. Tax breaks for families do not take central stage in Norway’s family policy framework (Chapter 3).
As part of its family policy environment, Norway offers different family leaves, consisting of paid parental and home care leave that taken together can typically last up to 101 weeks (Figure 1.4). In contrast to many other countries, there is no separate maternity leave and only an unpaid paternity leave, though many fathers receive some form of wage replacement through individual or collective agreements for two weeks after birth. Instead of maternity leave and paid paternity leave, Norway integrates reserved periods for the exclusive use by fathers and mothers into their wider paid parental leave system, which replaces income at 100% of earnings up to a ceiling of NOK 55 740 (about USD 5 636) per month, about as much as the average gross monthly earnings in the country (often these payments are topped up to full replacement through collective agreements, including in the public sector).
Note: The figure refers to paid leave entitlements in place as of April 2022 and does not account for changes in leave entitlements thereafter Data reflect entitlements at the national or federal level only, and do not reflect regional variations or additional/alternative entitlements provided by states/provinces or local governments in some countries (e.g. Québec in Canada, or California in the United States). For more information on periods labelled “earmarked for mothers”, “earmarked for fathers or partners” and “fully shareable” see notes to Figure 3.3 in Chapter 3. The “average payment rate” refers the proportion of previous earnings replaced by the benefit over the length of the paid leave entitlement for a person earning 100% of average national full-time earnings. If this covers more than one period of leave at two different payment rates, then a weighted average is calculated based on the length of each period. In most countries, benefits are calculated based on gross earnings, with the “payment rates” shown reflecting the proportion of gross earnings replaced by the benefit. In Austria, Chile, Germany and Lithuania, benefits are calculated based on previous net (post income tax and social security contribution) earnings, while in France benefits are calculated based on post-social-security-contribution earnings. Payment rates for these countries reflect the proportion of the appropriate net earnings replaced by the benefit. Additionally, in some countries maternity and parental benefits may be subject to taxation and may count towards the income base for social security contributions. As a result, the amounts actual amounts received by the individual on leave may differ from those shown in the figure. In Japan, the periods of parental leave that earmarked for fathers and mothers must be used simultaneously if both parents are to use the entirety of their entitlement. These periods are represented by turquoise bars. For more information, see the OECD Family Database.
Source: OECD Family Database Indicator PF2.1 and Indicator PF2.5.
Paid parental leave consists of maternal and paternal quotas that reserve 18 weeks for exclusive use by the mother (3 weeks before birth and 15 weeks afterwards) and 15 weeks for the father or recognised second-parent. If parents choose instead to be paid at 80% of pervious earnings, both quotas are extended by 4 weeks. On top of this, parents are entitled to a fully sharable period of 16 weeks (or 18 weeks when choosing 80% payment) that can be taken by either parent, though fathers face stricter eligibility conditions for this shareable leave, requiring that the mother works or studies at a minimum of 75% of full-time hours, while mothers face no such requirement. Despite wide‑spread use of parental leave by fathers – about 38% of all leave days taken in 2022 – they rarely take leave for longer than their paternal quota. In fact, while 90% of fathers take parental leave, 70% do so for precisely the length of the paternal quota and not a day longer (Chapter 3).
In addition to parental leave, Norwegian parents have access to paid home care leave if their child is not attending publicly funded ECEC services. While parental leave can be taken up until the third birthday of the child (e.g. through part-time leave option after initial 6 weeks of full-time leave or the splitting of leave in multiple blocks), home care leave is only paid for a child between one and two years of age. This leave is typically used to cover the time between expiration of parental leave entitlements and the right to a kindergarten place. Home care leave is paid through a flat-rate cash-for-care benefit of NOK 7 500 (USD 758) and thus substantially lower than parental leave benefits.
Norway provides universal and affordable childcare in their kindergartens with a statutory right to a childcare place from age one, though the exact date of enrolment depends on the birthdate (see more below). As a result, Norwegian children typically attend ECEC services after parental leave expires, which provide parents with good opportunities to reconcile work and family responsibilities. In fact, ECEC enrolment rates are among the highest in the OECD for under 3‑year‑olds (Figure 1.5) and almost universal among 3 to 5‑year‑olds, all the while hours of ECEC attendance are particularly long (about 35 hours per week). High ECEC enrolment is a direct result of a large‑scale expansion of the ECEC system between 2003 sand 2009, which culminated with the individual statutory right to a kindergarten place for all children aged 1 to 5 (Chapter 3).
Many parents previously relied on the cash-for-care benefit to care for their children after parental leave expired, but the number of recipients has fallen by more than 90% since the expansion of the ECEC sector. Today, the cash-for-care benefit is mostly used to bridge gaps in family support based on the children’s birthday. For example, while children who turn one between January and July are entitled to a kindergarten place in August of the same year and those who turn one between August and November have a statutory right to a place in the month, they turn one, children who turn one in December have to wait until the following August. As a result, children born on 1 December are 20 months old when they become entitled to a kindergarten place, while those born on 30 November are entitled when only 12 months old. This leaves many parents in need to find alternative care solutions.
The attendance of ECEC services is generally affordable, with regulated flat fees across pre‑school ages and a range of discounts for families. For example, while the maximum attendance fee is set at NOK 3 050 (USD 308), Norwegian parents do not have to pay fees more than 6% of gross household income. Families earning less than NOK 583 650 (USD 59 011) are also entitled to 20 hours of free kindergarten attendance for their children aged between 2 and 5 years old. At the same time, parents of multiple children attending ECEC services in the same municipality receive a discount of a minimum of 30% for fees of the second child and a minimum 50% reduction in fees for any additional child. However, application procedures and documentation requirements were previously considered as cumbersome, which created a significant gap between eligibility and take‑up of ECEC discounts. A newly introduced system that facilitates automated case management is intended to reduce the administrative burden and increase take‑up, especially among vulnerable families (Chapter 3).
