Educational attainment and maternity in Spain: not only “when” but also “how”


This paper describes how women with different human capital endowments behave differently with regard to the decision to bear their first child and disentangles the overall effects of education on the timing of the birth of the first child through a set of intervening variables. To do so, a decomposition technique that enables distinction between direct and indirect effects in logistic regressions is deployed. Education drives different fertility behavior patterns and delays fertility through several mechanisms: attachment to the labor market, non-traditional values, and the characteristics of the partner and the partnership. Nevertheless, the ability of the mechanisms explored in this study to explain the link between education and fertility timing is rather limited, and a large part of this relationship remains unexplained. In addition, we find evidence of differences across educational groups in the way women’s decisions to have their first child respond to several explanatory factors, which points at a structural change in the decision taking amongst women with different educational endowments.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3


  1. 1.

    The education field also has consequences with regard to the timing and number of children (Martín-García and Baizán 2006). Namely, there are several features of study disciplines that may shape fertility decisions (Van Bavel 2010): the expected starting wage, steepness of the earning profile, attitudes towards gendered family roles, and gender composition. Unfortunately, we are not able to observe the education field in our dataset, but we know that the careers of highly educated women may be affected by a steeper earning profile and sometimes better working conditions in terms of facilitating a work–life balance. Women may self-select into educational paths that lead to more family-friendly employers/careers/jobs. Obtaining a university degree increases the likelihood of working in the public sector, where the family–work balance is considerably easier to achieve than in the private sector.

  2. 2.

    Because parenthood decisions are made along the life cycle and fertility declines with age, age is expected to be related to maternity decisions. However, the direction of the relation is unclear in the presence of other observable variables and evidence related to the connection between age and maternity is diverse. De la Rica and Ferrero (2003) confirm that in Spain, women under the age of 35 years are more likely to have a child. Álvarez-Llorente (2002) states a decrease in the probability of a first birth in Spain from the age of 30 years in the case of non-working women and from 35 years for working women. Sheran (2007) confirms that the number of births declines with age in the United States whereas Del Boca et al. (2005) found the opposite trend in France and the United Kingdom.

  3. 3.

    It may also be the case that highly educated women suffer greater wage penalties when becoming mothers and this drives their postponement strategies. Recent research on motherhood wage penalties in Spain (Molina and Montuenga 2009; Fernández-Kranz et al. 2013) does not directly provide evidence of a larger earnings dip for more educated female workers. However, the findings in Fernández-Kranz et al. (2013) point in this direction, stating that women holding permanent positions experience greater wage dips.

  4. 4.

    Age-wage profiles reported in previous research on the Spanish labor market clearly shows that this is the case (Marcenaro-Gutiérrez and Navarro 2005, p. 81). This is consistent with a higher likelihood of permanent contracts and on-the-job vocational training in Spain for qualified workers.

  5. 5.

    In countries characterized by easier access to rental or property housing markets, residential independence and family formation occur earlier (Mulder 2006).

  6. 6.

    Prior to selecting the potential mothers for the multivariate analysis, we estimated the pairwise correlation coefficient between the interviewees’ number of siblings (i.e., children of their own mothers) and the total number of children they have had. The value was 0.245 for the entire sample, and slightly lower for tertiary education graduates (0.183) than for secondary education graduates (0.207). All were significant at the 95 % confidence level.

  7. 7.

    Premarital births were excluded from the analysis because they are very uncommon in Spain, particularly among women in the earlier cohorts of the sample (95 % of the children of the women in the initial sample were born to couples).

  8. 8.

    Within this sample selection, our youngest interviewees were 25 years old at the time of the interview. We made this decision to allow those women belonging to the most recent cohorts in the sample to complete their studies before the date of the interview. It is unlikely that we will miss first births as a result of this decision because, in Spain, pregnancies among students are uncommon. Moreover, among women born in the first half of the 1970 s, 80 % of children were born to mothers who were 25 years of age or older.

  9. 9.

    The log rank test (with a χ2 distribution with a value of 14.67 and significance at 99 %, with 1 degree of freedom) and Wilcoxon (Breslow) (29.42, equally significant) confirm that there are significant differences in survival functions across women with different levels of educational attainment.

  10. 10.

    The log rank test (with a χ2 distribution with a value of 75.94 and significance at 99 %, with 1 degree of freedom) and Wilcoxon (Breslow) (102.20, equally significant) confirm that there are significant differences in survival functions across women of different educational attainment levels.

  11. 11.

    This would initially challenge hypothesis H2B, therefore, greater attention has been paid to the distribution of this variable. Namely, although the average accumulated experience prior to marriage does not differ greatly across educational attainment levels, it is the result of the combination of two factors: less educated women are less likely to be in paid employment before becoming mothers (57 % compared with 71 % for their educated counterparts), however, when they are, they report greater employment experience (3.84 years in average) than highly educated women (3.27 years). Because women who are not employed before having children also often report no previous working experience, the average experience levels of both groups seem to be similar, although their distributions differ across educational attainment levels.

