2. Literature

There is a strand of literature exploring the relationship between fertility and (female) labor supply, that is, labor market participation. In broad terms, given the wide and long-standing interest in the link between childbearing and labor supply, hundreds of empirical studies have estimated this relationship.

The majority of these studies, for many years, found a negative association between fertility and female labor supply (i.e., Angrist & Evans, Reference Angrist and Evans1998; Del Boca & Sauer, Reference Del Boca and Sauer2009; Herrarte et al., Reference Herrarte, Moral-Carcedo and Sáez2012). The negative sign of this relationship was due to both economic and demographic issues. Having fewer children subsequently increased female labor market attachment, a trend that began in the postwar period (i.e., Blau & Grossberg, Reference Blau and Grossberg1992).

Some works found that decisions about fertility and labor market participation are interrelated and associated with both individual and household characteristics, especially the education of women (Hazan & Zoabi, Reference Hazan and Zoabi2015), the employment and earnings of all household members (Bar et al., Reference Bar, Hazan, Leukhina, Weiss and Zoabi2018), the availability of childcare services (Herrarte et al., Reference Herrarte, Moral-Carcedo and Sáez2012) and, more broadly, the role of institutions. As suggested by Del Boca (Reference Del Boca2002), for instance, several institutional rigidities limiting the availability of both part-time jobs and affordable childcare increased the costs of working for mothers, thereby explaining the low fertility and low labor market participation rates of women in the past. Moreover, the opportunity costs associated with childbirth, including the increasing labor market opportunities for women, may induce them to forgo having children to preserve their employment (Garibaldi & Wasmer, Reference Garibaldi and Wasmer2004; Bloom et al., Reference Bloom, Kuhn and Prettner2024).

However, many countries, including European and other developed countries, have experienced significant changes in female labor market participation and fertility over the last 40 years. Many countries have seen an increase in female labor market participation, especially among women with children (Del Boca & Sauer, Reference Del Boca and Sauer2009). As a result, the relationship between female labor force participation and fertility became even more complex over time (Bloom et al., Reference Bloom, Kuhn and Prettner2024).

In the last 20 years, empirical evidence on high-income countries has shown a weakening or even a reversal of the negative relationship (see, for instance, Michaud & Tatsiramos, Reference Michaud and Tatsiramos2011; Doepke, Reference Doepke, Hannusch, Kindermann, Tertilt, Lundberg and Voena2023). Researchers have highlighted various competing mechanisms contributing to the explanation of the reversed relationship.

These include the expansion of public education, the reduction of the opportunity cost of women’s time due to the marketization of childcare in high-income countries (e.g. Doepke et al., Reference Doepke, Hannusch, Kindermann, Tertilt, Lundberg and Voena2023), family policies – that is, benefit schemes (Gonzalez & Trommlerova, Reference González and Trommlerová2023) – and the availability of public childcare (Bauernschuster et al., Reference Bauernschuster, Hener and Rainer2016; Olivetti & Petrongolo, Reference Olivetti and Petrongolo2017), a stronger involvement of fathers in childcare (Doepke & Kindermann, Reference Doepke and Kindermann2019; Fanelli & Profeta, Reference Fanelli and Profeta2021), more favorable social norms that make more acceptable for women with children to work (Kleven et al., Reference Kleven, Landais, Posch, Steinhauer and Zweimüller2019), and the role of institutions and public policies (Del Boca & Sauer, Reference Del Boca and Sauer2009; Del Bono et al., Reference Del Bono, Weber and Winter-Ebmer2012, Reference Del Bono, Weber and Winter-Ebmer2015). Rindfuss et al. (Reference Rindfuss, Guzzo and Philip Morgan2003) and Brewster and Rindfuss (Reference Brewster and Rindfuss2000), for instance, argued that changes in the institutional context, such as changing social norms toward working mothers, evolving family policies (such as cash benefits and increasing childcare availability), all reduced the incompatibility between child-rearing and female employment (Kögel, Reference Kögel2004).

There is a strand of literature exploring the role of family/children policies (such as spending on cash and in-kind benefits) on the relationship between childbearing and female employment. However, the majority of studies available investigate the direct effect of such interventions on fertility, and not on the relationship between fertility and female employment. We try to fill this gap in the literature.

