This entry focuses on the number of births per woman in a population. The most commonly used metric is the Total Fertility Rate (TFR) – or often simply 'fertility rate' – which measures the average number of children per woman.1
The global average fertility rate is just below 2.5 children per woman today. Over the last 50 years the global fertility rate has halved. And over the course of the modernization of societies the number of children per woman decreases very substantially. In the pre-modern era fertility rates of 4.5 to 7 children per woman were common. At that time the very high mortality at a young age kept population growth low. As health improves and the mortality in the population decreases we typically saw accelerated population growth. This rapid population growth then comes to an end as the fertility rate declines and approaches 2 children per woman.2
The first section of the entry presents the global empirical perspective on the number of children per woman. The long second section presents the academic research that answers the question why the number of children per woman declined. Particularly important are 1) the empowerment of women in society and in relationships – through education, labor force participation, and strengthened women's rights – and 2) the increased well-being and status of children.
The number of children per woman was very high in the past and up to the second half of the 20th century. Globally, up to 1965 the average woman had more than 5 children.
As the first visualization below shows this changed over the last 50 years. Over the last 50 years the global fertility rate has halved and globally the average woman has fewer than 2.5 children today.
By clicking on '+ Add Country' below the chart you can compare this global decline with particular world regions or countries.
The previous chart shows the global average, the following chart shows the decline of fertility rates for all countries in the world from 1950 to today.
The chart below is a bit unusual, but once you wrap your head around it, it reveals a lot of information.
If you look at the red line you see the countries of the world ordered descending by the fertility rate in the period between 1950 to 1955. Rwanda, Kenya, the Philippines and also other countries that are not labelled in this chart had a fertility rate higher than 7 children per woman. China had a fertility rate of just over 6 and India a fertility rate of just under 6. On the very right of the red line you see that in 1950-55 there was only one country in the world with a fertility rate below 2: tiny Luxembourg. The width given to each country in this chart corresponds to the share of that country’s population in the total global population at that point in time – this is why China and India are so very wide. All countries in the world are plotted, but because the space is limited not all countries are labelled.
What we can see then is that in the 1950s, the world is clearly divided between countries with high and countries with low fertility rates. On the right-hand side of the chart we see countries where women have fewer than 3 children – in these countries the fertility rate had declined already in the decades before. As we will see below fertility rates were high in all countries in the distant past.
Looking at the orange line, you see that until 1975-80 some countries substantially reduced their fertility: China's fertility rate fell to 3 (this was largely before the introduction of the 'one child policy'). And other countries maintained very high fertility levels. In Yemen, the fertility rate was 8.9 children per woman in 1985. The global average was still close to 4 children per woman.
Since then the world has changed substantially. The blue line shows how. Globally, the fertility rate has fallen to 2.5 children per woman and low fertility rates are the norm in most parts of the world: The huge majority of the world population – 80% – now live in countries with a fertility rate below 3 children per woman.3 On the other end of the spectrum there are a few countries – home to around 10% of the world population – where women on average have still more than 5 children.
We also see convergence in fertility rates: the countries that already had low fertility rates in the 1950s only slightly decreased fertility rates further, while many of the countries that had the highest fertility back then saw a rapid reduction of the number of children per woman.
Comparing the red, orange and blue lines also makes it possible to see the change in single countries: In Iran for example, the fertility rate in 1985 was 6.2 children per woman; today women in Iran have fewer children than in the US, the UK, or Sweden: 1.7 children per woman. In Thailand (chart here), the fertility rate in 1950 was 6.1, in 1985 it was 2.6, and today it is 1.5 children per woman.
I have also made a second version of this chart where I included the projections for the 21st century. The UN expects global fertility to fall further in most countries so that the global fertility rate rate will be just below 2 children per woman by the end of the century.
World population by level of fertility over time, 1950-2010
The decline of the fertility rate is one of the most fundamental social changes that happened in human history. It is therefore especially surprising how very rapidly this transition can indeed happen.
The chart below shows the speed of the decline of fertility rates. As we see from the chart below it took Iran only 10 years for fertility to fall from more than 6 children per woman to fewer than 3 children per woman. China made this transition in 11 years – before the introduction of the one-child policy.
We also see from the chart that the speed with which countries can make the transition to low fertility rates has increased over time. In the 19th century it took the United Kingdom 95 years and the US 82 years to reduce fertility from more than 6 to less than 3.
This is a pattern that we see often in development: those countries that first experience social change take much longer for transitions than those who catch up later: Countries that were catching up increased life expectancy much faster, they reduced child mortality more quickly and were able to grow their incomes much more rapidly.
How long did it take for fertility to fall from 6 children per woman to fewer than 3 children per woman?
The UN data shown above only shows the change from 1950 onwards. By then the richest countries had already experienced substantial decreases of the fertility rate and it would be a mistake to believe that these countries did not see high fertility rates in the past.
The table below shows fertility rates in Europe before 1790. Back then one woman gave on average birth to 4.5 to 6.2 children. The population of a society does not increase when every woman is replaced on average by two children. As the tables presents fertility rates when the population in these countries did not yet grow rapidly we can infer that on average 2.5 to 4.2 children died per woman.
Age of Marriage of Women and Marital Fertility in Europe before 17904
|Country or Region||Mean age at first marriage||Births per married women||Percentage never married||Total fertility rate|
The following map shows the estimates published by Gapminder from the year 1541 onwards for countries where data is available. Fertility rates were estimated to be very high across the entire planet until very recently.
Here we show only data backed up with published estimates within the academic literature or United Nations Population Division. Gapminder also publishes long-run estimates for all countries – but stresses that these estimates come with high uncertainty. The full dataset can be found here.
The stacked area chart shows the number of births globally. In 1950, 97 million children were born, in 2015 the world saw 141 million births – these are 386,000 births every day.
In addition to the total fertility rate a second commonly used measure is the birth rate. The birth rate is expressed as the annual number of births per 1,000 people in the population.
Women's empowerment, the increasing well-being and status of children, technological and economic changes, changing norms, and opportunities for family planning matter for the reduction of the total fertility rate that I documented above. Below I will review both the theoretical explanations of how each of these aspects impacted the number of children women have and also present the empirical research that investigates these explanations.
