Hyunjeong Hwang
Economics Department of the OECD
Sunyoung Choi
Korea Institute for Health and Social Affairs
Sunyu Ham
Korea Institute for Health and Social Affairs
Yoonyoung Yang
Economics Department of the OECD
Jon Pareliussen
Economics Department of the OECD
Fertility rates have been steadily declining across the OECD, but the decline has been especially pronounced in Korea. As a consequence, the old-age dependency ratio is projected to increase faster than in any other OECD country, putting considerable strain on labour supply and public finances. Family formation patterns have evolved over time. In the decades up until around 2010 women increasingly delayed marriage and childbirth and reduced the number of children over their lifetimes, but almost all women eventually got married and had at least one child. From the 2010s, more women have chosen to remain unmarried or married and childless. The recent increase in births and marriages represents a catch-up after COVID-19. It is too soon to say if it also marks the start of a more durable increase in birth rates.
The average number of children women have over their lifetimes (Total fertility rate, TFR) has been steadily declining across the OECD, but the decline has been especially pronounced in Korea (Figure 2.1, Panel A). Korea’s TFR, which was six children per woman on average in 1960, dropped to just below one child per woman in 2018, and fell further year by year to reach 0.72 in 2023. This is the lowest in the world, well below the rate of 2.1 which is necessary to stabilise the population over the long term (abstracting from migration), and lower than Japan (1.2), China (1.2), Singapore (1.0) and Chinese Taipei (0.9) (Panel B; Yang et al., 2024). A fertility rate of 0.7 means roughly that for every 100 people in the current parent generation there will be only 35 children and 10 grandchildren (OECD, 2024).
Source: OECD Economic Surveys: Korea 2024.
Fertility only affects society with considerable lags. A large share of the Korean population is today in working age, so from a labour force, macroeconomic and fiscal point of view the challenges associated with low fertility and ageing have barely started. Some changes are nonetheless becoming apparent. The population already began to decline in 2021. Kindergartens are increasingly converted into nursing homes and wedding halls into funeral parlours (JoongAng, 2024) and the sales of pet strollers exceeded those of baby strollers on a leading e-commerce platform for the first time in 2023 (The Korea Times, 2023). This anecdotal evidence foreshadows the profound changes to come as large cohorts of baby-boomers reach retirement and working-age cohorts are continuously shrinking in size (Figure 2.2, Panel A).
Assuming that fertility remains around the current level and immigration follows past trends, the population is expected to halve over the next six decades with the elderly (aged 65 or older) then accounting for around 58% of the total population (Panel B). During this time, the old-age dependency ratio (the ratio of individuals aged 65 and over to those aged 20 to 64) will surge from 28% today to 155%. In the second and the third quarter of 2024, the number of births rose compared to the same period the year before for the first time in 14 years. This increase is at least in part a temporary rebound after the sharp drop in marriages and childbirths during the COVID-19 pandemic, but it cannot be ruled out that birth rates are now stabilising or even edging up on a more long-term basis (Statistics Korea, 2024a). However, even in a very optimistic scenario, where the fertility rate bounces back to 1.85, the dependency ratio would surge to 117%, still turning the country from one of the youngest in the OECD to the oldest (Panel C). The combination of a shrinking and ageing population poses a formidable challenge to sustaining social insurance systems and maintaining living standards. Labour shortages will intensify as retirees make up an increasing share of the population, and the fiscal cost of health, long-term care and pensions is set to more than double to 17.4% of GDP by 2060 (OECD, 2024).
Note: Panels A and B assume that employment rates by age group, the total fertility rate (0.7), net immigration inflows (30,000 per year), and employment rates by gender and five-year age cohorts all remain constant at current levels. Panel C assumes that the total fertility rate in all countries eventually converges to 1.85 children per woman, a medium-variant forecast from the United Nations’ World Population Prospects 2022. The old-age dependency ratio is defined as the ratio of individuals aged 65 and over to those aged 20 to 64.
Source: OECD Economic Surveys: Korea 2024.