Gauthier and Hatzius (Reference Gauthier and Hatzius1997) explore the effect of family-friendly policies, such as cash and maternity benefits, on fertility in 22 industrialized countries from 1970 to 1990. Their findings suggest that cash benefits in the form of family allowances are positively related to fertility. Maternity leave, instead, did not appear to be significantly related to fertility. Bokun (Reference Bokun2024) studied the effect of a cash transfer program in Poland named Family 500+, introduced in 2015 to increase fertility and alleviate child poverty. D’Albis et al. (Reference D’Albis, Gobbi and Greulich2017) that middle-income women are the most affected by the lack of access to formal and accessible childcare.

Nygård et al. (Reference Nygård, Lindberg, Nyqvist and Härtull2019) suggest that spending on family benefits in kind (such as childcare) primarily aims to facilitate parental employment (Ferrarini, Reference Ferrarini2006), and they are often coupled with different cash transfers that provide some income protection for households (Esping-Andersen, Reference Esping-Andersen2005). Dinale (Reference Dinale2024) on OECD countries suggests that in-kind benefit provision has the most positive association with female employment and fertility, thereby confirming previous findings (Thévenon, Reference Thévenon2011; Brinton & Lee, Reference Brinton and Lee2016).

Finally, a few studies explored the effect of specific policy/institutional interventions by country aimed at increasing fertility as well as enhancing female employment. For Spain, Sánchez-Mangas and Sánchez-Marcos (Reference Sánchez-Mangas and Sánchez-Marcos2008) investigate the effects of a public policy introduced in 2003 aimed at reconciling family and work; for the UK, Francesconi and van der Klaauw (Reference Francesconi and van der Klaauw2007) explore the effect of “in-work” benefits to single mothers on female employment rates.

3. Employment, childbirth, and public policies

In this Section, we describe the evolution of the female employment rate for women aged 20–44Footnote ² and total fertility ratesFootnote ³ from 1995 to 2021, as reported in Figs. 1 and 2 within welfare systems and family/children policies in the countries analyzed

Figure 1a.

Evolution of the female employment rate.

Source: Authors’ calculation on Eurostat data.

Figure 1a. Long description

The line graph illustrates the evolution of the female employment rate for women aged 20 to 44 from 1995 to 2021. The x-axis represents the years from 1995 to 2021, while the y-axis represents the employment rate in percentage, ranging from 40 to 85. The graph includes multiple colored lines representing different countries and regions: EU-27, Denmark, the United Kingdom, the Netherlands, France, Italy, Spain, and Poland. Each line shows the trend of the female employment rate over the years. Denmark and the Netherlands consistently have the highest employment rates, while Italy and Poland have the lowest. The EU-27 average shows a steady increase over the years. All values are approximated.

Figure 1b.

Evolution of childless women and mothers’ employment rates.

Source: Authors’ calculation on Eurostat data.

Figure 1b. Long description

The image contains four line graphs showing the employment rates of childless women and mothers with one, two, or three or more children from 2006 to 2021. Each graph represents different categories of women: childless women, women with one child, women with two children, and women with three or more children. The x-axis represents the years from 2006 to 2021, and the y-axis represents the employment rate percentage ranging from 35 to 90. The graphs include data for the EU-27 and several individual countries: Denmark (DK), the United Kingdom (UK), the Netherlands (NL), France (FR), Italy (IT), Spain (ES), and Poland (PL). Each country is represented by a different colored line. The employment rates for childless women show a general trend of stability with slight fluctuations. For women with one child, the employment rates are relatively high and stable. Women with two children show a gradual increase in employment rates over the years. For women with three or more children, the employment rates are lower but show a slight upward trend. The data indicates variations in employment rates across different countries and family structures. All values are approximated.

Figure 2.

Evolution of the total fertility rate.

Source: Authors’ calculation on Eurostat data.

Figure 2. Long description

The line graph illustrates the evolution of the total fertility rate in various countries from 1995 to 2021. The x-axis represents the years, ranging from 1995 to 2021, while the y-axis represents the total fertility rate, ranging from 1.1 to 2.1. The graph includes multiple lines, each representing a different country or region: EU-27, Denmark, United Kingdom, Netherlands, France, Italy, Spain, and Poland. Each line shows the trend of the total fertility rate over the specified period. The lines for France and Denmark are at the top, indicating higher fertility rates, while the lines for Italy and Spain are at the bottom, indicating lower fertility rates. The other countries fall in between, with varying trends over the years. All values are approximated.

