Introduction

The research on parental leave has shown that some countries have been more successful than others in promoting fathers’ use of leave. In all comparative studies on the effects of parental leave design on uptake, some Scandinavian countries emerge as having implemented the most successful reforms to increase the uptake of leave by fathers and to narrow the gender gap in its use. The effectiveness of parental leave reforms on male uptake is explained as the outcome of two main design features: non-transferability, i.e. take the leave-related cash benefit or lose it, and a high income replacement (Castro-García and Pazos-Morán 2016; Dearing 2016; Eurofound 2019; Margolis et al. 2019; Ray et al. 2010). In 2007, the introduction in Spain of a 2-week non-transferable paternity leave with 100% wage replacement—1 year before the start of the economic downturn—was a successful provision throughout the crisis years, comparable to the introduction of the first Swedish daddy month in 1995, also initiated at the onset of the Swedish recession in the 1990s. Both countries’ stories suggest that the element of non-transferable and highly paid design is sufficiently strong that it even works during periods of economic recession.

Yet, researchers show that the design of leave policy reforms often follows the idea of freedom of choice for the couple (transferable leave), and researchers call attention to policy reformers’ resistance to adopting a gender equality perspective, which favours individual, non-transferable and well-paid leave (Morel 2007; Morgan 2009). Rightly, political support for this simple design is not assured, as shown by European Council approval in June 2019 of a new European Union Directive on work–life balance for parents and carers. This Directive grants only 2 months of non-transferable leave instead of the 4 months proposed initially, and countries are free to set the wage replacement level instead of the initial proposal of a fixed level set at the level of sickness benefit payment (around 75% of the wage). To what extent are non-transferability and high wage replacement important in promoting male uptake of parental leave?

This article focuses on the effect of the economic recovery (2014–2016) and the extension of leave (2017–2018) on the use of the paternity leave by different types of fathers in Spain. Can changes in the economic and political context influence individual and couple-level barriers in the use of paternity leave? How far does the economic context and the increased social legitimation of paternal care through an extension of parental leave affect fathers’ opportunity costs in using it? The aim of this study is to understand which fathers use paternity leave less, and to test if the return to economic growth in 2014 or the extension of the Spanish paternity leave allowance in 2017 reduced any of the differences in its uptake across different groups of fathers. The study of patterns of paternity leave uptake across time, especially comparing the years of economic crisis with the economic recovery, allows us to assess possible contextual effects on leave use. In addition, the extension of paternity leave in 2017, during the fourth year of economic recovery, constitutes a good natural setting for studying, separately, the impact of an important institutional change. These two contextual changes may ease and legitimate, both within the family and in the workplace, fathers’ decisions and bargaining over this individual and statutory right to take care of the mother and the child(ren).

Pooling 44 quarters of the national Spanish Labor Force Survey enables us to estimate the probability of the use of leave according to fathers’ socio-economic characteristics and to see how different economic periods influence the likelihood of take-up. To our knowledge, this is the first study that explores for Spain how changing economic and institutional contexts influence social patterns in the use of paternity leave. Research has focused on gender and social differences in the uptake of leave, but there is a gap in the literature on how contextual changes, both economic and legislative, affect men’s opportunity costs and social legitimation.

Literature Review

What Drives the Uptake Rates Among Fathers? Evidence from Other Countries

To better understand how changes in economic and institutional contexts may affect fathers’ use of leave, we consider European countries with comparable ‘father-friendly’ designs: individual and non-transferable entitlement with high wage replacement rates (80–100%) and high ceilings. According to Blum et al. (2018), these are Iceland, Sweden, Norway, Spain, Slovenia and Portugal.Footnote 1 For reasons of space we compare only the first four countries. The evolution of national patterns of leave-taking by fathers in these countries is revealing. For Sweden, Duvander and Johansson (2012) show that the first non-transferable daddy month, paid at 90% of wage, increased the use of parental leave among all entitled fathers from 43% in 1994 to 75% in 1995, despite its coinciding with the economic crisis. The second daddy month introduced in 2002 and paid at 80% also increased fathers’ uptake, whereas the 2008 gender equality bonus had no effect (Duvander and Johansson 2012; Nyberg 2004). During the 1990s economic recession, the level of payment of parental leave decreased, but the daddy month was excluded from this decrease and reimbursed at higher levels (90% in 1995 and 85% in 1996) until the general payment level was set at 80% in 1998 (Johansson 2010). The first daddy month reform changed Swedish fathers’ behaviour without any interference from the economic recession, due to the political decision to exclude the high wage replacement for fathers from the general cuts in pay levels during the recession. The second daddy month confirmed the importance of the non-transferable and highly paid design elements.