What makes precise accounting difficult is that the different explanations of declining fertility are not mutually exclusive. But my sense from reading the literature is that over the long-run the two first explanations – women's empowerment and the increasing well-being and status of children – have been the two most important factors in most places.
The level of education in a society – of women in particular – is one of the most important predictors for the number of children families have. Before I am looking at the data and the empirical evidence in the research literature that establishes why increasing education is leading to a declining number of children per woman we should ask why and how exactly women's education is linked to the choice about children. We should look at the theory.
The choice for having a child is a question of opportunity costs and education changes them
Much of the theoretical work in recent decades on how families decide how many children they want rests on the models of the economist Gary Becker.5 His framework models the demand for children in the way the demand for other goods in life are modeled, the demand for children is tied to the ‘price’ of a child.
Prize, in this framework, is thought of as a much broader concept then just the monetary costs parents incur. It includes the direct costs of the child – much of which are monetary costs, such as the costs for childcare or schooling for example –, but it also includes the indirect costs such as the opportunity cost in time that is needed for pregnancy and upbringing of the child.
In many societies mothers spending more time with their children than fathers and so the opportunity costs in particular are mostly born on the mother’s part.
In the framework of Becker it can be understood why improving the education of women leads women to want fewer children. It is because the opportunity costs of better educated women are higher that they are less likely to want a large number of children: Women who are better educated have to turn down more opportunities than women who are less well educated and so the 'prize' they have to pay for having children is higher.
These effects of education on the fertility rates – which can amplify the effect on women's opportunity costs – is the topic of the following section.
Positive feedback via the health of children
There is evidence, which we discuss in our entry on child mortality, that better education of mothers is having a positive impact on better health and lower mortality of the children. Further below I will review the evidence that lower child in turn leads to a decrease of the total fertility rate. Taken together these two pieces of evidence suggest that better education of women reinforces the direct effect it has on fertility through an additional indirect effect via better child health.
Positive feedback via contraceptives
Education is also important for the knowledge and use of contraceptives and the ability of better educated women to reduce the gap between the desired and the actual number of children is an additional positive feedback effect by which better education reduces the number of children. Chicoine (2012) finds evidence for the importance of education in this regard.
Positive feedback via lower fertility
The shortest loop of positive feedback runs through fertility itself. Education is not only reducing fertility, lower fertility also allows for better education. Better education of women thereby reinforces itself, both within as well as across generations.
Evidence for this two-way reinforcing relationship can be found in the historical transition to lower fertility in Prussia which was studied by Becker, Cinnirella and Woessmann (2010).6
The effect that better education has on lower fertility can reinforce itself also over subsequent generations. As the fertility rate declines the education system faces smaller and smaller cohorts of school children for which it can better provide. And additionally parents with fewer children also have more opportunities to nurture and support each child. This is a kind of demographic dividend on education.
Better education makes it possible for social norms to change
In both historical and contemporary episodes of declining fertility researchers have found strong evidence that social norms are important in reducing the number of children that parents desire – I will present some of this evidence below. Education seems to be a key prerequisite for these changes to take hold.
Amartya Sen discusses this in his book ‘Development As Freedom’7 with regard to India. He writes: “There is, in fact, much evidence that the sharp decline in fertility rates that has taken place in the more literate states in India has been much influenced by public discussion of the bad effects of high fertility rates especially on the lives of young women, and also on the community at large. If the view has emerged in, say, Kerala or Tamil Nadu that a happy family in the modern age is a small family, much discussion and debate have gone into the formation of these perspectives. Kerala now has a fertility rate of 1.7 (similar to that in Britain and France, and well below China's 1.9), and this has been achieved with no coercion, but mainly through the emergence of new values—a process in which political and social dialogues have played a major part. The high level of literacy of the Kerala population, especially female literacy, which is higher than that of every province of China, has greatly contributed to making such social and political dialogues possible (more on this in the next chapter).” In 2016 – 17 years after Sen wrote this paragraph – the fertility rate in Kerala is still 1.7 children per woman.
Empirical findings on the link between women's education and the number of children
A great number of studies confirms that higher education of women is associated with lower fertility. Studies look at this relationship on both the social and on the individual level. While some studies achieve to only show a statistical correlation, others also establish a causal relationship between rising education and a decreasing number of children.
I am discussing a series of these studies below, but to understand how women's education and the total fertility rate are related the visualization below visualizes how these two aspects have changed over 6 decades. Each arrow in this plot shows for one country how the average number of children per woman (on the y-axis) and the years of education of women in the reproductive age (on the x-axis) have changed. It is possible to adjust the time slider below the chart and if you put start and end point to the same year you'll see a straightforward correlation.
What we see in the arrow-plot is that when women had on average fewer than 2 years of education back in the 1950s the fertility rate was between 5 and 8 children. Six decades later most women are much better educated and often have 8 or more years of education on average: As we would expect from the theory above this meant that they have much fewer children, where women have more than 8 years of education the fertility rate is less than 4 children per woman and mostly lower than 2.
You can search for particular countries with the search box in the bottom left – have a look at Iran: in 1950, when Iranian women had on average only a third of a year of schooling, they had on average 7 children. Sixty years later when Iranian women had on average 9 years of schooling they had on average 1.8 children (yes, women in Iran have today fewer children than women in the US).
In countries where women today still have only little access to education the fertility rates are still high. In Niger, the country with the highest reported fertility rate in 2010, women in the reproductive age had only 1.3 years of education on average. This is why, if you are concerned about population growth, you should be an advocate for giving women access to education.
Macro studies: Where women are better educated they tend to have fewer children
The correlation above is in line with what we should expect based on Becker's theory, but it is still a correlation only and far from conclusive evidence that there is a causal link between education and the number of children that women have.
Several studies go one stop further and are not only looking at two variables, but instead manage to additionally control for possibly confounding variables. These macro studies are often especially interesting because they can study historical changes where micro data – on individual families – is not available.