Note: Data generally include children enrolled in early childhood education services (ISCED 2011 level 0) and other registered ECEC services (ECEC services outside the scope of ISCED 0, because they are not in adherence with all ISCED‑2011 criteria). Data for, Iceland and the United Kingdom refer to 2018 and for Japan to 2019.
All Norwegian municipalities offer out-of-school hours services (OSH) once children enter school. While the take‑up among 6 to 8‑year‑old school children is comparatively high (50%), attendance drops sharply for children in higher grades of elementary school (11%). In part, this is related to the relatively parental fees payable for OSH services – which are often about has hight as for ECEC on a monthly basis, despite considerable fewer hours of participation. Indeed, about one‑third of parents who do not have their child in OSH-services state that price is a reason. Further, a previous lack of a unified national OSH curriculum, with wide ranging quality and content across OSH facilities, may have also contributed to relatively low attendance among older children. A new unified curriculum for OSD has recently been introduced (Chapter 3).
The Norwegian economy provides safe and secure labour market conditions for many families, also when children are young. However, lower-educated people and young adults face labour market insecurities despite tight labour markets. Indeed, economic security is often a prerequisite for family formation and fertility intentions in themselves. As such, changes in the labour market situation of young couples as well as their increased barriers to homeownership may be very important determinants of fertility rates over time. Across OECD countries, changes in women’s and men’s employment rates are important drivers of changes in fertility outcomes (Box 1.2).
The continuum of family supports in Norway, with high enrolment in ECEC at young ages, facilitates a “co-provider” model with high employment rates of men and women, even as they become parents. In fact, the employment rate barely differs between women aged 25‑54 without children under 15, of whom 80% were employed in 2020, and mothers whose youngest child is less than five years, of whom 78% were employed in the same year. Across most other OECD countries, the differences between mothers and non-mothers are far more pronounced, with those without children being employed much more often (Figure 1.6).
As in all OECD countries, Norwegian women still spend more time on unpaid work than men. However, with a comparatively modest gender gap in hours spent on unpaid work (59 minutes), Norway is ahead of the curve compared to other OECD countries when it comes to being “co-carers” at home, except for Sweden (49 minutes). Even though these gender gaps have declined noticeably over recent years, young Norwegian families may still be expecting more equality. Indeed, Norwegian mothers are somewhat less satisfied than fathers with the division of unpaid work in the household; they also struggle more than fathers in combining paid and unpaid work. Further helping families to achieve improved gender parity in the home may be important for fertility, as a more unequal distribution of unpaid work is linked to worse relationship satisfaction and a reduced likelihood of the first and subsequent births (Chapter 4).
Note: Data refer to 2019 for the United Kingdom.
Source: OECD calculations based on the EU Labour Force Survey. See more information in the OECD Family Database.
In the presence of a strong dual-earner norm – a result of a long-standing focus on enabling women to enter the labour market, but also high costs of living that necessitate dual incomes – Norway has one of the smallest gender employment gaps in the OECD. However, to accommodate family responsibilities with their career, some Norwegian women work part-time. While the share of Norwegian employed women aged 25‑54 working part-time (19%) is about as large as across the OECD on average (20%), it is greater than in other Nordic countries. While Norway’s labour market policy focuses on encouraging the norm of full-time work, the take‑up of part-time work limits the long-term earnings trajectories of women relative to men and is linked to both “motherhood wage penalties” and gender pay gaps. Instead of relying in part-time work in the early years of a child’s life, it may be feasible to increase flexibility in the timing of working hours.
Despite a smaller motherhood wage penalty than in many other OECD countries and an overall comparatively low and decreasing gender wage gap – just below 5% at median for full-time earners in 2020, down 3.7 percentage points from 2009 – uncertainties over the factors contributing to the gender pay gap could contribute to parents postponing the decision to have children. Survey evidence shows that, in general, a larger proportion of women than men who do not yet have children report expecting negative impacts on their career and work hours if they were to have a child. Mothers who expect such effects on their labour market trajectories might postpone the decision to have children, or have no children at all, especially if they are strongly career focused. A U-shape in the relationship between wage increases and the likelihood of having a child, which is more pronounced among younger cohorts, may mean that today women’s fertility decisions are more sensitive to wage changes than women were in the past (Chapter 4).
To address gender pay gaps, Norway is one of nine countries in the OECD that has a comprehensive equal pay auditing system. As such, every two years, all public and private employers – the latter only if they ordinarily employ more than 50 employees – are required to publish the pay differences between men and women. This system has only been in place since 2020, so it remains to be seen whether it is effective in further reducing gender pay gaps in Norway (Chapter 4).
Compared to other OECD countries, the Norwegian labour market is in a relatively good position. For example, unemployment among 25‑ to 54‑year‑olds was at one of the lowest levels across all OECD countries in 2020 (at 4.1% for men and 3.6% for women). At the same time, the quality of Norwegian jobs tends to be better than in other OECD countries, with very little reliance on temporary contracts overall. Although unemployment rates and the incidence of long-term unemployment rose a little over the past decade, with younger adults a little more exposed to temporary contracts than older workers are, the proportions affected are small relative to other OECD countries (Chapter 4).