Boosting fertility alone will not suffice to offset the rapid decline in the workforce. Even if a successful reform programme would almost miraculously succeed in tripling fertility to 2.1 children per woman by 2040, the level associated with a stable population in the long term, this would not prevent the workforce from declining. Lengthening working lives and increasing elderly employment are essential strategies to counteract the adverse consequences of ageing. Many workers in Korea express a desire to remain employed longer. According to a 2023 government survey of individuals aged 55-79, approximately 70% expressed a preference to continue working (Statistics Korea, 2024b). However, the average retirement age from their main career was only 52.7. Careers are often shortened by the seniority-based wage system, which makes older workers less appealing to firms when wages exceed productivity levels. Additionally, company-specific mandatory retirement ages below the legal pension age, along with practices encouraging older employees to retire early, contribute to premature retirement from primary careers. Reforming these systems, raising the statutory retirement age, and offering upskilling opportunities for older workers could help retain them in the workforce longer (OECD, 2024).
Increasing the supply of foreign labour is another vital strategy to remedy labour shortages going forward. Although immigration has increased and Korea has significant further potential to boost immigration, various visa hurdles for skilled immigration as well as poor working conditions for low-skilled immigrants hold back immigration and integration (OECD, 2024).
Lengthening working lives and boosting immigration, alongside a modest rise in the total fertility rate to 1.1, would result in a 12% higher level of GDP by 2070 (Figure 2.3, Panel A). These reforms could increase employment to 29 million by 2070, about 12 million higher than in the no-reform baseline (Panel B), and the employment rate would be 75%, around 18 percentage points higher than the no-reform baseline (Panel C). This would go a long way in alleviating fiscal burdens, with the structural primary revenue required to stabilise debt as a share of GDP settling slightly higher than today, but 7.8 percentage lower than the no-reform baseline in 2070 (Panel D).
1. Higher elderly employment scenario assumes increasing elderly employment by 1/3 of the difference with the previous age group by 2040 (e.g., the rate for the 60-64 group would rise to 1/3 of the 2022 rate for the 55-59 age group).
2. The baseline scenario is constructed by maintaining the current fertility rate of 0.7 throughout the projection period. In the higher immigration scenario, the net annual immigration inflow increases from the current 30 thousand to 250 thousand by 2040. In the longer working lives and boosting employment scenario, the following reforms are assumed: a) bringing up female employment rates to male employment rates by 2040; b) bringing up the youth employment rate to the current OECD average level by 2040; c) increasing elderly employment by one third of the difference with the previous age group by 2040 (e.g., the rate for the 60-64 group would close one third of the gap to the 55-59 age group in 2022); and d) increasing the legal retirement age to 68 by 2035, and raising it by two thirds of life expectancy gains thereafter. In the higher fertility rate scenario, the fertility rate is increased to 1.1 by 2040 which was the rate recorded in 2023 in Sejong city, the highest among cities in Korea. The scenarios are cumulative, so that, for example, the ‘higher fertility rate’ scenario also incorporates the effects from the ‘higher immigration’ and ‘longer working lives and boosting employment’ scenarios.
3. The rate of employment to adult population is defined as the ratio of total employment to population aged 20 and over.
4. The revenue that must be raised to keep the government debt ratio stable over time.
Source: OECD Economic Surveys: Korea 2024.
A low fertility rate does not mean that young people do not want children anymore. It rather reflects that obstacles are preventing would-be parents from having more children even though they want it and Korea as a nation needs it (OECD, 2024). Over the past two decades, the average number of children per married couple has decreased from around 1.5 to less than 1, while the ideal number of children married women desire has remained much higher (Figure 2.4, Panels A and B). This partly reflects a rapidly rising mean age of marriage, which has in turn increased the age when women have their first child to the highest in the OECD (Panel C). Foregone and delayed marriage have also led to a significant rise in childlessness. The share of women who have not experienced any live births by the age of 50 almost doubled from 7% in 2000 to 13% in 2020 (Panel D).