Becker, Cinnirella, and Woessmann (2013)8 study Prussia before the demographic transition in the 19th century and find that higher education of women is associated with lower fertility. After controlling for several other factors and after additionally employing instrumental-variable techniques the authors suggest that this relationship is indeed causal.
In another macro studies study that focuses on the decline of fertility rates in several countries over the period from 1870 to 2000 Fabrice Murtin (2013)9 investigates empirically which drivers are associated with the reduction of the number of children per woman. The author also looks at changes in health and prosperity, but ultimately finds that it is improving education that is the ‘main socioeconomic determinant of the demographic transition’. To give an indication of the effect size the author writes: “When average years of primary schooling grow from 0 to 6 years, fertility should decrease by about 40% to 80%.” The two-variable-association of education and fertility rates in the arrow plot is in the same ballpark – countries in which women's education increased from close to 0 to around 6 years, experienced a decline of around 40%.
Micro studies: Women who are better educated tend to have fewer children
Micro studies are not aggregating the measured aspects on the societal level – such as the country level – and instead allow to track the circumstances and behaviors of individuals. This is particularly helpful in our question because it is so difficult to disentangle the causal relationship between fertility and education – while better education is possibly related to lower fertility, it would also be reasonable to expect that the other way around, lower fertility increases the opportunities for women to receive better education and that this is the reason we see a correlation.
Micro studies can shed light on these seemingly intractable problem, particularly when they are set up as randomized evaluations. Duflo, Kremer, and Dupas (2015)10 conducted such a large randomized evaluation in western Kenya over a period of 7 years in which they found that subsidies to education decreased the rate of pregnancies in adolescent girls. Because we can study the effect of the policy intervention in this evaluation, we can see that the relation in question is not driven by an impact of the number of children on the education of women.
Luke Chicoine (2012)11 studied the impact of another policy change, this time in Kenya, that lengthened primary school by one year from 1985 onwards and finds that the additional education of women lowered the number of children they had. The author emphasizes that the increased early use of modern contraceptives of better educated women was instrumental to make this reduction of fertility possible.
Brievora and Duflo (2004)12 studied the link between education and the number of children women want by investigating the impact of a school construction program in Indonesia between 1973 and 1978. The authors again find that a better education of women is associated with a lower number of children.
And in yet another careful micro-study, Osili and Long (2008)13 investigated the introduction of universal primary education in Nigeria and equally find that an increase of women’s education by one year reduces early fertility by 0.26 births.
The evidence from these micro studies is exceptionally clear and suggests that increasing education for women is indeed leading to a decrease of the fertility rate.
And if we go beyond the country averages shown in the previous visualization we also see that within societies there is a strong correlation between women's education and the number of children they have. In the visualization below I show the evidence for all the countries in the world where the fertility rate is still above 5 children per woman. In these countries too it is true that more highly educated women have substantially fewer children. Those mothers with secondary education have typically fewer than 5 and often fewer than 4 children. Those with higher tertiary education have always fewer than 4 and sometimes even fewer than 2 children. A substantial reason why the fertility rates in these countries are so very high is just that very few women actually have secondary or even tertiary education – as is shown in the column on the left in all countries more than 90% of women have no tertiary education.
The increasing labor force participation of women is a second aspect of women's rising empowerment in society and this change too tends to lead to a decline of the number of children that women have. This change is so closely linked to the rising education of women discussed before that it is indeed impossible to separate from that. A substantial part of the increased opportunities that better education offered were realized in the labor market and it can be argued that the best way to understand how education matters for fertility rates is to view it together with women's increasing labor force participation.
The theoretical argument for why women want fewer fewer children as their chances in the labor market are increasing can be explained with reference to the same framework proposed by Gary Becker that I laid out above: As women increasingly participate in the labour market their opportunity costs for having children are rising so that they seek to have fewer children.
There is again a two-way relationship: In addition to the reasons to expect that increasing labor force participation of women leads women to have fewer children, it is also obvious to consider that the reverse is true. And indeed there is strong evidence that the decline of fertility rates increased women's participation in the labor market, we review this evidence here. This two-way-relationship leads to a cycle which can enforce itself: As women decide to have fewer children and are increasingly participating in the labor market they have yet more reason to have fewer children.
Women's labor force participation increased for several reasons. Beginning with the industrialization, labor markets underwent historical changes which made the increasing labor force participation of women possible. In the poorer economies of the past the huge majority of workers were working in the agricultural sector. Work on the farms was physically extremely demanding and men had a strong comparative advantage in this labour market over women. With the shift away from agriculture towards manufacturing and services and an increasing importance of education due to technological change this comparative advantage of men was eroded and over the long-run the labour force participation of women increased.
The close link between economic growth, structural transformation, education, and population growth are the focus of the Unified Growth Theory. The link between the increasing demand for female labor, which increased the opportunity costs of fertility, and thereby lead to a falling number of children per woman are a key component of this literature.14
In addition to the technological changes from which women benefitted, women's rights activists successfully fought for the possibility for women to work in professions from which women were previously banned without good reasons and once were allowed to become teachers or doctors the opportunities in the labor market increased even more.
We review the reasons for the increasing labor force participation of women in this post.
Studies that explore the historical decline from high fertility rates in largely agrarian economies to low fertility rates in industrialized countries are published in the already mentioned literature on the Unified Growth Theory. The economist Oded Galdor, one of the leading authors in this field, provides an overview in his book that bears the name of the theory as the title.15
There is also more recent empirical evidence from developing countries that increasing labor force participation of women contributed to the decline of the fertility rate. Jensen (2012)16 investigated this by conducting a study in India where the researchers provided three years of recruiting help to young women in randomly selected villages. The intervention thereby exogenously increased the labor market opportunities of these women and the author found that in response young women chose to either enter the labor market or obtained more education and both of these decisions meant that they postponed having children and they crucially also reported to want fewer children.
In a scatter plot of today's labor force participation of women and the fertility rate we do not see a close correlation between the two measures however. Part of the explanation for this is that women's labor force participation is not only high in rich countries today, but also in poor countries – the relationship to income is U-shaped. But if we look at the change over time of the Americas for example – a region with strong growth and increasing labor force participation of women – then we see indeed that the fertility rate declined as the participation of women in the labor market increased.