Since having a stable job might be a necessary condition for starting a family, the time it takes for young adults to potentially go through trial periods, temporary contracts, and initial periods of part-time work may force many to put off having children. Indeed, evidence suggests that the likelihood of a woman or her partner getting a permanent contract increases the possibility of having their first child more quickly, while temporary contracts are notably harmful to women’s likelihood to have a child. Unions and collective bargaining can help ensure that jobs across the economy are stable and that labour rights keep up with shifting workplace demands and labour market. While people covered by unions may enjoy more job security than those who are not, Norway differs slightly from other Nordic nations with high union coverage in that many of the rights won by unions have become enshrined in national legislation.
A continuum of comprehensive family policies throughout childhood and the objective state of the economy are not the only factors that determine whether people will choose to have children. Young couples’ expectations of future financial security as well as their perceptions and norms towards family formation, the perceived negative aspects of parenthood as well as norms around the intensity of parenting, are likewise important. Fertility trends also depend on changes in attitudes. While Norway has a long tradition of egalitarian gender norms, attitudes toward risk and family life have changed. Such changes might be just as important – or even more important – than the actual underlying changes in living standards and economic uncertainty.
Although Norway’s economy is robust, this does not necessarily mean that Norwegians feel the same way about “economic life”. Despite the potential importance of perceived or anticipated insecurity, it is complicated to measure such links properly. Nevertheless, there is some evidence that the belief of being able to bear a potential job loss or other financial setback is a particularly strong predictor of fertility, more powerful than objective economic security (Chapter 5).
To explore what underlying drivers are associated with changes in fertility rates, this report presents a cross-national regression exercise that estimates the effect of within-country and over-time changes in different aspects of the family policy framework and the labour market on fertility outcomes in a specific-country. For example, the associations found through the regressions are between within-country changes in female employment rates and within-country changes in the TFR or the MAB (mean age of mothers at the birth of their child). Importantly, the regressions do not provide evidence on a causal relationship between policies/outcomes and fertility but can nonetheless provide useful insights on which policies should be prioritised to impact fertility choices.
In terms of family policy, increases in public expenditure on ECEC services are associated with higher TFRs and a lower average age of mothers at first childbirth. This finding maps well to findings of quasi‑experimental studies on investment in the ECEC sector, although effect-sizes may vary across birth-parity. Changes in the ECEC enrolment rate are not associated with fertility outcomes, but the effect of ECEC availability may to some degree already be captured in public expenditure on the sector itself. Also, investment in the ECEC sector does not always reverse negative fertility trends – as, for example, seen with the large‑scale ECEC expansion in Korea (Chapter 3).
Increases in public expenditure on parental leave payments also have a positive association with the TFR and a negative link to the MAB. Increases in the length of parental and maternity leaves available to mothers is associated with a higher TFR, while earmarked parental and paternity leave for fathers has no discernible effect on fertility outcomes. The literature generally also finds a positive effects of parental leave on fertility rates, while effects of other non-leave lump-sum cash benefits that are designed to incentivise fertility (e.g. “baby-bonuses”) are often limited and temporary and ineffective in sustainably raising TFRs.
The labour market position of families is an important determinant of fertility. In the OECD-wide regression, increases in women’s employment rate are associated with a higher TFR and a lower average age of mothers at the birth of their children. This is well in line with findings in the international literature, which highlights the importance of women’s employment for the financial security of households. Increasing employment rates has raised the opportunity costs of childbirth for women, but family polices have co-developed to ease the combination of family life with labour market careers. The employment rate of men has similar associations with the TFR, but the effects size is somewhat smaller. The literature also bears out negative effects of increasing hours in paid full-time employment by women on fertility, which the OECD-wide regressions do not find, despite a strong negative effect that remains non-significant. Similarly, the incidence of part-time employment among women does not appear to be related to fertility outcomes.
The timing of the fertility rates decline suggests a link with the financial crisis that materialised in 2008‑09. However, the ensuing low birth rates have been more persistent than the economic consequences, which were shallow and short-lived in Norway. The Norwegian economy rebounded quickly after 2008/9 and remained one of strongest across the OECD throughout the 2010s, spurred by high and increasing oil prices, leading to quick recovery in employment and real wages. There is evidence to suggest that men and women in Norway have changed the way they consider their own economic situation after the financial crisis 2008‑09. People seem to ascribe greater value to being in a stable job during the 2010s than before. It has also become more important to have more work experience before having a first child, and this is especially so for women aged 26‑32 compared to younger women (Chapter 5). This suggests that women value having a stronger foothold in the labour market more today than 10 years ago.
At the same time, the 2010s were characterised by the perception of unpredictability and macro-level volatility, which might have contributed to a sense of perceived insecurity that did not entirely represent Norway’s solid economic performance. Public narratives propagated through (social) media – which may have accelerated at times of rapid globalisation – play a significant role in shaping the perception of economic strength or weakness. Beyond the financial crisis, broad-based negative narratives during the 2010s also included xenophobic responses to the 2015 refugee crisis, Euroscepticism, and the development of populism, all of which contributed to the spread of a sense of instability and unpredictability. As such, Norway may have imported some of the economic insecurity, which was more present in other countries, and which may have played a role in the fertility trends of the 2010s (Chapter 5).