This chapter describes the declining fertility trend, while its causes are discussed in more detail in Chapter 3. A main takeaway is that Korean women have over time postponed childbearing and reduced the number of children they had over their lifetime at least partially as a response to the conflict between employment and childbirth. Until at least around 2010, Korean women would generally marry and have at least one child, reflecting the societal norm towards the duty to marry and have children. They would typically give up employment for childbirth when faced with a choice between the two, reflecting traditional male breadwinner – female carer norms. However, since around the mid-2010s married women’s employment and fertility choices seem to have become increasingly based on their individual work-family preferences, while obstacles to combine the two remain. As a result, fewer people are getting married, and a larger share of married couples remain childless.
Source: OECD Economic Surveys: Korea 2024.
Fertility has fallen rapidly since the 1950s, over a period that took Korea from a rural society ravaged by war to a modern, urban society with per capita incomes at the OECD average (OECD, 2024). Falling fertility in the beginning of the period can be related to rising female employment, but also to high population pressure and a wish to escape poverty, and active family planning initiatives from the state.
Dramatic changes occurred in the 1970s and the early 1980s, with the TFR plummeting from 4.5 in 1970 to 2.1 in 1983, a drop of 2.4 over just 13 years (Figure 2.1, Panel A; Kwon 1997; Kwon and Kim 2002). Between 1965-70 and 1985, the fertility rate decreased in all age groups (Table 2.1). The fall in fertility rates for ages 15 to 24 can be attributed to the rising age at first marriage, while the decline for ages 25 to 34 can be interpreted as the result of fewer births within marriage.
Note: Age-specific fertility rates refer to the number of live births per 1,000 females in each age group.
Source: Kwon and Kim (2002), Statistics Korea.
The timing of this dramatic fall inevitably links it to the state’s family planning and population control initiatives from 1962 to 1979, but causality is unclear (Choi et al., 2024). Many developing countries, including Korea, adopted policies to slow population growth in the latter half of the 20th century. In Korea, these aimed to reduce women’s unwanted births through a programme of information, basic maternal and child health services, and the provision of family planning supplies and services. Early assessments of the policies vary from being essential for economic growth and modernization and highly successful (Korea Institute for Health and Social Affairs, 1991) to those lambasting them for violating women’s privacy, eroding their autonomy, and excessively expanding the reach of the state into personal domains, all in pursuit of birth suppression (Lee,1989; Kim, 1993). More recent literature questions the narrative that family planning and falling birth rates was caused by centralised pressure, and proposes that the initiatives rather satisfied a latent demand for birth control and family planning (Kwon and Kim 2002; Bae 2012). In contrast to the 1960s, the need for birth control became a widely shared social value, and the ideal number of children dropped to about two. A possible negative long-term effect of the family planning programme is that it may have contributed to the norm that workers are expected to let work take priority over family. The developmentalist population policy pitted economic growth against childbirth and failed to recognise the value and contributions of parenthood (Hwang, 2005). This historical legacy hindered public discourses about these issues even during the 1980s and 1990s, when economic growth and democratisation had been achieved and fertility had dropped well below two children per woman.
After a period of stabilisation and a temporary rebound around 1990, the TFR continued falling to reach a new low of 1.18 in 2002. This was the first time it fell below 1.3, marking the beginning of what is often called in Korea the era of ‘ultra-low fertility’. It temporarily edged up to around 1.3 in the early 2010s before declining again, falling below one for the first time in 2018. In 2021, Korea already had the lowest TFR in the world at 0.81, followed by Spain at 1.19, Italy at 1.25, and Japan at 1.30 (OECD, 2023). By 2023, the TFR had fallen to 0.72, which is remarkably low, both in historical and cross-country comparison (Statistics Korea, 2023a). The annual number of births, which had stayed above 600,000 before 2000, has also declined considerably since. While it fluctuated between 400,000 and 500,000 during the first 15 years or so of the millennium, it was down to 230,000 by 2023.
Since the start of the millennium, a number of studies have decomposed the fertility trend into various driving forces. They all point to marriage patterns and childbearing behaviour within marriage as key forces behind falling fertility rates. Childbearing outside of marriage has only had a very limited impact, as almost all births occur within marriage in Korea. From the 1960s to the 2000s, a reduction in the number of children per married woman and a continuous increase in the age at which women first got married both contributed to the decline in fertility. Before the 2000s the decline in the number of married couples having three or more children lowered the birth rate, while in the 2000s a decline in the number of married couples having two or more children was the driving factor. From the 2010s, however, falling fertility was increasingly driven by married women having no children and women foregoing marriage altogether.