The evolution in the use of leave in Norway also shows the strength of its design. In 1993, Norway introduced the first daddy month paid at 100% or 80% of wage depending on the total length of parental leave used, and thereafter the take-up rate for fathers rose sharply, from 4 to 78% during the first 5 years. Around 90% of eligible fathers used some of their entitlement or the entire quota (Kvande and Brandth 2017). Norway steadily increased the weeks in the fathers’ quota up to 14 weeks in 2013, but in 2014 a Conservative government reduced it to 10 weeks. The four times that the quota was increased in the years from 2000 onwards, the fathers’ uptake also increased, which means that more fathers were taking longer parental leave. So, in Sweden and Norway, fathers had the chance to experience reforms that extended the daddy quotas, while payment levels for fathers were not, or only temporarily and slightly, modified. In both countries, the extension of the non-transferable and highly paid leave led to growing uptake levels, which we interpret as this policy’s increasing legitimacy among fathers.

In contrast, Icelandic fathers have been hit by the economic crisis and austerity measures. Iceland introduced a daddy month in 2001, increased it to 2 months in 2002 and to 3 months in 2003. All of them were paid at 80% of the previous wage until the onset of the economic crisis in 2008, and, in general, fathers used their 3-month entitlement until 2007. This non-transferable and highly paid leave was a success story when Iceland was prospering (Arnalds et al. 2013). But the Icelandic case shows that it is not only the percentage paid but also the payment ceiling that plays an important role in uptake. According to the study by Sigurdardottir and Garðarsdóttir (2018), in the wake of the economic downturn the maximum payment for parental leave was reduced twice, in 2009 and in 2010. In addition, by that time fathers with higher earnings could only replace 75% of their wage. It is estimated that the maximum payment cut in January 2009 affected 26% of fathers who took parental leave, and the cuts in July 2009 and in January 2010 hit 36% and 46%, respectively. As a consequence, the average number of maximally paid leave days (full days) used by Icelandic fathers began to fall from 2009 onwards. The percentage of fathers who took at least 90 full days of leave declined from 78% in 2003 to 54% in 2011. Fathers from all income groups used less leave, but the uptake decreased most strongly among the fathers in the highest quartile of earnings. The authors conclude ‘money might be a major driving force behind the changes in the pattern of parental leave. Money presumably might thus play a large role if this trend is to be reversed’ (Sigurdardottir and Garðarsdóttir 2018, p. 354).

Although Spain went through an extremely deep recession from 2008 until 2013, there was no change in the wage replacement rate for paternity leave. Did fathers react as in Sweden, using their non-transferable and highly paid leave throughout the recession, or did they fear being penalised for absence from the job in a labour market context characterised by very high rates of unemployment and a large number of short-term jobs?

The Case of Spain

The first important reform to promote fathers’ use of paternity leave was implemented by a social-democratic government in March 2007 (Table 1), under the Spanish Law on Gender Equality. A non-transferable 2-week paternity leave and a 100% wage replacement level with a ceiling of 3074 € a month was introduced. Fathers could also use up to 10 weeks of transferable maternity leave, if the mother transferred her right. Until the present day, they have been able to take up their individual right to the subsequent parental leave, but this leave is not paid for by social security (see Table 1).

Table 1 Spanish statutory child care leaves, 2007–2018
Full size table

The Spanish economy had been growing fast since the beginning of the Millennium, to a great extent because of a boom in housing construction that lasted a decade, but which came to an abrupt halt in 2008. The recession mostly affected this masculinised economic sector. In fact, the male unemployment rate grew from 6% in 2007 to 10% in 2008 and increased steadily to 26% in 2013. In this situation of highly diffused layoffs of employees, fathers in temporary jobs and in companies under economic strain reported that they feared being penalised if they availed themselves of work–life balance measures (Abril et al. 2015). Furthermore, fathers in the private sector, self-employed men and those partnered with a non-employed woman made less use of their entitlement to paternity leave compared with their counterparts who had long-term jobs, worked in the public sector, were employees and were partnered with an employed mother. Fathers with a strong work-orientation and a more traditional gender ideology also used leave less frequently compared with fathers who were family-oriented or equally family-and-work oriented and fathers who did not agree that childcare is mostly a task for women (Escot et al. 2014; Meil et al. 2018; Romero-Balsas 2012). What happened to leave uptake with economic recovery and leave reform? In 2014, economic recovery began, and in 2017 paternity leave was extended to 1 month. Since 1 April 2019, paternity leave has been extended to 8 weeks and breastfeeding leave has become an individual entitlement. In addition, there are plans to increase paternity leave to 12 weeks in 2020 and 16 weeks in 2021 (Real Decreto-Ley 6/2019).