For Norwegians, there still is a strong norm to have children at some point in life, and “family” is at the top of many people’s lists of what brings well-being. Many Norwegians state that raising children is a rewarding experience that benefits them throughout their lives. Indeed, most working age Norwegians (88%) personally consider that the reason for having children is the lifelong happiness they bring. At the same time, it is becoming increasingly acceptable not to have children. For example, researchers in Finland, observe the rise of a new childfree ideal, and in the United States more and more young adults do not want to have any children. At the same time, life goals other than family and children, including career advancement and self-realisation, have gained importance in recent years (Chapter 5).
There is evidence that women feel more negatively about having kids than men do, and they often retain a kind of “veto” authority over the decision of having a child (see e.g. Stein, Willen and Pavetic (2014[5]) and Mynarska and Rytel (2022[6]). A reason for this may be stronger fears and anxiety about becoming parents than men may have, which could be related to the more direct effects on women of heavier care‑load and potential health issues related to pregnancy, birth, and childcare. Indeed, along with sleep deprivation and concerns about the potential negative effects on elder siblings, the physical challenges of pregnancy and childbirth are cited among the Norwegians as reasons among parents for choosing not to have another child (Chapter 5). At the same time the proportion of female and male (including both parents and non-parents) respondents rating reasons for not having a(nother) child as “fairly important” or “very important” tend to be similar across reasons, according to data from the Generations and Gender Programme (Figure 1.7). Even though the impact of these factors does not seem to have increased in recent years, health-related issues might also be more important if there is a greater hesitancy around having children overall. People who worry about the negative aspects of childbirth and childrearing, for example, are more likely to choose not to have children (Chapter 5).
Norms have changed towards more intensive parenting, with higher involvement of fathers and mothers in the upbringing of their child. This could mean that the matter of choosing a good timing to start a family, and thereby the relevance of navigating insecurities and competing life goals, has gained importance over the long term. While women historically have had to consider life changes due to the responsibility that is placed on them through children, this is a newer phenomenon for men, who are more often concerned that children are a burden to time and energy. It is perhaps unsurprising that young adults feel the need to take some time to navigate these new practices and norms before finding a balance between work, life, and partnerships that they are happy with. Indeed, the difficult responsibility of having and raising children is most frequently reported as an important reason for not having children (Figure 1.7).
Note: GGS data were obtained from the Generations and Gender Programme Data Archive and collected through the Survey on Family and Work by Statistics Norway, with contributions from Lars Dommermuth and Trude Lappegård. More information about model survey instruments can be found at: United Nations Economic Commission for Europe (2005). Generations & Gender Programme: Survey Instruments. New York and Geneva: United Nations. Respondents include parents and non-parents.
The potential link between gender equality and fertility is a topic of continuous discussion. Many people of childbearing age find it difficult to handle the responsibility of raising children, which places additional strain on the duties and responsibilities of prospective fathers and mothers. The decision to have children increasingly hinges on both parties feeling ready to take on this responsibility as both more and more often act as parents on an equal basis. According to research, actual and perceived fairness are crucial pieces of the puzzle when attempting to comprehend young couples’ decisions on fertility. However, growing gender equality could also be considered as a contributing factor to declining fertility, in part because of altered gender roles in relationships and new behavioural patterns (Chapter 5).
Compared with other countries, Norway still stands out as one of the OECD countries with the highest equality between men and women, reaching top levels across a range of gender equality measures in education, employment and governance. At the same time, egalitarian gender norms are common and growing over time, with strong progress over the years for Norwegian men (Chapter 5). For example, while across the OECD about 1‑in‑5 (21%) believe that men make better political leaders than women do, only 8% of Norwegian survey respondents think the same (Figure 1.8).
It is not always obvious whether such egalitarian gender norms encourage or discourage fertility. For instance, it has been suggested that one of the major reasons for postponing or avoiding having children is the extra time and emotional commitment required of fathers (Chapter 5). This is consistent with recent survey data from Norway showing that men are typically the ones who argue against having a(nother) child. At the same time, the explicit goal of Norwegian family and labour market policy is to encourage equal participation by men and women in the labour-force and the household (Chapters 3 and 4). Since women are not expected to quit their jobs and earn less money to care for children, this can lower the financial and personal expenses for women to have children. By extension, such policies can support a more equitable distribution of family and caring responsibilities among men and women. A delay in first births may then be an indication that young people are taking their time to find new ways of navigating life goals, careers, self-realisation, partnerships, and gender equality (Chapter 5).
Note: Data refer to 2020 for Canada, 2019 for Japan and New Zealand, 2018 for Australia, Chile, Mexico, Korea and Türkiye, as well as 2017 for all European countries and the United States.
Source: European Values Study and World Values Survey (EVS/WVS, 2021[7]).
Similar to most other OECD countries, Norway will face noticeable demographic pressures over the coming decades with a rapidly ageing population. This is both a result of an increase in life expectancy and birth rates that have fallen well below replacement levels. As such, the structure of Norway’s population is projected to undergo noticeable changes over the coming 40 years, with a particular increase in the share of elderly among the overall population (Figure 1.9). These trends might have strong implications on economic and social outcomes, such as rising fiscal pressures through an increasing share of the population that retires and leaves the labour market.
However, to what degree different possible scenarios on future fertility rates could avert those trends is not necessarily clear. Instead of aiming for fertility rates that rise back to previously seen levels, policy could focus on other factors – such as improving the health of the population and prolonging working lives, raising long-term investment in private pension savings, increasing the long-term care workforce and enhancing economic productivity – that are key to reduce the impact of demographic trends on economic and societal pressures (Chapter 6).