Notably, there are structural differences in EU countries both in the rate and in the evolution of the female employment rate. Among Mediter ranean countries, female employment rates were very similar in 2000 (49.6% in Italy and 51.9% in Spain), but the gap widened considerably over the following two decades. Italy recorded only a modest increase, whereas Spain saw a substantial rise (Fig. 1a). The rise in Spain’s female employment rate was driven by higher female educational attainment and the expansion of family–work reconciliation policies, which facilitated women’s access to skilled jobs and reduced barriers to maternal employment (Ibáñez, Reference Ibáñez2017; Duarte et al., Reference Duarte, Sarasa and Serrano2019). Conversely, Italy faces a “low-participation trap” due to high marginal taxes on second earners and chronic shortage of affordable childcare, which remains lower than in Spain (Reljic & Zezza, Reference Reljic and Zezza2025). In contrast, Denmark’s female employment rate has stayed high compared with other EU countries, though it declined slightly from 78% in 2000 to 75.9% in 2021, likely due to structural labor market factors such as widespread part-time work among women balancing childcare and family responsibilities (OECD, 2024; Eurofound, 2025). Female employment is also relatively high in the Netherlands and less affected by the recession (75.2% in 2000 to 79.4% in 2020). Overall, EU female employment followed a common trend: rising in the early 2000s, dropping during the Great Recession and austerity, then recovering in the late 2010s (Fig. 1).

Figure 1b reports employment rates for childless women and for mothers with one, two, and three or more children aged 20–49, documenting the motherhood employment penalty across the countries studied. Overall, childbirth is associated with lower female employment. Denmark displays a distinctive pattern, with employment increasing after the first and second child before declining among mothers with three or more children, reflecting extensive family-friendly policies (Pylkkänen & Smith, Reference Pylkkänen and Smith2004; Cebrián, Reference Cebrián, Davia, Legazpe and Moreno2019). By contrast, in Southern European countries such as Spain and Italy, employment drops sharply after the first child and continues to decline with subsequent births, reflecting more limited childcare provision and less flexible work arrangements (De Quinto et al., Reference De Quinto, Hospido and Sanz2021; Alderotti, Reference Alderotti2022). Across all countries studied, the transition to three or more children is consistently associated with reduced maternal employment, as cumulative care responsibilities outweigh policy support (Kleven et al., Reference Kleven, Landais and Leite-Mariante2025).

The fertility rate followed a similar pattern. After years of decline, fertility rates were increasing in late ‘90s in EU countries (later in the UK and Poland) until the Great Recession. Thereafter, fertility rates dropped, and only 10 years later, the fall stopped (Fig. 2).

Immigration contributed to aggregate fertility in Europe, especially in Southern countries such as Spain, where foreign-born women had higher fertility than natives in the late 1990s and early 2000s, partly offsetting the decline (Coleman, Reference Coleman2006; Sobotka, Reference Sobotka2008; Castro-Martín & Rosero-Bixby, Reference Castro-Martín and Rosero-Bixby2011). However, while immigration influenced fertility dynamics in specific periods, it does not explain persistently low fertility across Europe (Bloom et al., Reference Bloom, Kuhn and Prettner2024). The decline in fertility and female employment during the Great Recession reflects broader perceptions of uncertainty (Matysiak et al., Reference Matysiak, Sobotka and Vignoli2021), as economic downturns often reduce fertility and encourage postponement (Cherlin et al., Reference Cherlin, Erin Cumberworth, Morgan and Wimer2013). Such shocks affect fertility in both the short- and medium-long run, with rising unemployment historically linked to declines lagged by 1 to 2 years (Adserà, Reference Adserà2011; Neels et al., Reference Neels, Theunynck and Wood2013). Evidence after 2008 confirms these pro-cyclical patterns, with fertility rates declining or stabilizing across most European countries, particularly in Southern and parts of Central and Eastern Europe (Lanzieri, Reference Lanzieri2013).