The Evolution of the Paternity Take-Up Rate, 2008–2018

Through a natural experiment approach, Escot et al. (2012) compared the behaviour of male employees who had children aged less than 1 year old before and after the 2007 reform, using mothers as the control group. They found that the non-transferable and highly paid Spanish paternity leave was used by most eligible fathers. We have used the numbers of employed fathers in the Labor Force Survey (SLFS) and the register data from Social Security to approximate the uptake rates of paternity benefits from 2007 to 2018, because no data are available on the number of entitled fathers (Fig. 1).

Fig. 1
figure 1

Source own elaboration with data from the Labor Force Survey for employed fathers and the official social security statistics for the use of leave allowances

Paternity take-up rates, male employment rate and gap in maternity–paternity leave users, 2007–2018. Note take-up rates refer to fathers aged 20 to 49 who are employed. The first vertical line refers to the introduction of 2 weeks in 2007, the second to 4 weeks in 2017 and the 3rd to 5 weeks (July 5th, 2018). The gap fathers to mothers is the proportion of male over female leave allowance beneficiaries.

Full size image

According to Escot et al. (2014) in 2006, 5282 fathers took some transferable maternity leave. One year later, they numbered 5204, whereas 173,161 fathers used the new paternity leave introduced in March 2007. As shown in Fig. 1, the latter represented 50% of employed fathers, increasing in 2009 to 67% of employed fathers using their 2-week social security paternity leave, despite a steep decline in male employment, concomitant high unemployment rates and increasing feelings of insecurity. During the crisis and the first three recovery years, the uptake rates oscillated between 66 and 71%, and only with the extension of the leave to 4 weeks in 2017 and to 5 weeks since mid-2018 did it rise to 80%. So, the uptake did not change substantially during the time that it was a 2-week period of leave, which means that economic recession itself may not be what deters employed fathers from using a non-transferable and highly paid leave at an aggregated level, even if the composition of leave users may change. It may be that only changes in payment levels, as shown by the Icelandic case, affect the aggregated use of such a type of leave (Sigurdardottir and Garðarsdóttir 2018). To understand this Spanish success story, it is also important to know that extension of paternity leave to 1 month should have been implemented in 2009, but it was postponed, year after year because of budget restrictions. Fathers and employers were expecting that paternity leave would become longer as soon as the economy recovered. It was not until the fourth year of recovery, in 2017, that paternity leave was increased to 4 weeks, and in July 2018 to 5 weeks. By then, only 20% of employed fathers with a small child did not use the month-long leave.

As shown in Fig. 1, from 2009 to 2018 around 34% to 20% of employed fathers did not use paternity leave, and this may be for two regulatory reasons, besides the other socio-economic factors mentioned above. First, not all employed fathers are entitled to a paid leave, because some have informal jobs and do not contribute to social security, or they do not fulfil minimum contributory conditions for entitlement. Due to lack of data, we cannot know how many fathers were affected by such problems, but only that in the age group 20–49, the difference between the number of employed men and the number of employed men with social security increased significantly during the economic recession. Second, self-employed fathers are entitled to paternity leave, but they often did not and still do not take advantage of their entitlement. These fathers receive a benefit based on their social security contributions, which frequently do not reflect their monthly income but a lower compulsory minimal social security contribution. In such cases paternity leave does not replace income at 100% but at a lower rate.

A Success Story

Looking at how the beneficiaries of paternity leave have evolved in relation to employed fathers, it can be said that the non-transferable and highly paid Spanish paternity leave has been a success story since its introduction in March 2007. Different factors may explain its positive reception and high uptake. This leave is a statutory entitlement, a benefit paid by social security for employees and the self-employed with a previous record of social security contributions, and it is paid at 100% of the contribution basis. For employees this means that benefits are equivalent to their wage, and for employers it does not represent an economic cost, but only an organisational challenge.Footnote 2

It may come as a surprise that around 80% of employed fathers use paternity leave in a country with a strong tradition of gender segregation. In fact, employed mothers still do a double shift and are the main providers of care work (Gracia and Kalmijn 2016). Yet, this gap has diminished between 2003 and 2010, because fathers have become more involved in the physical care of children, especially in couples with very young children (Cano 2019). In addition, the gender culture has changed importantly in recent decades, as reflected by higher family diversity and the changes in family law. Spain has been among the countries with more liberal legislation on abortion, assisted reproduction, same-sex marriage and adoption by same-sex couples (Domínguez-Folgueras and Castro-Martín 2013). Most young Spanish adults consider that fathers are as well suited to look after children as mothers, and half of Spanish women are classified as approving an egalitarian gender ideology (Naldini and Jurado 2013; Grunow et al. 2018). These phenomena are embedded in other social changes driven by women, such as the rise in their levels of education and the increase in employment rates among mothers with small children.