Since the burden of child-birth and mostly also of the upbringing of children is borne by women, it is not surprising that fertility rates tend to be high where women have a lower social status and few opportunities outside the household. It is only when greater importance is given to the interests of women that this changes. Women’s better education and women’s increased employment opportunities both changed the role of women in society and increased the social status of women in society. With more outside-options to having a large number of children women opted increasingly to take advantage of these options and consequentially the total fertility rate declined. This can lead to virtuous cycles since lower fertility rates give women the freedom to do things other than childbearing and this in turn leads to a decline of fertility rates.
The good thing about this is that these changes are desirable in their own right. This makes clear that one of the best strategies to achieve lower fertility rates is to work towards reducing gender inequality and supporting women's empowerment and a rise of women’s power, status, and education relative to men.
Rapid population growth has been a temporary phenomenon in countries around the world. Rapid population growth starts when the health of the population improves and the mortality rate in a population decreases while the birth rate stays as high as before. Rapid population growth then comes to an end when after some time the birth rate follows the decline of the mortality rate. The model of the demographic transition formalizes this relationship between mortality, fertility, and population growth. Is the regularity of this co-movement of mortality and fertility rates just a coincidence?
The theoretical and empirical literature suggests that these changes regularly coincide with additional changes – such as those discussed above – and that the link is therefore partly driven by changes of third factors. But the literature additionally suggests that there is a important direct and causal effect of declining mortality – particularly of children – on the number of children parents are having. The sections below discuss first the theoretical reasoning and then the empirical evidence.
In an environment with high child mortality women will give birth to more children than they want to ensure against the loss of children
The theoretical argument for the link between high mortality – particularly at a young age – and families' decisions for a certain number of births becomes understandable when one considers how families who want a certain number of children are forced to think in a high mortality environment. If parents have a certain target for a number of surviving children then the number of children the women gives birth to will need to be higher when the level of child mortality is higher.17 Two mechanisms are at work – none of which have a particular pleasant name in the demographic literature:18
- "Child replacement": Parents that experience the death of a child might deliberately decide for an additional birth in order to “replace” a child that has died.
- "Child hoarding": Demographers speak of hoarding when a family decides to have more births than their desired number of children in order to protect themselves against the possibility of future high mortality in the family. The motivation for ‘hoarding’ is best understood in contrast to the motivation for replacement. Replacement is the response to the experience of a child death while hoarding is the response to the expectation of child mortality. This thinking becomes understandable when one considers that in pre-modern societies more than one third of all children die before they are five years old.
Of these two effects only hoarding can explain why a decline of fertility follows a decline of child mortality in the way it is described by the model of the demographic transition. As child mortality decreases, parents gradually learn that the risk of child death is decrease and see that hoarding is not needed anymore. Finally in the absence of child mortality parents can decide directly the number of children that they want. Consistent with the model of the demographic transition there might be a time lag between declining mortality and declining fertility as parents will only in retrospect learn that the environment has changed and the risk of child death is declining.
Empirical findings on the relationship between child mortality and the number of children
Some empirical research studies the link in question on the aggregate level, and we will turn to this below, but there is also research that studies decisions at the household level and asks whether there is evidence for the two mechanisms laid out above.
There are a number of publications that find evidence for child replacement. A recent one is the study by Reher et al. (2017)19 which studies European countries through the demographic transition. And there is also empirical evidence for child hoarding: Raaj Say (1991)20 and Sebnem Kalemli-Ozcan (2003)21 both find evidence for the importance of hoarding and therefore for the reduction of child mortality rates as a driver of decreasing fertility rates.
Several macro studies look at the relationship and provide evidence:
Luis Angeles (2010)22 investigates the relationship in a large set of developed and developing countries from 1960 onwards and finds a large effect. Crucially the author finds a lag of about 10 years for the decline of child mortality to translate into declining fertility.
Palloni and Rafalimanana (1999)23 investigate the effect of declining child mortality on fertility in Latin American countries between 1920 and 1990 and while they find that lower child mortality leads to lower fertility rates the authors caution that this effect “may be too small to support the hypothesis that changes in child mortality are of more than modest importance in the process of fertility decline in Latin America in the late twentieth century”.
In the visualization below, we can see correlation that we expect. Countries with high child mortality rates tend to have much higher fertility rates, while countries with low child mortality rates experience lower fertility rates. All countries in which the child mortality rate is higher than 9% have fertility rates higher than 4.5 children per woman.
When you press the 'play'-button in the bottom left of the chart you can see how these demographic aspects changed over time: Countries start out in the top-right corner when children die frequently and women give birth often. But as child health improves and fewer children die we see in country after country that the fertility rate falls and countries move into the bottom left corner of the chart. Consistent with the explanation of hoarding we see that women do not immediately reduce the number of birth and only after a lag of some years will the fertility rate adjust to the lower mortality rate.
The decline of child labor
An aspect emphasized already is that the high number of children in the past is not an accident. On the contrary, families wanted many children because they needed many children. In the agricultural, poor economies of the past children were contributing to the household productively from a young age on. Child labour was very widespread as we show in our entry on child labor.
This changed when the economy modernized. Hazan and Berdugo (2002)24 document that with technological progress and the structural change of the economy away from agriculture children's wages were falling relative to the wage adults and argue that this declining importance of child labor contributed to the decrease in the number of children that parents want.
Technological changes were closely tied to political changes and indeed economic and technological development, with the shift from a low tech to a high tech economy, amplified the changing moral perspective on child labour and contributed to a decreasing employment and thereby to a declining demand for children.
Döpke (2004)25 studies how government policies affected the decline of the fertility rate and finds that the increasing restrictions for child labor in particular were crucial because they increased the opportunity costs of having children as parents did not benefit from the child’s contribution to the household. The fact that restrictions in child labor had an effect on fertility rates is also strongly suggesting that indeed the importance of child labor before was an important incentive for the high number of children parents had before.