France maintained relatively high fertility, supported by state policies for working mothers, childcare, and child transfers, while female employment rose from 64.8% in 2000 to 69.3% in 2020, only marginally affected by the Great Recession (Thévenon & Gauthier, Reference Thévenon and Gauthier2011). Poland saw rising female employment (59.3% in 2000 to 67.1% in 2008) and a modest fertility increase before the recession, but fertility remained low due to demographics, motherhood costs, and limited labor participation (Suwada, Reference Suwada2021). In the Nordic countries, fertility declined during the Great Recession despite mild economic shocks, though Denmark experienced pre-recession and 2016–2017 fertility rises, remaining above Poland (Comolli et al., Reference Comolli, Neyer, Andersson, Dommermuth, Fallesen, Jalovaara, Jónsson, Kolk and Lappegård2019; Fig. 2). These trends unfolded within fragmented welfare systems and varying levels of family policy expenditure. For the family/children function, social (protection) benefits provide support in cash or in kind in connection with the costs of pregnancy, childbirth, and adoption, bringing up children, and caring for other family members (for details, see Eurostat, 2008). For the definitions of cash benefits and benefits in kind for the family/children function, see Table 1.

Table 1.

Classification of benefits in the Family/Children function and country-specific information

Table 1. Long description

The table presents a comparison of various types of benefits related to family and children support across different countries. It is structured into three main columns: Type of benefit, Description, and Country-specific information. The table has multiple rows detailing different benefits such as cash benefits, including income maintenance benefits, birth grants, parental leave benefits, family or child allowances, and other cash benefits. It also covers benefits in kind, such as child day care, accommodation, home help, and other benefits in kind. Each type of benefit is described, and country-specific information is provided for Denmark, the United Kingdom, the Netherlands, France, Italy, Spain, and Poland. The table highlights the variations in family support policies and expenditures across these countries.

Source: ESSPROS, Eurostat (2008) and Social protection statistics, Eurostat (2023).

Notably, the composition of the overall expenditure on family/children benefits, according to Eurostat data, differs importantly across the countries explored.Footnote ⁴ Figure 3 shows the composition and evolution of the expenditure for family policies, namely cash and in-kind benefits. Overall, we note an increase in social expenditure in all countries from 2004 to 2020 and a prevalence of the cash benefits component. For the year 2020, we see that in France, Italy, the Netherlands, Poland, and the UK, periodic cash benefits are the most important component of the total expenditure on family/children benefits. Denmark and Spain, instead, see a prevalence of the in-kind component.

Figure 3.

Composition and evolution of expenditure for family policies.

Source: Authors’ calculation on Eurostat data. Note: PPS euros per inhabitant.

Figure 3. Long description

The bar graph compares cash and in-kind expenditures for family policies across various countries and years. The x-axis lists countries and years, including Denmark 2004, Denmark 2020, United Kingdom 2004, United Kingdom 2018, Netherlands 2004, Netherlands 2020, France 2004, France 2020, Italy 2004, Italy 2020, Spain 2004, Spain 2020, Poland 2004, and Poland 2020. The y-axis measures expenditure levels ranging from 0 to 1400. Each country-year pair has two bars: one for cash expenditures in black and one for in-kind expenditures in gray. Notable trends include significant increases in both cash and in-kind expenditures from 2004 to 2020 for most countries, with the United Kingdom 2020 showing the highest in-kind expenditure. All values are approximated.

Nygård et al. (Reference Nygård, Lindberg, Nyqvist and Härtull2019) show a stable growth in GDP-related and real levels of spending on both cash benefits and in-kind benefits. From Fig. 3, we see that all the countries analyzed show an increase in spending on cash benefits, in some cases even exceeding the increase in in-kind benefits (e.g., the Netherlands, Spain, and Italy). However, spending on benefits in kind was more efficient for alleviating child poverty than spending on cash benefits (e.g., Esping-Andersen, Reference Esping-Andersen2005, and Ferrarini, Reference Ferrarini2006).