Long-term projections that link fertility scenarios with the outlook on economic and social outcomes can help to identify whether any policy aims of increasing future fertility rates are appropriate and/or necessary. The projections in this report, presenting outcomes with a horizon of 2060, are based on the OECD Long-Term Model as well as various fertility scenarios based on the population projections from Eurostat and the United Nations. For Norway, these baseline projections assume a convergence to a TFR of 1.62 by 2060. The presented fertility scenarios are chosen to converge either to 0.5 below (“low fertility”) or to 0.5 above (“high fertility”) this baseline, following similar scenarios used in the UN World Population Prospects. The obtained estimates result from a purely mechanical simulation and assume that any changes in fertility outcomes do not interact with factors outside of the model or have any indirect effects beyond its scope (Chapter 6).
Note: The OECD total is a total aggregated across all OECD member countries.
Source: For Norway Eurostat Population Projections 2019, for OECD Total UN World Population Prospects 2022.
Demographic projections involve noticeable changes in the population shares of young, working age and elderly population groups by 2060. Mounting demographic pressures are already expected in the baseline scenario, but a lower fertility rate would exacerbate these trends. While the absolute number of people in the working age population (aged 15‑64) will increase either way, it is outpaced by increases in the absolute size of the elderly population (aged 65 and above) in each scenario. This would thin out younger populations over time and likewise increase the projected old-age dependency ratio in 2060, ranging from 44% at baseline to 48% under low fertility and 40% under high fertility – a strong increase from the 27% recorded in 2020. Regardless, as the absolute size of the working-age‑population increases until 2060, the Norwegian labour force is projected to grow by 10% at baseline. Even low fertility scenario would still see a 1% growth in the size of the labour force (Figure 1.10). However, these positive projections on the size of the labour force are likely to subside beyond the 2060 projection horizon (Chapter 6).
With a growing labour force, the economy is expected to continue to grow in absolute terms, irrespective of future fertility rates. However, when economic output is expressed on a per capita basis, future projections on production highly depend on the size of the working-age population. As the entry into the workforce comes many years after birth, lower than projected birth rates would initially mean that the size of the labour force and economic output would initially remain stable. Indeed, under a low fertility scenario for Norway, economic output per capita would increase until the effect of fewer births reduces the size of the labour force relative to the baseline scenario, with negative effects on economic growth per capita towards 2060 (Chapter 6). Overall, however, lower than baseline fertility would result in annual potential GDP per capita growth of 0.96 percentage points on average between 2020 and 2060, slightly higher than the 0.89 percentage points at baseline. This would culminate in aggregate growth of potential GDP per capita of 43% under low fertility, relative to 38% at baseline and 34% under high fertility (Figure 1.10).
Note: The baseline scenario refers to baseline fertility rates in the Eurostat population projections, which converge to a TFR of 1.62 by 2060 for Norway. “Low fertility” refers to fertility rates that converge to a TFR 0.5 below baseline. “High fertility” refers to fertility rates that converge to a TFR 0.5 above baseline.
Source: Eurostat Historical Population Projections 2019 and the OECD Long-Term Growth Model
With an ageing society, public expenditure on retirement income as well as health and long-term care is expected to increase. Irrespective of the development of fertility rates, Norway thus faces substantial fiscal pressures in the future. In fact, Norway’s primary public expenditure as a share of GDP in 2060 is set to increase by 12‑13% relative to 2020 irrespective of whether fertility rates will follow baseline or alternative scenarios (Figure 1.10). While the proportion of expenditure on health and retirement among all public spending increases, low fertility would decrease other spending, mainly because public expenditure on ECEC, school and general family benefits is likely to decline (Chapter 6).
The question of whether all of these trends will lead to more fiscal strain on Norway in the future cannot be answered by this report, as it cannot provide answers on the development on the public revenue under different fertility scenarios. National projections of Statistics Norway and the Norwegian Ministry of Finance point to a slightly lower fiscal pressure under lower than baseline fertility rates. However, Norway will likely lose some of its former fiscal flexibility in the future as petrol revenues are expected to decline. At the same time, the sizable Norwegian Government Pension Fund Global, is expected to continue to finance a major proportion of state expenditure (Chapter 6).
As such, fiscal pressure may mount, generally independent of future fertility rates. Instead of aiming for higher fertility rates to avert economic and social pressures, prolonging working lives, increasing long-term investment in private pension savings and improving productivity might thus be more effective. Fewer births than in previous decades are therefore not necessarily a serious fiscal and economic concern, as long as Norway prepares for the future and keeps its population well informed about necessary adjustments in policies.
Independent of any fiscal pressures, ageing societies will require increased long-term care provision. For Norway, this would mean that the required Long-Term Care (LTC) workforce would have to be more than twice as large than what is projected for 2060 if the aim would be to keep provision at the current level. Even when factoring in productivity increases in the LTC sector, the required workforce would still be 69% larger than what is projected. However, the Norwegian Helsepersonellkommisjonen (Health Personnel Commission), for example, assumes that a better utilisation of the available health and care workforce and the increased use of technology and digital solutions would suffice in averting a looming LTC crisis (Ministry of Health and Welfare (2023[8])). Nevertheless, such a productivity boost in the health and care sector would have to be noticeably larger than the projected productivity increase for the aggregate Norwegian economy by 2060 (Chapter 6). Higher fertility rates would somewhat reduce these pressures as more individuals would flow into the labour force over time – and thus also enter LTC professions – but this would not be sufficient to avert a looming LTC crisis in the future, even with higher productivity. Along with a better utilisation of the workforce and the increased use of technology, Norway could therefore aim for an expansion of the LTC workforce by steering more into training and education related to geriatric care – including boys and men – or specific migration channels to recruit foreign LTC workers, to avert a looming LTC crisis in the future (Chapter 6).