With respect to the gap between paternity and maternity leave beneficiaries, Fig. 1 also shows that its evolution is similar and the gender gap has closed. In 2008, 79 fathers used paternity leave per 100 mothers using maternity leave (100 is gender-equal uptake). During the crisis years, the fathers’ rate relative to mothers averaged 82 per 100; by 2016, the third year of recovery, it had risen to 88, and with the extension of leave, there are slightly more paternity than maternity leave beneficiaries (101). No gender gap now exists in the uptake rates of non-transferable and highly paid leave in Spain.Footnote 3 Yet, we note the difference in the duration of maternity versus paternity leave. The small gap in the uptake rates of maternity versus paternity leave only refers to the number of beneficiaries, but the Spanish leave system formally discriminates against fathers because of the much shorter length of the paternity leave compared with maternity leave (Table 1). According to the last available statistics offered by Spanish social security, until September 2018 mothers who gave birth to children and did not transfer any of their maternity days to their partners (99% of mothers) took an average of 113 days compared to the average of 30 days in the take-up of paternity leave.Footnote 4 Thus, there is a strong gender gap for both the days available and the days used. This study does not focus on the gender gap but on uptake gaps across different fathers.

Theoretical Framework and Hypotheses

A mix of factors at individual, couple and contextual level that are related to money and gender may affect fathers’ use of paternity leave (Risman and Davis 2013). At the institutional level, design influences fathers’ use of leave in two ways: first, by defining the opportunity costs of carrying out the care in terms of income loss, and, second, by increasing the social legitimation of men caring for their children. Welfare state institutions and laws are used by couples as reference points for making decisions on how to provide care and how to divide care work between them (Grunow and Veltkamp 2016). Our argument is that social policy reflects and creates social norms, which have the capacity to transform gender norms and the gendered use of time.

The non-transferable element sends a strong prescriptive and moral signal to fathers, because at one point the legislator decided to give men an individual entitlement to care for their new-born children, despite a long-lasting tradition of assigning care of small children to women. Non-transferability ensures that fathers can resist deep-seated inner resistance to taking up a traditionally female task, as well as external pressure by employers, co-workers and significant others (Bygren and Duvander 2006; Kvande 2009). The second element, a high wage replacement (and high ceiling) during absence from the job reduces opportunity costs by guaranteeing that men continue to be breadwinners, which is still a basic ingredient of most adult male identities, and it also ensures that families across social classes can afford the use of parental leave (Jurado-Guerrero et al. 2018; Margolis et al. 2019). For instance, the level of education and earnings were both found to affect the likelihood of using leave in Sweden (Bygren and Duvander 2006; Duvander and Johansson 2016). Yet, it is not only the leave design but also the economic context that can influence men’s decisions, as higher job insecurity during recessions may increase penalties to leave users through non-extension of temporary contracts and dismissals.

To sum up, the experience from the three northern European countries described above shows that parents are responsive to changes in payment levels and payment ceilings, which may come about because of the economic crisis and recovery. It also seems that high-earning fathers may be the most affected by lowered payment ceilings. Yet, what happens if payment levels are not changed, but the economic situation alone changes? Did economic recovery change the social gaps found in the use of Spanish paternity leave, or were these gaps only affected by the paternity leave expansion in 2017? Sweden and Spain are very different countries, but both introduced a non-transferable and highly paid leave for fathers at the beginning of an economic recession. They therefore provide a good scenario for testing the importance of leave design over many other contextual factors.

At the individual and couple level, we posit that factors related to both the family division of work and money drive uptake rates. We expect that fathers with higher economic opportunity costs in using leave—due to higher potential penalizations such as job loss and even the loss of their residence permit in the case of immigrants—and lower family opportunity costs—because the mother is not employed—will avail themselves of paternity leave less. In addition, the educational level of fathers may be positively related to family opportunity costs, because highly educated fathers may be more aware of the importance of developmental activities with children from an early age. In fact, they have been observed to spend more time on these (Sullivan et al. 2014; Baizán et al. 2014; Escot et al. 2014). Thus fathers who are self-employed, working in low-unskilled jobs or in the private sector, have a fixed-term contract, are less educated, were born abroad, or have a non-employed partner will be less likely to use paternity leave (Hypothesis 1).