More specifically, according to the family policy regime literature, countries in Europe used a mix of “transfer-based” and “service-based” policies (i.e., Korpi, Reference Korpi2000; Ferrarini, Reference Ferrarini2006). Countries of the Nordic regime, such as Denmark, have generally invested heavily in in-kind benefits, but also in transfers to families. These countries display relatively higher levels of female labor force participation. Countries in the Continental regime, such as France, have traditionally used relatively generous cash benefits and tax credits, whereas spending on in-kind benefits has been low (Mahon, Reference Mahon2002; Hemerijck, Reference Hemerijck2013). In the Anglo-Saxon market-based family regime, for instance, in the UK, governments generally spend less on direct cash benefits to families, but more on tax subsidies for market-provided childcare services. This has made mothers more inclined to part-time rather than full-time employment (Esping-Andersen, Reference Esping-Andersen2009). In Mediterranean countries, family cash benefits have generally been modest, and in-kind benefits almost non-existent, with low levels of female employment (Nygård et al., Reference Nygård, Lindberg, Nyqvist and Härtull2019). However, in some countries, such as Spain, there was a shift toward de-familizing family policies, including higher childcare provision and increased family benefits, in the 1990s and early 2000s (Ahrendt et al., Reference Ahrendt, Blum and Crepaldi2015). By contrast, Eastern European countries (i.e., Poland) share a legacy of broad coverage of public social services from the communist era. They have also used a transfer-based strategy, although the generosity level has been lower than in the Nordic and Continental countries (Saxonberg & Szelewa, Reference Saxonberg and Szelewa2007).

On the whole, the importance of in-kind benefits seems to have grown since the 1990s, largely driven by the expansion of childcare/day-care provision. This, together with the austerity era, as suggested by, among others, Farnsworth and Irving (Reference Farnsworth, Irving, Farnsworth and Irving2015), made in-kind benefits even more relevant. There is a substantial body of literature suggesting that in-kind benefits, particularly childcare services, positively influence female labor force participation (Del Boca et al., Reference Del Boca, Pasqua and Pronzato2005; Bárcena-Martin et al., Reference Bárcena-Martin, Lacomba-Arias, Moro-Egido and Pérez-Moreno2014; Ahrendt et al., Reference Ahrendt, Blum and Crepaldi2015; Chzhen Reference Chzhen2017).

4. Econometric approach

In line with the life cycle approach proposed by Nakamura and Nakamura (Reference Nakamura and Nakamura1985), we adopt a dynamic approach to model employment and fertility choices. We tackle their dynamic interrelation and the existence of endogeneity issues (e.g., Del Boca et al., Reference Del Boca, Pasqua and Pronzato2005) by incorporating state dependence, endogenous initial conditions, and potential feedback effects from employment to future fertility decisions (e.g., Carrasco, Reference Carrasco2001, Michaud & Tatsiramos, Reference Michaud and Tatsiramos2011). Because both employment and childbirth are binary variables, our specification results in a dynamic bivariate probit model, which we apply to each analyzed country.

Let e _it represent the employment status of individual i = 1…n at time t = 1…T, the benchmark specification of female employment reads:

(1) ${e_{it}} = 1\left\{ {\gamma {e_{it - 1}} + \beta {b_{it}} + \delta {b_{it - 1}} + \omega {x_{it}} + \vartheta {z_i} + {a_i} + {u_{it}} \gt 0} \right\}$

where 1 is a binary indicator function, e _it−1 is the lagged employment status, b _it is a dummy variable indicating whether a child was born in the current (surveyed) year or not, b _it−1 is the lagged childbirth dummy variable, while x _it and z _i are vectors of strictly exogenous time-variant and time-invariant (respectively) individual and household characteristics.Footnote ⁵ γ is the state dependence parameter, β is the parameter of interest describing the impact of childbirth on female employment, δ measures the impact of past childbirth, while ω, ϑ, and φ are sets of parameters to be estimated. Finally, a _i and u _it represent the unobserved time-invariant individual effect and the idiosyncratic error term; we assume that these are both normally distributed and that u _it is not serially correlated.

The childbirth equation reads as:

(2) ${b_{it}} = 1\left\{ {\alpha {e_{it - 1}} + k{b_{it - 1}} + \tau {x_{it}} + \varphi {z_i} + \lambda {m_{it}} + {h_i} + {\varepsilon _{it}} \gt 0} \right\}$

where m _it is a variable indicating if the first-born is male or not and introduced for identification purposes, h _i is the random effects term, and ε_it is an idiosyncratic error we assume to be normally distributed. α, κ, τ, φ, and λ are parameters to be estimated.