Norway is doing well in terms of its family policy approach and labour market outcomes, and there is little evidence that the country could introduce or change specific policies that would bring total fertility rates back to two children per women or just above – the replacement TFR of 2.1. This is not necessarily problematic, as demographic change will substantially affect Norway largely independently of the exact trajectory of future fertility rates. Thus, the country is well advised to prepare for demographic change, rather than having a narrow focus on fertility trends.
Despite the overall good performance of Norway on a wide range of indicators, there are some outstanding issues that, while not necessarily affecting fertility much in themselves, may improve the continuum of care over the early life of a child as well as the parental labour market position, particularly of young and vulnerable Norwegians. Concrete recommendations identified in this report include:
Parental leave: Norway could consider aligning the eligibility requirements for shareable parental leave, so that mothers do not have to be in employment or study for fathers to take sharable leave. Shareable parental leave could also be phased out over time, while eventually granting each parent half of the entire leave at default with the option of transferring a certain number of weeks to their partner, similar to the approaches followed in Iceland and Finland.
More scope for flexibility in working hours when children are very young: At present, part-time work during parental leave requires agreement of the employer. Legislation could be redesigned to grant a unilateral right to parental leave on a part-time basis. If leave rights were also individualised this could further promote a more equal “co-caring” approach among fathers and mothers. Financial incentives in the form of “bonus months” could be used to increase take‑up. If parents both used the system with similar intensity, sharing of part-time work and caring could be one way to address the potential gap between parental leave expiry and enrolment in ECEC.
Early childhood education and care: another option to address the gap between the end of parental leave entitlements and effective enrolment in ECEC facilities would be to grant the right to a kindergarten place to all children at the end of the month in which they turn one. This could also facilitate a phase‑out of the cash-for-care benefits, which is frequently used to cover for the time between paid parental leave and kindergarten enrolment.
Gender Pay Gaps: The Norwegian Government could extend the nature of pay audits to contribute to a better understanding of drivers of the parenthood wage gap and continue to put pressure on employers to address parenthood wage gaps as well as wider gender wage gaps in their wage‑setting strategies. For example, it could expand the requirements to indicate parenthood penalties by reporting wages of parents at a given period after returning from parental leave.
Housing market: To improve the efficiency and fairness of taxes on housing assets Norway could cut back on tax related concessions provided to homeowners, introduce capital gains tax from home sales and scrap special wealth tax discount rates for housing. Norway could enhance the performance of rental markets by, for example, removing the income‑tax concessions for owner-occupiers renting out parts of their primary residences or second dwellings. Norway could also focus on increasing the stock of social housing, especially in cities, for instance by increasing loans for construction.
Supporting the disadvantaged in the labour market: The vast majority of Norwegian workers are provided with job security, fair contracts and reskilling opportunities, but this does not hold for all workers. Norway should carry on with supporting disadvantaged workers into good and secure jobs, though continued investment in upskilling and reskilling programmes for adults and broader active labour market supports.
[14] Adema, W., N. Ali and O. Thévenon (2014), “Changes in Family Policies and Outcomes: Is there Convergence?”, OECD Social, Employment and Migration Working Papers, No. 157, OECD Publishing, Paris, https://doi.org/10.1787/5jz13wllxgzt-en.
[11] Ahn, N. and P. Mira (2002), “A note on the changing relationship between fertility and female employment rates in developed countries”, Journal of Population Economics, Vol. 15/4, pp. 667-682, https://doi.org/10.1007/s001480100078.
[18] Beck, N. and J. Katz (1995), “What To Do (and Not to Do) with Time-Series Cross-Section Data”, American Political Science Review, Vol. 89/3, pp. 634-647, https://doi.org/10.2307/2082979.
[19] d’Addio, A. and M. Mira d’Ercole (2005), “Trends and Determinants of Fertility Rates in OECD Countries: The Role of Policies”, OECD Social, Employment and Migration Working Papers, https://www.oecd.org/social/family/35304751.pdf.
[12] D’Addio, A. and M. Mira d’Ercole (2005), “Trends and Determinants of Fertility Rates: The Role of Policies”, OECD Social, Employment and Migration Working Papers, No. 27, OECD Publishing, Paris, https://doi.org/10.1787/880242325663.
[17] Driscoll, J. and A. Kraay (1998), “Consistent Covariance Matrix Estimation with Spatially Dependent Panel Data”, Review of Economics and Statistics, Vol. 80/4, pp. 549-560, https://doi.org/10.1162/003465398557825.
[7] EVS/WVS (2021), “European Values Study and World Values Survey: Joint EVS/WVS 2017-2021 Dataset (Joint EVS/WVS).”, JD Systems Institute & WVSA. Dataset Version 1.1.0, https://doi.org/10.14281/18241.14.
[10] Fluchtmann, J., V. van Veen and W. Adema (Forthcoming), “Fertility, employment, and family policy: A regression analysis”, OECD Social, Employment and Migration Working Papers, https://doi.org/10.1787/1815199X.
[4] GGS (2020), Generations and Gender Survey 2020 Norway Wave 1, https://ggp.colectica.org/item/int.example/e101b705-934c-4dd0-8f09-d2d8b979c311/26.