  12. 12.

    Observations are independent across individuals but not across time. For this reason, we use clustering, which specifies that the observations are independent across individuals (clusters) but not necessarily within individuals (over time). Clustering affects the estimated standard errors and variance–covariance matrix of the estimators but not the estimated coefficients (Gutiérrez-Domènech 2005).

  13. 13.

    The type of partnership is a time-varying feature. Thus, in each yearly observation, we may determine whether interviewees are married or cohabiting.

  14. 14.

    The Spanish territory has been divided into four large areas: North (Asturias, Cantabria, Galicia, Navarre, the Basque Country, and La Rioja), East (Aragon, Balearic Islands, Catalonia, and Valencia), Central (Castilla y Leon, Castilla-La Mancha, and Madrid), and South (Andalusia, Canary Islands, Extremadura and Murcia).

  15. 15.

    The logic/rationale behind these two methods is exactly the same, but they do not have to result in exactly the same estimates. They differ in the comparison group defined to study the counterfactual scenario to disentangle direct and indirect effects: In the first method, it is assumed that low or moderately educated women are actually highly educated women (namely, they have the same distribution of the intervening variables as highly educated women). In the second method, it is assumed that highly educated women are actually low or moderately educated women. Given that both methods run parallel and only differ in the determination of the counterfactual group, the results of both methods are quite similar.

  16. 16.

    In Erikson et al. (2005), M had to be a normal variable whereas the generalization of Buis (2010) allows for any distribution. In our case, all of them are dummy variables except one: experience in the labor market, which is measured in the number of years since the beginning of the employment career (i.e., the first job) until the beginning of the first partnership.

  17. 17.

    To make our results more comparable across educational groups, we display marginal effects instead of coefficients. Furthermore, to test the significance of differences in the profiles of first maternity across low to moderately and highly educated women displayed in Table 1, we performed an assessment of the differences in the contribution of all explanatory variables to the predicted probabilities across both educational groups following the strategy defined in Long (2009). We verified that most of differences in the impact of explanatory variables via the marginal effects in Table 1 point are significant; the only exceptions being birth cohort and being in the first 4 years of marriage/cohabitation. The results are not shown because of space limitations, but are available to the interested reader upon request. We thank one of the referees for pointing out the need for testing significance in the observed differences across educational groups.

  18. 18.

    Unfortunately, we are unable to control for contract type because the dataset at hand has a large share of missing data (approximately 75 % of the observations of employment spells do not report type of contract or employment status).

  19. 19.

    Interestingly, the category “no answer” is very marginal here: only 0.95 % of women refused to reply to this question.

  20. 20.

    This result needs to be qualified as the differences in the contribution of the birth cohort variables to the delay in the first maternity are not significant. However, as long as they do not follow the expected trend, H3F is not confirmed.

  21. 21.

    We thank one of the referees for the suggestion to plot the results in this way.


  1. Adserà, A. (2006). An economic analysis of the gap between desired and actual fertility: The case of Spain. Review of Economics of the Household, 4(1), 75–95.

    Article  Google Scholar 

  2. Adserà, A. (2004). Changing fertility rates in developed countries. The impact of labor market institutions. Journal of Population Economics, 17(1), 17–43.

    Article  Google Scholar 

  3. Ahn, N., & Mira, P. (2001). Job bust, baby bust?: Evidence from Spain. Journal of Population Economics, 14(3), 505–521.

    Article  Google Scholar 

  4. Alba, A., Álvarez-Llorente, G., & Carrasco, R. (2009). On the estimation of the effect of labour participation on fertility. Spanish Economic Review, 11(1), 1–22.

    Article  Google Scholar 

  5. Álvarez-Llorente, G. (2002). Decisiones de Fecundidad y Participación Laboral de la Mujer en España. Investigaciones Económicas, 26(1), 187–218.

    Google Scholar 

  6. Amuedo-Dorantes, C., & Kimmel, J. (2005). The motherhood wage gap for women in the United States: The importance of college and fertility delay. Review of Economics of the Household, 3(1), 17–48.

    Article  Google Scholar 

  7. Baizán, P., Aassve, A., & Billari, F. C. (2003). Cohabitation, marriage, and first birth: The interrelationship of family formation events in Spain. European Journal of Population, 19(2), 147–169.

    Article  Google Scholar 

  8. Becker, G. S. (1960). An Economic Analysis of Fertility. In Universities National Bureau Committee for Economic Research (Ed.), Demographic and economic change in developed countries (pp. 209–231). Princeton, NJ: Princeton University Press.