Earlier decomposition exercises pooled age-specific marriage- and fertility rates at a given point in time (synthetic cohorts), thereby masking behavioural changes within age cohorts. These studies tended to conclude that delayed marriage was the main cause of falling fertility, while falling fertility within marriage had a relatively small effect (Jun, 2002) or even that the birth rate within marriage increased during this period (Lee, 2012; 2018; Table 2.2). In other words, these studies concluded that couples had approximately the same number of children as before, but only later in life, so the TFR was bound to catch up in the future.
Source: Lee (2018).
In general, analysing age-specific TFR based on synthetic cohorts has limitations for capturing rapidly changing fertility behaviours. Breaking down changes in the TFR into components such as delayed childbearing and changes in fertility levels using synthetic cohorts underestimates the actual decline in the TFR, since delayed childbirth in older cohorts can wrongly be interpreted as a temporary delay also for younger cohorts, while childbirth does not necessarily recover over time within cohorts. This is especially important to understand the decline in fertility rates in Korea and other East Asian regions (Frejka et al., 2010, as cited in Shin, 2020).
Figure 2.5 shows that the fertility of women aged below 30 was higher in 2002 than in 2012, while fertility for women above this age was considerably higher in 2012 than in 2022. This illustrates why confounding the two trends of women delaying childbirth and having fewer children over their lifetime could lead researchers to interpret falling fertility as a mere delay when relying on synthetic cohort data in the past. In contrast, by 2022, fertility was either considerably lower or approximately the same as a decade earlier for all age groups, removing with the benefit of hindsight any earlier misconceptions that women would “catch up” by having more children later in life.
Source: Statistics Korea, Vital Statistics of Korea.
To overcome the methodological issues of synthetic cohorts, Yoo and Sobotka (2018) focus on a tempo-adjusted TFR, namely the counterfactual TFR that would have been achieved if there had been no significant changes in the mother’s age of childbirth by birth order (first child, second child, third child and so on). In line with previous findings, they showed that mechanically, without the shift to later childbearing, the TFR would have stayed higher. For example, the TFR was just below 1.2 in 2002. Adjusting for the tempo-effect, the TFR would have been around 1.7. In contrast to previous conclusions, they highlighted that a trend towards fewer children per woman had contributed decisively to lowering the TFR. They showed for example that having more than two children had become rare already by the late 1980s, and that the further falling incidence of having three or more children therefore had little impact on the subsequent decline in the overall TFR. The tempo-adjusted TFR for second births, which was around 0.9 (approximately 90% of women would have had at least two children, adjusting for the tempo-effect) in the 1980s and early 1990s, decreased steadily from the late 1990s to below 0.6 in 2015. The adjusted TFR for first births sat slightly below 1.0 in the early 1980s, and started to decline from the early 2000s. It dropped further to levels approaching 0.8 in the early 2010s.
Woo and Jang (2018) support these findings by studying the parity progression ratio, the share of married women with a certain number of children who later have an additional child (zero children to one, one to two and so on), with data from the Korean Longitudinal Survey of Women and Families. They show that the share of women (age 15-49) having at least one child fell only modestly from 97% during the 1965-1983 period to 92% during the 1984-1997 period to 87% during the 1998-2014 period. The likelihood of having a second child has fallen steadily from one time period to the next, while the likelihood of having three or more children was severely diminished already in the 1984-1997 period (Table 2.3).
Note: Parity progression ratios are defined as the proportion of women with a certain number of children that later have an additional child, measured for the total population of women aged 15-49 in any given year.
Source: Woo and Jang (2018).