Döpke also argues that child labor restrictions and public provision of education were crucial for the rapid growth of the East Asian economies. These government policies were important for the reduction of the number of children per woman and the fast economic growth in the region.
In today's rich economies children have vastly more education than in the poor agrarian economies of the past. The basic argument for why the increase of education contributed to the decline of fertility rates derives again from the seminal work of Becker (1960) who argued that because of the costs of bringing up a child parents have to make a decision between the number of children they want (quantity) and the resources they want to spend on each child (quality). Limited resource force parents to decide to either have many children – but then have few resources (time & money) available for each child – or to have fewer children and then to be able to have more resources available for each child. This tradeoff is described in the literature as parents' choice between the "quantity" or the "quality" of their children.26
The argument in a nutshell is that educating children is very costly and since parents have limited resources the increasing costs of having children forced them to have fewer children.
Lower child mortality increases the incentives to invest more resources into children
A first extension of the argument linking fertility rates and children's education explains that this association is partly driven by the decline of child mortality. The model developed by Soares (2005),27 argues that declining child mortality changes parents incentive in the quantity-quality-tradeoff described above. In the high mortality environment of the past, investments in the education of children had low returns since there was a high risk that the child does not survive. Parents therefore did not want to spend resources in educating children who are at a high risk of premature death and therefore might not benefit from that education. With little incentive into the uncertain future of their children, parents instead hoped to maximize the contribution from children to the household by increasing the quantity of children.
Falling child mortality then changed the incentives of the quantity-quality-tradeoff: The reductions in mortality – of children in particular, but over the life course more generally – increased the incentive for parents to invest in the education of their children (‘quality’) and with limited resources to spend on education they are thereby compelled to give up the goal of maximizing the number of children (‘quantity’). Lower child mortality increases the incentives to invest more resources into each child and with limited resources parents therefore chose to have fewer children
Technological and structural change increased the importance of education
Falling mortality of children is of course not the only reason why children in societies with better living conditions are better educated.
Technological change killed many of the low-skill, routine tasks that kept our ancestors busy and meant that workers in a modern economy need a much higher level of education. The economic change driven by technological improvements meant that the returns to education increased and these in turn increased the incentives to invest in the education rather than the number of children, as Becker, Murphy, and Tamura (1990) argue.28 This tilted the quantity-quality tradeoff yet further away from a high number of children.
Conclusion on education
Rich societies moved far away from their past after more than a century of this development. Today children are an economic drain rather rather than an asset. This is essentially because in the modern high-skilled economy we require a massive investment in education over many years that only pays off with a long delay. With respect to child labor and children's education technological change powerfully changed the status of children.
The rising investments of parents in their children do not only have monetary costs. Additionally the rising importance of children could have increased the time and effort of parents in raising their children.
We have no empirical data that could give us an idea of how parents' effort and time investment in their children changed since the 18th or 19th century. And so we have to resort to qualitative assessments here. The social historians Alter and Clark write29 about the view in their profession “Most historians agree that attitudes about children changed by the nineteenth century. Europeans developed a more romantic view of childhood and of domestic life in general. Educational theorists like Froebel and Pestalozzi emphasized that early childhood is a special time in which play stimulates learning. Women were expected to stay at home to create a refuge for children from the competition and conflict of the world of work.“
And while we lack data on how much time parents spent with their children, the much higher working hours in the past make it reasonable to believe that parents had less time with their children and over the course of modernization the parents' time investment increased. For the period since 1960 Gauthier et al (2004)30 document that the time parents spend with children increased substantially in a sample of 16 industrialized countries for which they examined the data.
In popular accounts, it is often argued that in the past and in poor economies parents aim for a large number of children as they will depend on them for support in old age. This argument cannot have played a large role before the onset of the demographic transition as the stagnation of population growth implies that on average only two children will reach the reproductive age.
But even for the time thereafter and more generally there is surprisingly little evidence for this argument given how well it is known. In reviewing the literature Bloom and Luca (2016)31 conclude that “the empirical evidence supporting this hypothesis is surprisingly scarce.”
In a study in Indonesia Cameron and Cobb-Clark (2001)32 find only very limited importance of the transfers of children to their parents and emphasize that instead the elderly are mostly relying on their own labor income even at an old age.
During the period of the demographic transition when mortality was low and fertility high however there is some evidence for the importance of children for old-age support from the US where Sundstrom and David (1988)33 documented the importance of children for the old-age support of the parents before the civil war.
Also Billari and Galasso (2009)34 examine the effect of a pension reform in the 1990s in Italy and find evidence that pensions and children are to some extent substitutes when it comes to support in old age.
Overall however the demand for children arising out of a need for support in old age is likely to be less important than other factors examined here.
The following plot shows the close relation between the income level (measured by GDP per capita) and the total fertility rate. Shown are not just country averages of the fertility rate and income, the visualization is also showing the within-country inequality. Each population is split into 5 quintiles, from the poorest 20% to the richest 20%. Unfortunately there is no data available to do this perfectly and I had to combine income and wealth data to approximate the relationship: The fertility rate by the wealth quintile in each country is plotted against the respective income quintile in each country. The match is imperfect because I am assuming that each household is in the same income and wealth quintile and this is only an approximation of reality.
The reflex of many economists when thinking about the fertility rate is to point to income as the likely determinant. And sure enough, between countries and over time we see that higher incomes are associated with lower fertility. But good things come together – richer countries are also healthier and better educated – and so this correlation between high incomes and low fertility alone is surely not evidence that it is increasing income that is responsible for the decrease in fertility.
In fact we have already explored several third factors. Most obviously, a higher level of education of a population is a factor that contributes to higher prosperity and a lower number of children. And a second set of changes – technological change, lower child labour, and the structural change of the economy – comes along with economic growth and lowers the demand of parents for children, as we have seen.
But still there might also be a direct effect of increasing prosperity on the declining demand for children. Higher incomes make different, more varied lifestyles possible, which might convince prospective parents to have fewer or no children. With respect to the increasing prosperity in Europe over the last century the historians George Alter and Gregory Clark write:35 “New products and new lifestyles in the growing metropolitan societies created by the Industrial Revolution expanded choices. Wealthy families responded by consuming more of these new products and services instead of producing children.”