The so-called initial conditions problem (Heckman, Reference Heckman, Manski and McFadden1981), due to the potential misalignment between the start of the observed data and the beginning of the analyzed process, is addressed using an alternative conditional maximum likelihood (CML) estimator proposed by Wooldridge (Reference Wooldridge2005).Footnote ⁶ This requires specifying the conditional densities of the unobserved effects via the following auxiliary models:

(3) ${a_i} = {\theta _0} + {\theta _1}{e_{i1}} + {\theta _2}{b_{i1}} + {\theta _3}{\bar x_i} + {\mu _i}$

(4) ${h_i} = {\pi _0} \pm {\pi _1}{e_{i1}} + {\pi _2}{b_{i1}} + {\pi _3}{\bar x_i} + {\varepsilon _i}$

Where e _i1 is the initial employment status, b _i1 is the value of the childbirth dummy variable at time 1, ${\bar x_i}$ is a set of time-averaged time-variant control variables calculated from periods 2 to T.Footnote ⁷ Finally, θ _k and π _k are parameters to be estimated.

Finally, because unobservable factors such as norms, employer discrimination, and sectoral shocks may affect both equations,Footnote ⁸ we model the correlation between unobserved heterogeneity terms to reduce the potential risk of bias when estimating the effect of childbirth on employment. Their association is captured by the correlation term $\rho = {{corr}}\left( {{\mu _i},{\varepsilon _i}} \right) $ .Footnote ⁹

Based on the benchmark specification, we run two augmented models. The former estimates the change in childbirth effects on female employment over time, the latter analyses how family policies shape childbirth effects.

In the former specification, the childbirth indicator in the employment equation and the lagged employment indicator in the childbirth equation interact with a categorical variable identifying four distinct periods: the pre-Great Recession period (2003–2007), the Great Recession period (2008–2009), the Austerity period (2010–2014), and the post-Great Recession period (2015–2020). The two equations are then modified as follows

(5) ${e_{it}} = 1\left\{ {\gamma {e_{it - 1}} + \beta \left( {{b_{it}}*perio{d_i}} \right) + \delta {b_{it - 1}} + \omega {x_{it}} + \vartheta {z_i} + {a_i} + {u_{it}} \gt 0} \right\}$

(6) ${b_{it}} = 1\left\{ {\alpha \left( {{e_{it - 1}}\;*perio{d_i}} \right) + k{b_{it - 1}} + \tau {x_{it}} + \varphi {z_i} + \lambda {m_{it}} + {h_i} + {\varepsilon _{it}} \gt 0} \right\}$

In the latter specification, the childbirth indicator in the employment equation interacts with several family policy indicators, which are included in the childbirth equation as controls. We consider the total expenditure for family policies, expenditure for cash and in-kind benefits, and expenditure for specific measures, including family allowances, income maintenance, birth grant, child day-care, and accommodation. In a supplementary exercise, we consider labor market conditions and characteristics, namely female part-time employment, female fixed-term employment, and female unemployment.

In this case, we pool information for the seven countries (c subscript) to exploit the geographical variations. The related equations read:

(7) ${e_{itc}} = 1\left\{ {\gamma {e_{itc - 1}} + \beta \left( {{b_{itc}}*family\;polic{y_{tc}}} \right) + \delta {b_{itc - 1}} + \omega {x_{itc}} + \vartheta {z_{ic}} + {a_{ic}} + {u_{itc}} \gt 0} \right\}$

(8) ${b_{itc}} = 1\left\{ {\alpha {e_{itc - 1}} + k{b_{it - 1}} + \tau {x_{itc}} + \varphi {z_{ic}} + \lambda {m_{itc}} + \varsigma \;family\;polic{y_{tc}}\; + \;{h_{ic}} + {\varepsilon _{itc}} \gt 0} \right\}$

Finally, we emphasize that although the estimated coefficients indicate the sign of the relationship, they are not suitable for determining the magnitude of the impact between the outcome and explanatory variables. Therefore, we compute and report the average marginal effects (AMEs).

5. Data

Our data are from the EU-SILC survey, which is conducted in most European Union countries by the relevant national institutes of statistics using harmonized definitions and survey methodologies. The topics covered by the survey include living conditions, income, social exclusion, housing, work, demography, and education.