To overcome the shortcomings of previous studies and give a more accurate and complete picture of fertility behaviour, a number of studies are now focusing on estimating and analysing cohort fertility rates (both completed and uncompleted) (Shin et al., 2019; Yoo, 2020, 2022; Oh, 2018; 2019; Kye et al., 2022). These studies have all found that since the 1969 birth cohort, the age of women’s first childbirth has steadily increased and fertility levels have decreased (Figure 2.6). For instance, Yoo (2020) showed that while the most frequent childbearing age has shifted from the mid to late twenties to the early to mid-thirties, the fertility levels of women in their early to mid-thirties today do not compensate for the past fertility levels of women in their mid to late twenties.
Source: Shin et al. (2019).
In the 2000s, fertility was mainly related to the childbirth behaviour of women born from the late 1960s to the late 1970s, while childbirth behaviour in the 2010s was associated with women born between the late 1970s and the early 1990s. There are important differences between these two generations, with rising education levels among the most obvious. It is still too early to study the fertility decline from the mid-2010s until today based on completed cohort fertility rates, as women in childbearing age at any given time span decades of birth cohorts.
However, past patterns and the composition of current childbearing cohorts point to a continued downward trend in fertility. First, childlessness among married women is increasing in every relevant age group (Figure 2.7). This trend is also evident from the perspective of birth cohorts. According to the analysis by Cho et al. (2024), the rate of childlessness among married women born before 1970 was very low at around 3%. However, women in more recent birth cohorts are far more likely not to have children. For instance, 9.6% of married women born in 1980 did not have children until they became 40 years old (Figure 2.8).
Source: Statistics Korea, Population Census.
Source: Cho et al. (2024).
Second, childbirth outside of marriage remains highly uncommon, and the share of the population which is married has kept falling across age groups and gender. In 2000, over 90% of men in their early forties and women in their late thirties were married (Figure 2.9). For most of the 20th century, marriage was common in Korean society. However, the portion of married individuals has declined significantly across nearly all age groups from the early 2000s. By 2020, the marriage rate did not surpass 80% even in the age groups most likely to have tied the knot. Even taking into account that marriages in 2020 were delayed due to the pandemic would likely not affect these numbers much, since this is the stock of married people, which will not change drastically based on one single year for most age groups. The reduction was particularly dramatic in age groups where first marriages were most common, namely people in their late twenties and early thirties. For instance, in 2000, 59.6% of women aged 25-29 were married, but by 2020 this had dropped to 18.9%. Over the same period the marriage rate for women in their early thirties fell from 87.2% to 53.1%. The rate for women in their late thirties dropped from 90.9% in 2000 to 72.9% in 2020. For men of the same age, it fell even more sharply from 86.9% to 60.1%.
Source: Statistics Korea , Population Census.
Some question marks remain, which could potentially change this assessment as more data becomes available in the future. In the third quarter of 2024, the birth rate rose compared to the same period the year before for the first time in 14 years. The marriage rate has also been on the rise after reaching all-time lows during the Covid-19 pandemic. This increase is of course good news, but both marriage rates and birth rates remain well below their pre-pandemic levels (Figure 2.10). It seems clear that the rebounding marriage rate is largely a catch-up effect after the pandemic, when many marriages were postponed. Childbirth typically follows marriage with a lag, and the rebound in birth rates can therefore also be interpreted as a lagged effect of the pandemic, which is likely to be temporary in nature. However, it is too early to quantify the effect on fertility of the COVID-19 pandemic in 2020 and 2021, and thereby too early to conclude that this is the full story. In 2018, the statutory work week was limited to 52 hours. This momentous change in favour of family-work balance is not obviously visible in fertility statistics, but positive effects on fertility could have been masked by the subsequent pandemic. Designating low fertility as a national emergency along with a steady expansion of family policies can also have helped to give young couples the needed confidence to bring children to the world.
However, as discussed in subsequent chapters, any fertility revival will at best be slow and gradual until policies, gender norms and working practices taken together enable a large majority of women to pursue career and family in tandem. In the meantime, the cohorts of women in childbearing age keep shrinking and the large cohorts born in the 1950s, 60s and 70s keep ageing. Even in a very optimistic scenario, where birth rates more than double, Korea will turn from one of the youngest countries in the OECD to the oldest in a few decades.