The correlation that we see in the visualization below is therefore partly driven by a direct link between income and the number of children parents want, but to a large extent also by changes that correlate with prosperity and fertility rates.
The change of fertility are a prime example for changing social norms. In many places around the world the practice of having more than 5, 6, 7, or 8 children, which was the norm for millennia, was replaced by the norm of having 2 children or fewer.
We have explored socio-economic and technological changes that contributed to the declining rate of fertility, but there is also evidence for the importance of cultural changes which had a direct impact on how this norm changed.
The influences on social norms are hard to study as social changes can be rarely separated into causes and consequences. But sometimes there are possibilities to investigate how social norms change. A number of researchers looked at the effect of new media becoming available:
Television is not only providing entertainment, it also exposes the viewers to very different ways of life. For rural, poor, illiterate populations television is the primary way to be informed about the ways of life of people that live very differently. The lives depicted in television shows are in stark contrast to the reality back home. They are often in urban settings and feature female characters with higher education that work outside the household and who have smaller families
The researchers La Ferrara, Chong, and Duryea (2012)36 studied the impact of television on the fertility rate in Brazil. Brazil has their own format of soap operas called ‘telenovelas’ which are hugely popular in the country.
The authors focus on the 8pm telenovelas of Globo – Brazil’s largest network – which are watched by the “vast majority of the Brazilian population” according to the authors. They analyzed 115 novelas aired by Globo between 1965 and 1999 and found that 72% of the main female characters (age 50 and lower) had no children at all, 21% had only one child, so that only 7% had more than one child. These are much fewer children than the average in Brazil over the same time: The fertility rate in 1965 was 5.7 children per woman and fell to 2.4 in 1999.
The researchers were able to identify the effect of the telenovelas as the TV network was only gradually expanded through the country. The entry in each new area allowed the researchers to study the impact of television again and again.
After controlling for socio-economic variables the authors find that the exposure to coverage by Globo is associated with a decrease in the probability of giving birth of 0.5 percentage points, which is 5 percent of the mean. To put this in perspective the authors explain that this size of the effect corresponds to that of an increase of women’s education by 1.6 years.
This effect was strongest for the poorer, less educated women for which the difference to the lifestyles shown on TV is largest.
To illustrate the direct impact of telenovelas on norms the authors also study how the names parents give to their children change and found that the names of the telenovela's main characters increased in popularity.
Similarly, Jensen and Oster (2009)37 studied the effect of the introduction of cable television on women’s status in rural India. They find that getting access to cable TV “leads to approximately a 3.7-percentage-point decrease in the likelihood of pregnancy”, a small but not negligible effect.
And for the US, Kearney and Levine (2015)38 similarly reported in the American Economic Review that the MTV show “16 and pregnant” led to a 4.3 percent reduction in teen births.
All these effects are not large when compared to the very substantial reductions of the number of births per woman over the long run. But it should be kept in mind that these studies look at very isolated aspects of how social norms have changed and as such they already indicate just how important social norms and information about different life styles are parents' decision on the number of children they want.
The role of social norms in lowering the fertility rate might not be only at work in poorer world regions today, but might have actually contributed to the decline of the fertility rate of the countries which first underwent the demographic transition.
The decline of the fertility rate started in France earlier than in other countries. Even countries where the populations were richer and more educated at the time had a higher fertility rate than the French population.39 This scatterplot shows that by 1870 the fertility rate in France was much lower than in countries in which the population was better educated.
The fact that France was the first European region in which women started to have fewer children is shown in the map below. All of the European regions that experienced a 10% decline of the fertility rate were in France.
But the claim is not only that France was special in first experiencing this decline, but also the origin of social norms that lead to declines elsewhere. An indication for the cultural diffusion of social norms originating in France is the empirical finding that language barriers mattered for the timing of the decline of the total fertility rate.
Alter and Clark40 summarize several findings of this research which support this claim: “In Spain, for example, regional differences in fertility followed the boundaries between areas that spoke dialects of Spanish or different languages, such as Basque or Catalan (Leasure, 1963). In France fertility decline was later in Breton-speaking départements. Belgium, divided into Flemish- and French-speaking regions, shows most clearly the importance of linguistic borders. Lesthaeghe (1977) studied seventy Walloon villages matched to neighboring Flemish villages with identical economic conditions. Fertility fell first in the Walloon village in sixty-two of the pairs, and the fertility decline began twenty years earlier on average in the Walloon village (Lesthaeghe, 1977). Flemish couples in the city of Leuven were more likely to have small families if a migrant from Wallonia lived nearby (Van Bavel, 2004).”
Ron J. Lesthaeghe41 has suggested that the cultural change towards lower numbers of children might not only be due to socio-economic changes, but might have been also driven by the spread of Enlightenment ideas and the simultaneous turning away from the pro-natalist doctrines of the Christian religion in Europe. Alter and Clark write “Enlightenment ideas about reason and humankind’s role in nature, as well as opposition to religious authorities, made birth control within marriage ethically and socially acceptable.”
Earliest date of a 10% decrease in fertility in Europe42
Many religious teachings are asking the believers to have a large number of children. The Christian bible for example teaches to “be fruitful and multiply and fill the earth and subdue it”.43
The visualization below shows the children per woman plotted against the share of children that die in the first 5 years of life. Each country here is colored according to the largest religious group in that country. The obvious relationship here is that in countries where more children survive, fertility is lower. Which religion dominates in a country has no clear relation to the fertility level – and even if does have some importance the correlation is much weaker than that with the health of children. Countries with a majority Christian population have fertility rates as high as 6 (DR Congo) and as low as 1.25 (Portugal) children per woman. Across countries fertility rates vary within and not between religions.
And what is true between countries is even more obvious for the change over time. Religious background cannot explain the rapid change in the level of fertility that we saw. Explanations that refer to the cultural background of a population regularly run into this problem that they can hardly explain the very fast socio-economic changes over time. Despite the ascribed values of a particular religion we have seen changes that are at odds with these: In catholic Italy the fertility declined from 2.5 in 1966 to 1.2 at its lowest rate in 1997, and in Muslim Iran the fertility declined from 6.5 children per woman in 1982 to 1.8 in 2005!