How have these gaps between fathers evolved? At the contextual level, we expect that both the economic recovery and parental leave extension will have a positive effect on reducing differences across fathers. On the one hand, the end of the economic crisis in 2013 may lead to a decrease in fear of being economically penalised. Thus, we expect that during the economic recovery the gaps in the probabilities of uptake across fathers diminished compared with the crisis period, especially, between employees who are more and less protected by social class position, type of work contract, and public/private sector (Hypothesis 2). On the other hand, we interpret the leave extension in 2017 as an increase in the social legitimation of taking leave. Therefore, we expect that extension of paternity leave to 4 and 5 weeks has the potential to reduce all existing gaps in the use of this leave (Hypothesis 3).

The test of these hypotheses will allow an understanding of which micro-level factors drive the observed rise in the uptake of paternity leave among employed fathers. Hypothesis 2 will be tested looking at the recovery years 2014–2016, a 3-year period which allows for a time-lagged effect, compared to the crisis years 2008–2013. The expansion of paternity leave from 2 to 4/5 weeks in 2017/mid-2018 may carry over some positive effects from recovery. In this sense, if the economic recovery (2014–2016) impacts positively on the use of paternity leave, but not on the reduction of the differences between fathers (i.e. Hypothesis 2 is rejected), then a significant influence of leave extension period on the gaps will be mainly a net effect of the legal expansion (Hypothesis 3).

Data and Methods

Sample

We use data from the Spanish Labor Force Survey (SLFS), conducted quarterly by the National Institute of Statistics since 1976. The original sample consists of about 65,000 households, which means information about approximately 170,000 people is available. For this study we select a sample composed of men who are fathers (21 to 52-year-old men), who are employees and who have children 3 months old or younger. The identification of households with children of that age is possible because the SLFS collects information about the month of birth. We select this sample to work with a group of fathers who are potential users of paternity leave. Unfortunately, the characteristics of the SLFS prevent identifying all parents who use this leave, because the field work of the survey is carried out over 13 weeks (the week of reference for each person is the one prior to the interview), while parental leave lasted 2 weeks until 2016, 4 weeks in 2017, and 5 weeks in 2018. This implies that in each quarter we may find fathers who are not using parental leave during the week of reference, although they may have used it before or after this observation window.

This limitation prevents us from measuring the precise number of fathers using paternity leave by socio-economic background. However, the data allow us to distinguish which parents took leave to a greater or lesser extent compared to others. This is possible because the probability of being interviewed while taking leave is the same for any person. That is, the limitations of the SLFS when it comes to identifying every father who used paternity leave affect everyone, regardless of the personal characteristics of the individual. For the same reason, the likelihood of being observed while on leave increases with leave extension, but this probability affects all in an equal way.

The original sample of the SLFS (I/2008–IV/2018) of men living with children who are 3 months old or younger results in 12,393 observations. This figure is reduced to 8661 when we select only those working as employees, i.e. excluding inactive (n = 314), unemployed (n = 1482) and self-employed (n = 1936), and to 8628 after restricting the sample to men between 21 and 52 years old (33 cases dropped). After dropping the missing values from our control variables, the final sample of employee fathers is reduced to 8397 observations.

Additionally, we use another sample that includes the self-employed. The original sample in the SLFS of male self-employed living with children 3 months old or younger results in 1936 observations. After excluding those outside the range 21–52 years old (n = 18) and dropping the missing values, the final sample of self-employed fathers consists of 1867 observations. Analysis of this second sample includes the ‘sector of activity’ variable (see below), for which our first sample (i.e. 8397 employees) contains 93 new missing values, resulting in 8304 employees. Thus, this second final sample is composed of 10,171 observations (8304 employees + 1867 self-employed).

Variables

The SLFS asks about ‘why he/she has not worked [in the week of reference], having employment’. Among other answers, the questionnaire offers the option ‘leave for the birth of a child’. Therefore, our dependent variable includes the following two categories: not using paternity leave (0) versus using it (1).

To test our hypotheses, we consider three sets of key independent variables (see Table 5 for descriptives). First, we control for the period in which the fathers were interviewed: (i) during the economic crisis (2008–2013); (ii) during the economic recovery (2014–2016); and (iii) from the start of the parental leave extension to 4/5 weeks (2017–2018). In order to distinguish the different effects of the economic recovery and the leave extension period, the former will be used as the category of reference. Second, we also introduce factors related to family opportunity costs, such as the level of education (tertiary, upper secondary, lower secondary or less) and the labour status of the partner (partner employed, partner non-employed). Third, our independent variables for opportunity costs related to job characteristics are type of contract (permanent vs. temporary), type of sector (public vs. private) and occupational attainment. To measure this last variable, we use the EGP social class scheme (Erikson et al. 1979), which contains eleven categories. We have followed the operationalisation proposed by Ganzeboom and Treiman (2011) and adapted it to the National Classification of Occupations: (i) higher professionals (professionals, owners of large enterprises and higher managers); (ii) lower professionals (associate professionals, lower managers and higher sales workers); (iii) routine non-manual employees, manual supervisors and skilled workers; and (iv) low-unskilled workers and farm labourers. Fourth, we introduce immigrant status according to the country of birth because it is thought to increase opportunity costs due their often-unsure residence status, which is linked to having a job.