Rates of birth and marriage (12-month moving average) per 1000 women aged 20-45
Source: Statistics Korea Population Statistics (database).
Bae, Y. (2012), Human Reproduction in the Korean Modernity, Sigan Yeohaeng (in Korean).
Choi, S., S. Ham, Y. Yang and J. Pareliussen (2024), “Women's employment and fertility in South Korea: a review of recent research”, OECD Economics Department Working Papers, No. 1825, OECD Publishing, Paris.
Cho, S., Y. Shin, K. Kim, D. Hyun, J. Choi (2024). A Comparative Study of Life Course Transitions Across Birth Cohorts: Focusing on Young Adulthood. Korea Institute for Health and Social Affairs, Statistics Research Institute (in Korean) (forthcoming).
JoongAng Daily (2024), “As birthrates plummet, kindergartens become nursing homes”, Korea JoongAng Daily, 17 January.
Jun, K. (2002), “Fertility”, Population of Korea, Statistics Korea (in Korean).
Kim, E. S. (1993), The Making of the Modern Female Gender: The Politics of Gender in Reproductive Practices in Korea, PhD Thesis, University of California, Berkeley.
Korea Institute for Health and Social Affairs (1991), Thirty Years of Population Policies, Korea Institute for Health and Social Affairs (in Korean).
Kwon, T. et al. (1997), “The process and meaning of fertility transition in Korea”, Understanding Fertility Transition in Korea, Ilsinsa (in Korean).
Kwon, T. and D. Kim (2002), Understanding Population (revised edition), Seoul National University Press (in Korean).
Kye, B., S. Yoo and S. Choi (2022), “Trends in marital fertility, 2005~2020: duration-based approach”, Korea Journal of Population Studies, Vol. 45/4 (in Korean).
Lee, C. (2018), “Did pro-natal policy in Korea fail?: a decomposition of fertility change from 2000 to 2016”, The Korean Journal of Economic Studies, Vol. 66/3 (in Korean).
Lee, C. (2012), “A decomposition of the decline in the total fertility rate in Korea: effects of changes in marriage and marital fertility”, Korea Journal of Population Studies, Vol. 35/3 (in Korean).
Lee, M. (1989), “A feminist analysis of the state policy in birth control”, Women's Studies Review, Vol. 6 (in Korean).
OECD (2024), OECD Economic Surveys: Korea 2024, OECD Publishing, Paris.
Oh, J. (2019), “Forecast of and factors driving a double-dip in Korea’s fertility rate”, The Korean Journal of Applied Statistics, Vol. 32/4 (in Korean).
Oh, J. (2018), “A Study on forecasting cohort incomplete fertility for Korea”, Korea Journal of Population Studies, Vol.41/1 (in Korean).
Shin, Y., M. Lee and S. Park (2019), Analysis and Outlook of Korea’s Long-term Trends in Lowest-low Fertility, Korea Institute for Health and Social Affairs (in Korean).
Statistics Korea (2024a), Population Trends for September and the Third Quarter of 2024.
Statistics Korea (2024b), Statistics Korea Supplementary Surveys on the Economically Active Elderly Population, Statistics Korea.
The Korea Times (2023), Sales of Pet Strollers Surpass Baby Strollers for 1st time, The Korea Times, 27 December.
Woo, H. and I. Jang (2018) “Fertility transition in South Korea: application of survival analysis”, Journal of the Korean Official Statistics, Vol. 23/2 (in Korean).
Yang, Y., H. Hwang and J. Pareliussen (2024), “Korea’s unborn future: lessons from OECD experience”, OECD Economics Department Working Papers, No. 1824, OECD Publishing, Paris.
Yoo, S. (2022), “Postponement and recuperation of fertility in South Korea: estimating and projecting completed fertility”, Journal of the Korean Official Statistics, Vol. 27/1 (in Korean).
Yoo, S. (2020), Analysis of Total Live Birth Changes, The Population Association of Korea (in Korean).
Yoo, S. and T. Sobotka (2018), “Ultra-low fertility in South Korea: the Role of the tempo effect”, Demographic Research, Vol. 38/1.