As it is so often the case with explanations for social phenomena, the importance of culture is easily overestimated.
While the big differences between countries today and changes over time are not determined by religion it would be wrong to say that religion has no importance for the number of children women have.44 There is evidence that, everything else being equal, religious people have more children, so that religion matters for differences at the same socio-economic level. Still, the differences between religions within the same country are much smaller than the differences between different countries in different socio-economic conditions. We can see how much more living conditions matter than the religion.
And then there have been some small highly religious sub-populations in which religious teaching had a very substantial effect on fertility rates. The highest total fertility rate that I found for any population in human history was for the Hutterites – a group of Christians with origins in Tyrol – where, in the early twentieth century, married Hutterite women in North America had on average 10.4 children.45
Family planning refers to all active efforts to chose actively the number of children a woman or family wants. While the changes discussed before changed the incentives for having a larger or smaller number of children, family planning is focussed on the decision making and implementation of that decision on the personal level.
Family planning involves the use of contraception as well as counseling by experts.
Why family planning is highly important today is to consider the share of pregnancies that are unwanted. The number of pregnancies that are unintended is very high. For the year 2012 – the last year for which we have data – it is estimated46 that 85 million pregnancies were unwanted. These are 40 percent of all 213 million pregnancies in that year.
Of these 85 million pregnancies 50 percent ended in abortion, 13 percent ended in miscarriage, and 38 percent resulted in an unplanned birth. This means that 32 million children are born unplanned every year.
A second way of looking at this is to study the discrepancy between the wanted fertility rate and the actual fertility rate:
- This visualization shows the two measures country by country.
- This visualisation shows the two measures over time.
Access to family planning and safe and confidential access to modern methods of contraception can reduce the number of unintended births.
Women’s empowerment and the increased status of children reduce the number of children that parents want. But a goal of lower fertility is irrelevant if there are no means to achieve it. Methods of contraception give parents the chance to get the actual fertility closer to their desired fertility.
Today there is a range of methods of contraception available which are referred to as "modern methods":47
- oral hormonal pills,
- the intra-uterine device (IUD),
- the male condom,
- the implant (including Norplant),
- vaginal barrier methods,
- female and male sterilization,
- the female condom
- and emergency contraception.
This map shows where women have access to these methods.
Better education of women matters again as research shows that better education can increase the understanding and acceptance of contraceptive methods and the ability to use contraception effectively.48 In a study of the decline of the fertility rate in Bangladesh van Ginneken and Razzaque (2003)49 find that "of the several demand factors (indicators of socio-economic status) studied, women's education has the largest impact on the fertility decline." and more concretely that "a very important role in the fertility decline is played by changes in attitudes towards feasibility and acceptability of birth control."
Martha Bailey (2010)50 studied the timing of legal access to birth control across US states and finds that the availability of the Pill substantially accelerated the post-1960 decline in marital fertility. Based on her analysis Bailey argues 40 percent – or even more – of the total change in the marital fertility rate in the decade between 1955 and 1965 can be attributed to the availability of the Pill. And again, this driver of lower fertility rates is complementing other factors discussed above that contributed to lower fertility rates. Goldin and Katz (2002) argue that the availability of modern contraceptive methods contributed to the increased opportunities of women.51
With the development of low-cost, safe methods of contraception have become more widely available in recent decades. But as the map below shows there is still a substantial unmet need for contraception in many parts of the world. The share of married women of reproductive age who do not want to become pregnant but are not using contraception is higher than 20% in many parts of the world. If one takes into account which substantial impact the availability of modern contraceptive methods had in many parts of the world it seems likely that meeting this "unmet need" is a promising way to further decrease the rate of unintended pregnancies.
In the chart below we see the relationship between fertility rates and the use of contraception. This is shown as the average children per woman versus contraception prevalence (based on the use of any method of contraception). Here we see a negative relationship: in countries where contraception use is low, the number of children per woman is higher. As contraception becomes more widely used, the number of children per woman declines.
We can also see this when we compare the unmet need for contraception with fertility rates: the average children per woman is high where the unmet need for contraception is also high.
Joshi and Schultz (2013)52 analyze the impact of a family planning program that was implemented in Matlab, Bangladesh in 1977. The authors find that villages that were part of the program experienced an additional decline of fertility by about 17%. This effect persisted for at least two decades and additional positive effects on the health and nutritional status of the children.
Family planning advice surely can work as the impressive case of Iran shows, in 1982 Iran had a fertility rate of 6.5 children per woman and only 2 decades later – after this and that advice – the fertility rate was down to 2 children per woman.
A common claim—and one originated by the Chinese Government—is that China’s one-child policy has prevented approximately 400 million Chinese births. The view of many has been that this policy shaped a population age structure that contributed to economic growth (through the effect of the “demographic dividend”) and even contributed to global efforts to address climate change. But was the policy necessary to drive down fertility?
But is it really? The chart shows fertility in China since 1945. The striking decline and rebound of fertility around 1960 is due to the Great Leap Forward famine. But otherwise fertility in China was over 5 and even as high as 7 children per woman in the 1950s and 60s. Then, fertility started to decline – and as we see from the chart this decline started in 1970, long before the introduction of the one-child-policy. China promoted family planning policy in the 1960s and 70s, but the one-child policy was phased in between 1978 and 1980. By the time of the introduction of the one-child-policy, fertility in China had already more than halved. The huge reduction in fertility happened irrespective of the one-child-policy. Until 1980 child mortality – which we saw is an important determinant of fertility – had already halved from 12% (in 1969) to 6%.
In 2013 the researchers Wang Feng, Yong Cai, and Baochang Gu examined what China’s fertility rate would have been in the absence of the one-child policy.53 Using data from countries that had a similar birth rate to China’s in 1970, they compared the trajectories of change in those countries with that of China. The study found that “in other countries without a one-child policy the birth rate also declined, and it declined below the level predicted for China.” Additionally, the researchers estimated what China’s fertility would have been without the one-child policy by using the UN’s 2011 population projection model. The results showed that China’s fertility rate, which was already on a rapid decline in 1970, would have continued to decline after 1980 and by 2010 “fertility would have fallen to its currently observed level.” The continuation of the decline is due to the continuations of improving living conditions in China over this period.