Finally, we also consider some control variables which have been found to influence use of leave in previous studies, such as age, age2 (to capture the potential non-linear effect), and the number of children under 16 living in the household. Given that information about the type of contract and the (public/private) sector are only available for employees, the sector of activity is also considered for the sample of fathers that includes both employees and self-employed. The sector of activity variable is based on Singelmann’s classification (1978): extractive and transformative sectors; construction; distributive services; producer services; public administration; social services; and personal services.

Analytical Strategy

Our analytical strategy consists of the estimation of logistic regression models on the probability of using paternity leave versus not using it (reference category). We chose this technique instead of linear probability modelling because the distribution of the probabilities for our dependent variable is close to 0 (Hellevik 2009). To ease the substantive interpretation of the coefficients, we also plot the average adjusted probabilities.

We are interested in the effect of the economic recovery period and the parental leave extension (vs. the economic crisis) on the gaps between those fathers more and less likely to take paternity leave. To study the changes between these groups of fathers across periods, we proceed in three steps. First, after presenting descriptive results, we estimate six different models with main effects and without controls for each gap between fathers: by type of worker, by education, by country of birth, by partner labour status, by social class, by type of contract and by type of sector. Second, we run two full models—one with all workers and another only with employees—controlling for all our independent variables. Finally, we estimate the seven social gap models with interactions between the period and the fathers’ profiles controlling for all independent variables.

Results

Descriptive Findings

Table 2 presents, by individual/couple characteristics and period, the percentage of employed men who were found to be using paternity leave. Figures are much lower than the register data, because the likelihood of finding a father on leave in the SLFS is systematically sub-estimated (see “Sample” section). We can interpret differences across the uptake rates in terms of trend (increasing or decreasing gaps between fathers) and statistical significance, but not measure the magnitude of those changes. These descriptive results show three interesting findings. First, during the Great Recession there were significant differences between fathers in the use of paternity leave. As previous empirical literature had indicated, it is confirmed that the leave is used more by employees, highly educated individuals, natives, employees in higher professional occupations, those in the public sector, those with a permanent contract and by men living with an employed partner. Second, the changes experienced during the first 3 years of the economic recovery differ depending on the gap. On the one hand, the differences in use by type of worker, educational level, country of birth and social class have increased during recovery in comparison with the years of the economic crisis. On the other hand, the gap between fathers by partner status, type of contract and sector has decreased or remained almost similar. Third, in contrast with the period of economic recovery, since the extension of leave in 2017, all gaps have diminished.

Table 2 Percentage of men using parental leave disaggregated by period and individual characteristics
Full size table

However, we wonder to what extent the changes between periods and gaps are statistically significant and, if they are, to what extent they are explained by composition effects. To clarify these two issues, we will estimate different logistic regression models.

Multivariate Analysis

Social Gaps in the Use of Paternity Leave

Table 3 presents the bivariate log odds of using paternity leave, controlling for different types of fathers and historic period. The seven main effects models (M1–7) show two results. First, the gaps expected by Hypothesis 1 are confirmed and suggest that fathers with higher economic opportunity costs and lower family costs take up paternity leave less. Fathers who are self-employed, in low-skilled occupations, with fixed-term contracts, working in the private sector, born abroad, or with a non-employed partner are less likely to use paternity leave. Second, the probability of taking leave during the economic recovery does not change in comparison with the Great Recession, but fathers are more likely to use paternity leave from 2017 onwards with the leave extension compared to the period of the economic recovery. Once we introduce all independent and control variables into our models (M8–9), most gaps between fathers and the differences between periods remain.

Table 3 Logistic regression on the likelihood of using parental leave
Full size table

The educational gradient of leave uptake appears in the bivariate model (M2), but disappears after controlling for all independent and control variables (M8–9). This suggests that the negative effect of having a lower secondary education or less may be a composition effect of these fathers being more frequently in employment situations with higher economic opportunity costs and higher risks of being penalised when using their leave entitlement. For instance, the lower educated may have more frequently low-skilled jobs with temporary contracts, which might explain that education ceases to be significant when one controls for social class, type of contract and other features.