We explore EU-SILC longitudinal data files covering the period from 2003 to 2020. Since each longitudinal data file in the EU-SILC survey covers 4 years, we decided to use multiple data files to cover a longer period.Footnote ¹⁰ For each subperiod, that is, 4-year period, we consider the sample of individuals/households interviewed in at least three of the four successive waves. We select data for seven European Countries, that is, Denmark, the UK,Footnote ¹¹ the Netherlands, France, Italy, Spain, and Poland. We focus on women of childbearing age, defined as those between 18 and 45 years old, that is, fertile females. Our unit of analysis is therefore the fertile female, and our sample includes both mothers and childless women. This selection results in a total of 327,377 observations for the period investigated. We model both employment and childbirth (benchmark analysis), which are our variables of interest. For the employment equation, we estimate our model using a dummy variable for being employed (employee, excluding the self-employed).

Regarding childbirth/fertility, in line with existing literature (see, for instance, Barbieri & Bozzon, Reference Barbieri and Bozzon2016; Mussida & Sciulli, Reference Mussida and Sciulli2023), childbirth is defined as a new birth in the current (surveyed) period.

Table 2 reports the weighted summary statistics by country for both dependent variables and covariates used in the econometric analysis. The dependent variables in our investigation are female employment status (employed or not) and childbirth for our sample of women aged between 18 and 45. We now briefly describe the covariates included in each equation, keeping in mind that due to the model used, we also include lagged employment status, lagged childbirth dummies, initial employment condition, initial childbirth, and the average of time-varying covariates (for details, see Section 4).

Table 2.

Weighted descriptive statistics

Table 2. Long description

The table presents weighted descriptive statistics for employment and childbirth across Denmark, the UK, the Netherlands, France, Italy, and Poland. It includes dependent variables such as employment rates and childbirth percentages, as well as covariates like the age and education level of females in childbearing age, presence of children, number of persons with disabilities, homeownership, employment status of husbands, presence of other employed members, and the presence of a boy as the first born. The table has 18 rows and 7 columns, with each row representing a different variable and each column representing a different country. Notable trends include higher employment rates in Denmark and the UK compared to Italy and Poland, and varying childbirth rates across the countries.

Note: Given the composition of the sample, including mothers (and childless women), the indicators “presence of children aged 2–5” and “presence of children aged 6–15” do not sum to 100%, as some households include more than one child.

Source: Authors’ elaborations on EU-SILC data. Note: Mean (percentages). We also control for regions, where available.

For both equations, we control for certain characteristics of the fertile femaleFootnote ¹² as well as household characteristics. The former includes age (splitting into age ranges within the 18–45 years considered), education, and consensual union (whether legal or not). Household characteristics include the presence of children of different age ranges (2–5 and 6–15 years),Footnote ¹³ the number of disabled individuals in the household, home ownership, the employment status of the husband (either employee or self-employed versus not employed), and the presence of other employed household members (other than the husband). We also control the geographical area of residence, where available, for the entire period under investigation.Footnote ¹⁴

Additionally, for identification purposes in the childbirth equation, we include a dummy variable indicating the presence of a first-born boy (male child) in the household. According to the literature (e.g., Ichino et al., Reference Ichino, Lindström and Viviano2014), this variable is expected to increase the likelihood that women will have additional children and would be independent from employment probabilities. Finally, since we analyze data over a long period, we also include yearly dummy variables.

The analyses of the evolution of childbirth effects and the role of family policies required additional variables aimed at identifying specific economic periods (i.e., a categorical variable distinguishing among pre-Great Recession, Great Recession, Austerity, and post-Great Recession periods), and characterizing family policies. For the latter, we employ family policy indicators of the social expenditure on family/children function gathered by Eurostat.Footnote ¹⁵ As for the social expenditure on the family/children function, since this indicator is somewhat associated with fertility rates and the share of children in the population, we consider the expenditure for each resident. The expenditure is expressed in purchasing power standard (PPS), which eliminates the price level differences between countries (measured in PPS per inhabitant). ^{Footnote 16} Table A1 in the Appendix shows the figures for the first and last year available for each policy in each country.

In detail, we merge our data with the indicators provided by Eurostat by obtaining a pooled dataset for all seven countries explored. We consider the role of the overall expenditure on family/children functions and its components of cash and in-kind benefits. We then also investigate specific cash and in-kind benefits, namely family allowances, income maintenance, birth grant, child day-care, and accommodation.