Also shown in the chart is the evolution of fertility in Taiwan. Taiwan – which is claimed by China as part of China – never introduced a one-child-policy. Yet, Taiwan experienced the same decline that China did. From around 7 children per woman to fewer than two. Today the fertility in Taiwan is even lower than in China. In fact the fertility is close to 1 child per woman – just the aim that China had and never reached despite this being the planned outcome of the Chinese government. The point here is that economic and social development is truly important and ultimately what influences women’s decision about how many children they want to have.
Did the one-child-policy work? Fertility in China and Taiwan (1945-2015)
We have already seen that as a country develops – child mortality declines and incomes grow – the fertility declines rapidly.
The demographers Mikko Myrskylä, Hans-Peter Kohler & Francesco Billari studied what happens at very high levels of development. To measure development they relied on the Human Development Index – a measure published by the UN that combines with equal weight indicators of a country's health, material living standards and level of education.
In their study – published in Nature in 200954 – they found "a fundamental change in the well-established negative relationship between fertility and development as the global population entered the twenty-first century."
The visualization below shows their finding. Again, we can see the strong negative association between a country’s level of development and the fertility level. But at very high levels of development—HDI over 0.85 or even 0.9—this association is reversed. While causality cannot be established in this relationship, it is evident that after a given point, higher development is associated with increasing fertility. Not only do the authors show this relationship cross-sectionally, but also over time: after reaching the lowest Total Fertility Rate at HDI values between 0.85 and 0.9, fertility then increases again as countries advanced to the highest development levels.
It is a finding with important consequences. The authors note that this reversal "has the potential to slow the rates of population ageing, thereby ameliorating the social and economic problems that have been associated with the emergence and persistence of very low fertility".
For an interactive version of the above visualization, see here.
The relationship between fertility and HDI through time (1980, 2000, 2014)55
The number of births in a population is measured in different period and cohort measures:
The total fertility rate (TFR)56 – or simply Fertility Rate – is defined as the average number of children that would be born to a woman over her lifetime if:
- The woman were to experience the exact current age-specific fertility rates (ASFRs) through her lifetime, and
- The woman were to survive from birth through the end of her reproductive life.
It is expressed as children per woman.
The definition of the crude birth rate (CBR) – or simply birth rate – is "the number of live births occurring among the population of a given geographical area during a given year, per 1,000 mid-year total population of the given geographical area during the same year".57
The age-specific fertility rate (ASFR) "measures the annual number of births to women of a specified age or age group per 1,000 women in that age group."58
The net reproduction rate (NRR) is defined as "the average number of daughters a hypothetical cohort of women would have at the end of their reproductive period if they were subject during their whole lives to the fertility rates and the mortality rates of a given period. It is expressed as number of daughters per woman".59
Civil registration systems register vital events – births, deaths, fetal deaths, marriages and divorces – for governments. As such they are a key source for vital statistics and fertility measures. As the following map shows the civil registration coverage of births is incomplete in many countries in Africa and Asia.
A second aspect about country level data of fertility to keep in mind is that there can be considerable heterogeneity within countries, which are hidden in the mean fertility which were discussed in this entry. The mean Total Fertility Rate for India in 2010 was 2.8 (UN Data): But this average hides the fact that the fertility in many Southern Indian regions was below 1.5 (which is similar to the mean fertility in many European countries), while the fertility in Northern India was still higher than 5 children per woman (which is as high as the mean of the African countries with the highest fertility).
Total fertility rate map: average births per woman by districts, 201160
United Nations World Population Prospects
- Data: Number of annual births, the Crude Birth Rate (CBR), Total fertility (TFR), Net Reproduction Rate (NRR) among other measures.
- Geographical coverage: The entire world – countries and world regions.
- Time span: From 1950 onwards.
- Available at: Online here
- The League of Nations records are available via Northwestern University and these records include population figures, birth rates, and death rates. They are available by world region and country (a global total is missing however).
Human Fertility Database (HFD) and the associated Human Fertility Data website
- Data: The data is detailed and includes Data fertility rates (by age (ASFR), cohort and period) and mean ages at childbearing.
- Geographical coverage: More than 70 countries
- Time span: Mostly the 2nd half of the 20th century although data for the 1st half of the century is available for some countries
- Available at: Human Fertility Database and Human Fertility data website
The International Historical Statistics (IHS)
- Data: Birth rates
- Geographical coverage: Various countries around the world
- Time span: From 1750 onwards
- Available at: Published in three volumes covering more than 5000 pages.61 At some universities you can access the online version of the books where data tables can be downloaded as ePDFs and Excel files. The online access is here.
- These statistics – originally published under the editorial leadership of Brian Mitchell (since 1983) – are a collection of data sets taken from many primary sources, including both official national and international abstracts dating back to 1750.
- Data: The total fertility rate
- Geographical coverage: Data is available for a great number of countries.
- Time span: Data goes back to 1800
- Available at: The data documentation and the spreadsheets available for download can be found here.
- The documentation includes a discussion of the quality of the available Data fertility.
World Development Indicators (WDI) published by the World Bank.
- Data: Total fertility rate and on the crude birth rate
- Geographical coverage: Entire world, by country and world region
- Time span: Data is available only for the last decades.
- Available at: Data for total fertility rate is here, and data for crude birth rate is here.
The Demographic and Health Surveys (DHS)
- Data: Fertility and related data.
- Geographical coverage: More than 90 countries – here is the list of countries.
- Time span: Data is available from the early 1990s onwards
- Available at: Online here
- For the time preceding the DHS data timeframe the World Fertility Survey is available at Princeton University here.
The UN's World Fertility Report 2012 is online here. The Eurostat website 'Statistics Explained' publishes up-to-date statistical information on fertility here. For a comparison of the latest available data of different data sets, see Wikipedia here.