Period Effects on the Paternity Leave Use Gaps

In Table 4, we look at the interaction of the historic period with the different gaps in uptake for all fathers controlling for all independent and control variables. The results point to two different patterns. On the one hand, Models 12, 13 and 16 show no changes in some of the fathers’ gaps over time. The influence of the economic recovery and the leave expansion do not have any significant effect on the different likelihoods of leave uptake between natives/immigrants, fathers with an employed/non-employed partner, and those working in the public/private sectors. These findings reject our Hypotheses 2 and 3 with respect to the positive impact of the economic recovery and the context of leave extension on the tendency of fathers of immigrant origin, with a non-employed partner or in the private sector, to catch up with their counterparts controlling for all other variables. This means that working in the private sector continues to be a handicap to using paternity leave, being an immigrant and thereby experiencing many disadvantages which complicate making use of the leave and the existence of a strong gender division of work in the home all deter fathers from using their entitlement.

Table 4 Logistic regression on the likelihood of using parental leave
Full size table

On the other hand, Table 4 shows a reduction in the gaps in leave uptake across some other types of fathers, but only from 2017 onwards. For example, Model 10 indicates that the self-employed are less likely than employees to use paternity leave during the economic recovery (− 1.327***). The interaction effects show that the differences between workers were smaller during the economic crisis (0.289), although this change in the gap with respect to the economic recovery is not statistically significant. Nevertheless, self-employed workers reduce their distance from employees’ uptake probabilities somewhat with the extension of leave to 4 and 5 weeks in 2017/2018 (0.627*). We find similar patterns in model 15 for differences by type of contract for the group of employees. During the crisis and the economic recovery, employees with fixed-term contracts are less likely to use their entitlement in comparison with fathers having permanent contracts, but this gap decreases with the paternity leave extension.

The changes in the gaps across fathers by educational level and social class show not only lower opportunity costs with leave extension for the more disadvantaged but reversing effects. On the one hand, the results of Model 11 show that during the economic crisis and the recovery there were no differences between the more and less educated, as shown in the full model for the whole period studied (M9, Table 3). However, in Model 11 we find that fathers with lower secondary education are more likely to take paternity leave than those with tertiary education during the years when paternity leave was extended, net of all other factors. In the same sense, Model 14 shows that with the extension of leave, unskilled employees increase the probability of their using their entitlement, up to the point of reversing the gap with respect to high professionals (0.781**), controlling for all other variables.

We can imagine two potential explanations for fathers in unskilled occupations and with lower secondary education having increased the likelihood of their taking leave when it was extended. First, it may be that fathers with lower secondary education are getting more conscious about the importance of becoming involved in care from early childhood, as hypothesised for fathers with tertiary education. This may be related to a social diffusion process (i.e. imitation effect) from tertiary and upper secondary to lower secondary education in the use of paternity leave. Second, at company level we could be moving from a culture of resistance to the acceptance of an extended father’s leave period as a new labour right. In 2017 with the extension of paternity leave to 1 month, the Spanish employers’ organisation (CEOE) stated that work–life balance should not be improved through men’s absence from the job: ‘Although it is true that men tend to use the existing work–life balance measures less frequently, which is mainly due to cultural reasons and not just to how these measures are structured, the goal of balancing work and life cannot entail men leaving or limiting their participation in the labour market to ease women’s participation therein, regardless of the fact that it is necessary to encourage the adoption of measures that promote co-responsibility’ (CEOE 2017). In March 2019, the government decided to reform the Spanish leave system by suppressing the terminology ‘maternity’ and ‘paternity’, renaming them as one unique ‘leave for birth and infant care’ and increasing the non-transferable and 100% ‘daddy part’ up to 16 weeks by January 2021, equivalent to the leave for mothers (Decreto-Ley 6/2019). Currently, the CEOE has accepted this reform implemented by the government under the pressure of a strengthened feminist movement in Spain in recent years.

In order to ease the interpretation of log odds and to visualise the above findings, Fig. 2 shows the average adjusted probabilities of using paternity leave for fathers by type of worker, educational level, social class and type of contract, the four changing social gaps identified. As we saw above, the changes in the gaps are statistically significant only during the period of leave expansion with respect to the economic recovery and no significant effects appear compared to the crisis. In Fig. 2, the gap between employees and self-employed seems to increase from 7.3 percentage points in 2014–2016 to 8.4 in 2017–2018. Yet, because the confidence intervals (95%) in the extension period are larger than in the recovery and crisis, a statistically significant reduction of the gap by type of worker emerges (M10). For education, the differences between fathers with tertiary education and those with lower secondary or less changes from 2.9 percentage points during the economic recovery to 5.1 points during the paternity leave extension, but in favour of the less educated. In the case of the higher professionals and the unskilled workers, the gap changes from 5 points to 2.5, but also here in favour of the unskilled fathers. Finally, the gap between workers with a permanent contract and with a fixed-term contract shrinks from 4.7 to 1.1 points.

