2025/2026 Fertility job market papers

A summary of new research papers

Job market season is upon us which means that lots of new research is coming out. I am trying to make a list of all job market papers related to fertility and family formation. Let me know if you come across something that I should add or if your own job market paper should be on the list!

This idea comes from Matt Clancy and Lauren Gilbert.

This post contains short summaries of the job market papers on fertility and family formation that I have found so far by searching keywords like “fertility”, “family”, “child”, “birth”, and “marriage” in the paper titles on econ.now.

We have:

• Increases in the price of childcare in the US lead to decreases in fertility

• A month of mandated paid paternity leave in South Korea increased births

• Rising housing costs have contributed to the fertility decline and building more large units would be particularly helpful for fertility

• Parental leave policies increase fertility but decrease children’s human capital

• How and why marriage, cohabitation, and separation dynamics have changed in the US over four decades

• Unplanned C-sections decrease future fertility

“The Price of Parenthood: Childcare Costs and Fertility”

Abigail Dow, Boston University

This paper uses data from the US to get at the causal effect of childcare costs on fertility. The idea is to use changes in childcare regulations as an instrumental variable. For example, if regulations in a state change so that the maximum group size decreases or staff-to-children ratios increase, we would expect the price of childcare to increase. We can then see what the effect of this estimated increase is on birth rates.

Using this methodology, the author finds that a 10% increase in childcare prices is associated with a 5.7% decrease in birth rates for women aged 20 to 44. Given that the mean birth rate is 70 per 1,000 women, this amounts to 4 fewer births per 1,000 women.

The effects on fertility are concentrated among women over 30. The author builds a model to explain this heterogeneity: Older women have higher incomes and therefore outsource more childcare, making them more responsive to childcare prices. They are also more likely to already have a child and higher parity births are more sensitive to childcare costs as well.

“Mandated paternity leave and fertility: Evidence from South Korea”

Tammy Sunju Lee, University of Michigan (joint with Jungmin Lee)

The authors use administrative data from South Korea to study the effects of mandated paternity leave in this very low fertility setting. While South Korea has generous parental leave in principle, take-up of paternity leave remains very low in practice. A large conglomerate implemented a mandated paternity leave of one month in 2017, meaning that the authors can compare outcomes at the firms belonging to the conglomerate with outcomes in similar non-treated firms.

The intervention had a large effect on take-up of paternity leave:

For the control firms (on the right), less than 10% of fathers take any parental leave throughout the study period. In the treated firms (on the left), the share taking leave shoots up after the policy implementation and goes from roughly 2% of fathers to almost 68%.

Event study plots show that the intervention increased the probability of having a baby by about 15%:

They also run a survey of male employees at treated and non-treated firms and find that those in treated firms are more likely to recommend taking paternity leave to a hypothetical co-worker and think that their co-workers are more likely to take leave and more likely to recommend taking leave. Additionally, employees at treated firms are less likely to believe that a hypothetical co-worker would face career penalties from taking leave or that other employees would face an increased workload because of someone taking parental leave.

“Build, baby, build: How housing shapes fertility”

Ben Couillard, University of Toronto

This paper investigates the effects of rising housing costs since 1990 on fertility in the US by combining census tract data with a structural model of housing demand and location choice.

The main result is that if housing costs had stayed at the (inflation-adjusted) 1990 level, then 13 million additional children would have been born in the three following decades, which is roughly 11% of all births during that time. Rising housing costs explain roughly half of the decline in the total fertility rate between the 2000s and 2010s. Additionally, housing costs are responsible for 7 percentage points of the decline in family formation among young people, aged 20-29.

The graph below depicts these findings. The red line shows what happened in the real world during this time and the dashed blue line is what the paper estimates would have happened had housing costs not risen since the 1990s. The dashed purple line is what you would get if you naively applied results from the data without the structural model. This overestimates the effects of increasing housing costs because families select into cheaper locations.

The author then compares two different changes in the housing supply to see what the effects on fertility would be in the model. First, we could build a larger amount of additional small units, decreasing average rents more. Second, we could build a smaller amount of additional large units, decreasing average rents by less but increasing the supply of housing units particularly desired by families. The model shows that the latter option generates 2.3 times more extra births than the former option.

“The distributional effects of paid parental leave policies on fertility and children’s human capital”

Giorgia Conte, Trinity College Dublin

The US is the only high-income country without widespread paid parental leave. This paper investigates the effects of introducing paid leave in this context, by comparing what happened in New Jersey after it introduced 6 weeks of paid leave to what happened in Maryland as the control state.

First of all, the author finds that the introduction of paid leave increases birth rates by 17%:

In order to understand the effects of the policy more closely, the author then builds a theoretical model that incorporates decisions on fertility, labour supply, and parental investment in each child for agents with different levels of education. The model is calibrated using different US data sets and importantly some data on child cognitive development.