[1] Hart, R. and Ø. Kravdal (2020), Fallende fruktbarhet i Norge: Hva kan det skyldes og hva kan man gjøre med det hvis det oppfattes som et problem?, https://www.fhi.no/contentassets/5e954d6441b045bc9b53a8e2d702b529/fallende-fruktbarhet-i-norge_rapport.pdf.
[13] Luci-Greulich, A. and O. Thévenon (2013), “The Impact of Family Policies on Fertility Trends in Developed Countries”, European Journal of Population / Revue européenne de Démographie, Vol. 29/4, pp. 387-416, https://doi.org/10.1007/s10680-013-9295-4.
[8] Ministry of Health and Welfare (2023), Tid for handling — Personellet i en bærekraftig helse- og omsorgstjeneste, https://www.regjeringen.no/no/dep/hod/id421/.
[6] Mynarska, M. and J. Rytel (2022), “Childbearing motivation at the onset of emerging adulthood”, Journal of Youth Studies, https://doi.org/10.1080/13676261.2022.2080536.
[15] OECD (2019), Rejuvenating Korea: Policies for a Changing Society, Gender Equality at Work, OECD Publishing, Paris, https://doi.org/10.1787/c5eed747-en.
[16] Rubin, D. (ed.) (1987), Multiple Imputation for Nonresponse in Surveys, John Wiley & Sons, Inc., Hoboken, NJ, USA, https://doi.org/10.1002/9780470316696.
[2] Statistics Norway (2022), Research Project: Fertility Decline, https://www.ssb.no/en/forskning/befolkning-og-offentlig-okonomi/fertility-decline.
[5] Stein, P., S. Willen and M. Pavetic (2014), “Couples’ fertility decision-making”, Demographic Research, Vol. 30/63, pp. 1697-1732, https://doi.org/10.4054/DemRes.2014.30.63.
[3] University of Oslo (2022), Falling Fertility and Rising Social Inequalities, https://www.sv.uio.no/iss/english/research/projects/Falling%20Fertility%20and%20Social%20Inequalities/.
[9] van Buuren, S. (2018), Flexible Imputation of Missing Data, Second Edition, Chapman and Hall/CRC, Second edition. | Boca Raton, Florida : CRC Press, [2019] |, https://doi.org/10.1201/9780429492259.
Effects sizes of a two‑way fixed-effects model with heteroskedasticity-corrected standard errors
|
Total fertility rate (log) |
Mean age at birth (log) |
---|---|---|
Public spending on family leaves (USD 1 000 PPP per child aged 0‑5) |
0.0117*** |
‑0.0 017** |
(0.0043) |
(0.0008) |
|
Public spending on ECEC (USD 1 000 PPP per child aged 0‑5) |
0.0128*** |
‑0.0011*** |
(0.0027) |
(0.0003) |
|
Public spending on family allowances (USD 1 000 PPP per child aged 0‑17) |
0.0130** |
‑0.0003 |
(0.0061) |
(0.0005) |
|
Weeks of paid maternity and parental leave available to mothers |
0.0005** |
0.0000 |
(0.0002) |
(0.0000) |
|
Weeks of paid paternity and earmarked parental leave for fathers |
‑0.0003 |
0.0000 |
(0.0004) |
(0.0000) |
|
ECEC enrolment rate 0‑2 years (%) |
‑0.0002 |
0.0000 |
(0.0004) |
(0.0001) |
|
ECEC enrolment rate 3‑5 years (%) |
‑0.0001 |
0.0001 |
(0.0004) |
(0.0001) |
|
Employment rate of women (%) |
0.0070*** |
‑0.0006** |
(0.0021) |
(0.0003) |
|
Employment rate of men (%) |
0.0043** |
‑0.0002 |
(0.0019) |
(0.0003) |
|
Share of part-time employees among women (%) |
‑0.0022 |
0.0003 |
(0.0037) |
(0.0003) |
|
Weekly usual hours worked by women in full-time jobs |
‑0.0307 |
‑0.0017 |
(0.0340) |
(0.0016) |
|
Weekly usual hours worked by men in full-time jobs |
0.0046 |
0.0005 |
(0.0130) |
(0.0007) |
|
Log GDP per capita |
0.7418 |
0.2991*** |
(0.9865) |
(0.1112) |
|
Log GDP per capita (squared) |
‑0.0442 |
‑0.0127** |
(0.0480) |
(0.0052) |
|
Average years of schooling among women aged 25+ |
‑0.0038 |
0.0024 |
(0.0146) |
(0.0024) |
|
Average years of schooling among men aged 25+ |
0.0079 |
‑0.0033 |
(0.0199) |
(0.0036) |
|
N |
421 |
421 |
Note: The table shows regression coefficients that capture effects of within-country, over time variation between labour market outcomes and log-transformed fertility outcomes. Estimates are based on a two‑way fixed-effects regression, with year and country fixed-effects as well as linear time trends for each country. The model is estimated over the period 2002‑18 using country-level data from Australia, Austria, Belgium, Canada, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Korea, Luxembourg, the Netherlands, New Zealand, Norway, Poland, Portugal, the Slovak Republic, Spain, Sweden, the United Kingdom and the United States. Missing values for average years of schooling and ECEC enrolment rates are handled through multiple imputation, using five iterations of predictive mean matching (van Buuren (2018[9])). The standard errors are heteroskedasticity- and panel-corrected and shown in parenthesis. ***, ** and * represent significance at 1%, 5% and 10% level, respectively. Detail on the methodology is available in Annex 1.B. and Fluchtmann, van Veen and Adema (Forthcoming[10]).