Fig. 2
figure 2

Source own calculations from SLFS (2008–2018)

Average adjusted probabilities of using paternity leave. Men (21–52) with children 3 months old or less*. *Probabilities calculated from models in Table 4 controlling for all other variables. Note differences between fathers in each period must be interpreted considering the 95% confidence intervals (thin lines).

Full size image

To sum up, the results described confirm Hypothesis 1 (except for education), which suggests that the existing gross gaps in the use of paternity leave in Spain are an outcome of factors related to opportunity costs. With respect to the period effects, Hypothesis 2 is rejected in every case, because no gap between fathers is reduced during the economic recovery compared to the economic crisis. Thus, the crisis period does not show significantly higher opportunity costs of taking a 2-week leave. On the other hand, Hypothesis 3 stated that the use of paternity leave becomes a new social norm that is followed by many more fathers because institutions are used as reference points by couples and by society as a whole. We can confirm Hypothesis 3 at least in four cases, because the leave expansion period closes or even inverses gaps between fathers by type of worker, social class, educational level and type of contract.

Conclusion

The implementation of a 2-week non-transferable paternity leave with 100% wage replacement in Spain in 2007 has been very well received, despite the Great Recession. Employed fathers increased their uptake rates from 66% at the moment of recession to 80% when the period of leave was extended, and formerly reluctant fathers were encouraged to use, on average, 30 days of paternity leave and benefit. In this article, we have studied the evolution of the use of this right from 2008 until 2018, and have analysed the extent to which the social gaps in uptake rates have changed. We hypothesised that the economic recovery (2014–2016) would decrease fathers’ opportunity costs in the workplace (loss of job) and overall (loss of residence permit) when taking leave, and would thus increase the opportunity to care for their babies. We also posited that the paternity leave extension (2017–2018) would lead to closing most social gaps in the use of this non-transferable and 100% paid fathers’ leave, because taking up the leave is becoming a new social norm: fathers should get involved in the care of their new-born children.

Our results show that the end of the economic crisis and the subsequent less-threatening context has not served to significantly moderate the differences in the uptake rates of fathers. Instead, the extension of leave in 2017/2018 to 4 and then 5 weeks has been the decisive event in reducing or reversing various socio-economic gaps in the use of the paternity leave. The positive influence of the extension has been effective for those who had higher opportunity costs in relation to their jobs and for fathers with lower secondary education, independently of their job situation. The political and social legitimation of paternity leave has been fundamental in unlocking reluctance to taking it among the self-employed, employees with temporary contracts, employees in unskilled occupations and among fathers with lower secondary education or less. In the case of those last two groups, the gaps with respect to the employees in higher professional occupations and the higher educated employees actually reversed. The limits of this new social norm to use paternity leave are seen in immigrant fathers who are less able to use this right. Also, fathers with a non-employed wife and those working in the private sector continue to be nearly as unlikely to take paternity leave as they were before. In these cases, neither the economic recovery nor the extension to 4 and then 5 weeks has significantly reduced the gap with respect to other fathers.

One limitation of this study is the difficulty of excluding from the analysis employed fathers without the right to take paternity leave. Some employed individuals may be working in the informal economy, which will impede them from accessing the leave. Another limitation is the deliberate exclusion of unemployed fathers from our analysis, because only some of them are entitled to paternity benefit. During the crisis, the number of unemployed men without entitlement to unemployment or paternity benefit sharply increased, but no data on the exact number are available.

Despite these limitations, our analysis confirms the importance of a non-transferable and highly paid leave design in a southern European country for encouraging the uptake of leave among fathers. In comparison with recent reforms to fathers’ leave in other countries, where large proportions of fathers do not use their entitlement, the Spanish daddy month has had a great reception among new fathers. Nevertheless, it has to be remembered that the closure and reversal of several social gaps in the use of paternity leave does not mean that Spanish fathers use as many leave days as mothers. In 2018, on average, they used one quarter of the days that mothers used. This reflects the gender gap in the entitlement to paternity leave (5 weeks) compared to maternity leave (16 weeks). It is likely that this gap will disappear by 2021, as the reform passed in March 2019 has the aim of expanding the father’s element of the new leave for birth and infant care to match the 16 weeks of non-transferable and fully paid entitlement that mothers already have. Since this last reform, the Spanish case has become a unique social experiment in pushing gender equality in leave entitlement to its fullest extent.