The model can then be used to simulate what happens with the adoption of paid parental leave. Total fertility rises by 4.5% which is lower than the effect in the data. One possible explanation is that some of the effect in the data is on the timing of births, i.e. births that would have happened anyway happen earlier, while the model only captures the effect on total fertility. The fertility increases happen mostly in the middle of the income distribution and lead to declines in parental investment per child. Child human capital decreases by 1% in early childhood, 1.3% in middle childhood, and 0.5% in late childhood. This is due to more children being present in the household: If fertility stayed constant, parental leave would increase child human capital.

“Marriage, cohabitation, and separation: A dynamic approach to the second demographic transition”

Sean Elliott, University of Toronto

This paper first establishes three broad facts about household formation in the US from 1980-2023:

1. Marriage rates have declined while cohabitation rates have risen:

2. Marriages and cohabitations have become more stable over time:

3. The patterns change very clearly at two points in time. In the mid-1990s, substitution from marriage to cohabitation accelerated and after 2008, remaining single altogether became increasingly common.

The author then builds a model of relationship formation that can capture match surplus (the value both partners get from being in the relationship), match-formation costs (for example, the effort you have to put in to find a relationship), and dissolution costs (for example, the hassle of finding new apartments when you live together and break up).

When calibrated to the data, the model shows that average match surplus has risen over time which explains why relationships are more stable. This is particularly true for marriages, while for cohabitations, the increased stability comes more from rising dissolution costs. At the same time, match-formation costs have increased which means that fewer relationships are formed in the first place.

The shift from marriage to cohabitation in the mid-1990s can be explained by changes in the relative attractiveness of marriage and cohabitation due to easier paternity establishment in cohabitations and better support for single mothers.

Relationship formation patterns differ by the education level of both partners. This graph summarises changes in the net gain from relationship formation compared to 1983:

Negative values indicate that a given relationship is less likely to form compared to 40 years ago. While couples where both partners or the woman are highly educated have not experienced big changes over time, those where both partners or the woman have low education are much less likely to enter into cohabitation or marriage.

“After the cut: Cesarean delivery and subsequent fertility”

Haley Wilbert, University of Notre Dame (joint with Carson Homme)

The authors have detailed individual-level medical data for members of the US army and their dependents. They use this data to estimate the effects of C-sections that are likely planned and those that are likely unplanned.

In order to get at the causal effect of an unplanned C-section, they use the fact that patients are quasi-randomly assigned to a managing doctor based on when they arrive at the hospital and who is available at that point. Different doctors are more or less likely to perform C-sections, as the graph below shows:

The idea is to use this variation between doctors to create an instrumental variable that predicts whether someone has an unplanned C-section and to then estimate the effects of this unplanned C-section. You can think of this as comparing women giving birth who are very similar in all respects but some of them have an unplanned C-section because they got assigned a doctor who is more likely to perform C-sections.

Using this strategy, they find that the probability of a subsequent birth within 42 months is reduced by 28% (8.7 percentage points) and the probability of a subsequent birth within 48 months is reduced by 34% (11.9 percentage points) due to having an unplanned C-section.

The paper also estimates the effects of likely planned C-sections by using the fact that most babies presenting in the breech position (bottom down instead of head down) are delivered via planned C-sections. They find no statistically significant effects of planned C-sections on future fertility. This indicates that it is not the procedure itself that seems to be affecting future fertility but rather the fact that an unplanned C-section is presumably much more stressful and a more negative birth experience overall.

“Understanding variation in Cesarean section use: Supply-side drivers and maternal health effects”

Helen Kissel, Stanford University (joint with Helena Roy)

This paper is less focused on fertility outcomes but also has a short section on how unplanned C-sections influence subsequent fertility. The authors of this paper use nationwide Medicaid data and focus on women in a very low-risk group: Those who have never given birth before, have carried the pregnancy to term, are pregnant with a singleton, and the baby is in vertex position (head down) prior to birth. They then use the same strategy of exploiting variation in the likelihood to perform C-sections between randomly assigned doctors.

They also find that there is lots of variation between hospitals and between doctors in the likelihood of C-sections:

Additionally, they show that C-sections are not always well targeted. This graph shows the distribution of vaginal births and C-sections by underlying C-section risk, i.e. based on clinical indicators for C-sections:

While most C-sections happen in patients where they are appropriate, there is also a considerable share happening in patients with very low risk scores. This further validates the interpretation of unnecessary C-sections in some patients happening only because of being assigned to a doctor with a higher propensity of performing a C-section.

This paper finds a 28% reduction in the probability of having a second birth in their study period as a result of an unplanned C-section. I think it’s really cool to have these two papers side by side and see that they both come to a very similar conclusion on the magnitude of the effect! This is even though they use different data sets covering different populations and make different decisions as to what patient subset to focus on.