Source: OECD calculations based on data from the OECD Family Database, the OECD Employment Database, the OECD Social Expenditure Database, the OECD National Accounts, the UNESCO UIS Database and the UN World Population Prospects.
The regression in this report mainly explores the link that fertility outcomes have with family policy and labour market outcomes. This builds on earlier work, such as Ahn and Mira (2002[11]), d’Addio and Mira d’Ercole (2005[12]), Luci-Greulich and Thévenon (2013[13]), Adema, Ali and Thévenon (2014[14]) and OECD (2019[15]). All presented regressions use country-level data, rather than data on the level of the individual, findings therefore must be interpreted with caution and some caveats are discussed below. For a more detailed overview of the methodological approach, see Fluchtmann, van Veen and Adema (2023), forthcoming.
The specific outcomes and policy variables of interest are the Total Fertility Rate (TFR) and the Mean Age of women at the Birth of their child (MAB). The TFR is defined as the average number of children that would be born to a woman over her lifetime if she were to experience the exact current age‑specific fertility rates throughout her lifetime and if she lives from birth until the end of her reproductive life. It is obtained by summing the single‑year age‑specific rates at a given time. The MAB is defined as the average age of the mothers at the time of birth of their children, standardised for the age structure of the female population of reproductive age.
The base data obtained from the OECD Family Database, the OECD Employment Database, the OECD Social Expenditure Database, the OECD National Accounts, the UNESCO UIS Database and the UN World Population Prospects contains a substantial degree of missing observations. For this reason, only countries with widely available data over the period 2002 to 2018 are selected. In practice, this means that countries that became OECD members from 2010 and onwards are not included (Chile, Costa Rica, Colombia, Estonia, Latvia, Lithuania, Slovenia) as well as some other countries that are outliers in terms of their fertility rates (Israel, Mexico and Türkiye). Japan is not included in the regression because of the lack of comparable data on average usual working hours in full-time employment, while Iceland is dropped for a lack of comparable data on average years of schooling and Switzerland for insufficient data on social expenditure. Data for the remaining 25 OECD countries were included in the regressions (Australia, Austria, Belgium, Canada, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Korea, Luxembourg, the Netherlands, New Zealand, Norway, Poland, Portugal, the Slovak Republic, Spain, Sweden, and the United Kingdom).
Most data that enter the regression is generally widely available for the included countries, particularly for family policy expenditure and labour market outcomes. However, there is a substantial degree of missing data for childcare enrolment rates among 0 to 2 and 3‑ to 5‑year‑olds, which are an important indication of childcare availability, as well as for the average years of schooling of men and women, which are important control variables. In order to still include these variables in the regression without reducing the number of available observations through a listwise‑deletion process (i.e. dropping observations with any missing data entirely from the regression), missing data for these variables is imputed a using multiple imputation model based on predictive mean matching (see e.g. Rubin (1987[16]) and van Buuren (2018[9])). This maximises the number of regressors that can be utilised as well as the available statistical information – which is especially important in this small-sample cross-country setting. While these results need to be viewed with the necessary caution, there are no qualitative differences with a model that does not impute any observation and excludes childcare enrolment rates and average years of schooling. This approach is discussed in detail in Fluchtmann, van Veen and Adema (2023), forthcoming. The results are robust to excluding the imputed variables (regression tables are available upon request).
The regression controls for country specific effects on fertility to account for unobservable factors that differ between countries – for example due to cultural differences – by including country fixed effects. In addition, the regressions also control for linear movement in the outcome variable over time – such as changing attitudes towards parenthood – as well as for specific shocks to the outcome in each particular year, using linear time trends for each country as well as aggregate year fixed effects.
As laid out in in Adema, Ali and Thévenon (2014[14]), panel regressions as used in this report typically suffer from non-constant error variances across countries and years (i.e. heteroskedasticity) and autocorrelation. To account for this, the standard errors used in the regressions are heteroskedasticity- and panel-corrected following the approach laid out by Driscoll and Kraay (1998[17]) (the often used standard error correction for panel datasets by Beck and Katz (1995[18]) can be sensitive to panels with short time periods, as is the case here). As the data contains several country clusters, it would be ideal to estimate cluster robust standard errors. However, with many countries dropped and an overall small number of countries, cluster robust standard errors are unlikely to converge towards true standard errors, so that inference using the cluster robust estimator may be incorrect more often than when using the simple robust estimator. For this reason, the regressions do not correct for within-country correlation and only report heteroskedasticity-robust panel-corrected standard errors instead.
A possible limitation of the regression models is that the relationship between a regressor and fertility outcomes can also work in the opposite direction. For example, it is possible to assume that long working hours are discouraging women to have children, while it is also possible that high fertility is pushing women to work fewer hours on an aggregate level. For this reason, all independent variables in the regressions are lagged by one year. The rationale is that the decision of having a child (in most cases) occurs at least 9 months before the actual birth, thus the variables affecting the choice should be measured at the time of the decision, and not when the child is born. A further limitation of the model is that the unit of observation is the country average for each indicator. Thus, heterogeneity along economic, education, racial and regional lines between individuals in a country is not accounted for. Individual data would allow for a more robust identification: for example, the fact that having a secure job has an impact on a woman’s individual intentions of having children. On the contrary, the fact that a certain percentage of women in a country has a secure job may have a relationship with the total fertility rate of such country, but the link is less strong.