    Google Scholar 

  9. Becker, G. D. (1991). A Treatise on the family (Enlarged ed.). Cambridge, MA: Harvard University Press.

    Google Scholar 

  10. Blackburn, M. L., Bloom, D. E., & Neumark, D. (1993). Fertility timing, wages and human capital. Journal of Population Economics, 6(1), 1–30.

    CAS  Article  PubMed  Google Scholar 

  11. Bloemen, H., & Kalwij, A. S. (2001). Female labor market transitions and the timing of births: A simultaneous analysis of the effects of schooling. Labour Economics, 8(5), 593–620.

    Article  Google Scholar 

  12. Brañas-Garza, P., & Neuman, S. (2007). Parental religiosity and daughter’s fertility: The case of Catholics in Southern Europe. Review of Economics of the Household, 5(3), 305–327.

    Article  Google Scholar 

  13. Buis, M. L. (2010). Direct and indirect effects in a logit model. The Stata Journal, 10(1), 11–29.

    PubMed Central  PubMed  Google Scholar 

  14. Castro-Martín, T. (1992). Delayed childbearing in contemporary Spain: Trends and differentials. European Journal of Population, 8(3), 217–246.

    Article  PubMed  Google Scholar 

  15. De La Rica, S., & Ferrero, M. D. (2003). The effect of fertility on labour force participation: The Spanish evidence. Spanish Economic Review, 5(2), 153–172.

    Google Scholar 

  16. De la Rica, S., & Iza, A. (2005). Career planning in Spain: Do fixed-term contracts delay marriage and parenthood? Review of Economics of the Household, 3(1), 49–73.

    Article  Google Scholar 

  17. De Wit, M. L., & Ravanera, Z. R. (1998). The changing impact of women’s educational attainment and employment on the timing of births in Canada. Canadian Studies in Population, 25(1), 45–67.

    Google Scholar 

  18. Del Boca, D., Pascua, S., & Pronzato, C. (2005). Fertility and employment in Italy, France and the UK. Labour, 19(S1), 51–77.

    Article  Google Scholar 

  19. Del Boca, D., & Sauer, R. M. (2009). Life cycle employment and fertility across institutional environments. European Economic Review, 53(3), 274–292.

    Article  Google Scholar 

  20. Erikson, R., Goldthorpe, J. H., Jackson, M., Yaish, M., & Cox, D. R. (2005). On class differentials in educational attainment. Proceedings of the National Academy of Science, 102(27), 9730–9733.

    CAS  Article  ADS  Google Scholar 

  21. Fernández-Kranz, D., Lacuesta, A., & Rodríguez-Planas, N. (2013). The motherhood earnings dip: Evidence from administrative records. Journal of Human Resources, 48(1), 169–197.

    Article  Google Scholar 

  22. Francesconi, M. (2002). A joint dynamic model of fertility and work of married women. Journal of Labor Economics, 20(2), 336–380.

    Article  Google Scholar 

  23. Goldstein, J. R., Sobotka, T., & Jasilioniene, A. (2009). The end of lowest-low fertility?. Population and Development Review, 35(4), 663–700.

    Article  Google Scholar 

  24. Gustafsson, S. (2005). Having kids later. Economic analyses for industrialized countries. Review of Economics of the Household, 3(1), 5–16.

    Article  Google Scholar 

  25. Gustafsson, S., & Worku, S. (2005). Assortative mating by education and postponement of couple formation and first birth in Britain and Sweden. Review of Economics of the Household, 3(1), 91–113.

    Article  Google Scholar 

  26. Gutiérrez-Domènech, M. (2008). The impact of the labour market on the timing of marriage and births in Spain. Journal of Population Economics, 21(1), 83–110.

    Article  Google Scholar 

  27. Haan, P., & Wrohlich, K. (2011). Can child care policy encourage employment and fertility? Evidence from a structural model. Labour Economics, 18(4), 498–512.

    Article  Google Scholar 

  28. Hakim, C. (2003). A new approach to explaining fertility patterns: preference theory. Population and Development Review, 29(3), 349–374.

    Article  Google Scholar 

  29. Herrarte, A., Moral-Carcedo, J., & Sáez, F. (2012). The impact of childbirth on Spanish women’s decisions to leave the labor market. Review of Economics of the Household, 10(3), 441–468.

    Article  Google Scholar 

  30. Hotchkiss, J. L., Pitts, M. M., & Walker, M. B. (2011). Labor force exit decisions of new mothers. Review of Economics of the Household, 9(3), 397–414.

    Article  Google Scholar 

  31. Jenkins, S. P. (1995). Easy estimation methods for discrete-time duration models. Oxford Bulletin of Economics and Statistics, 57(1), 129–138.

    Article  Google Scholar 

  32. Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations. Journal of the American Statistical Association, 53(282), 457–481.

    MathSciNet  Article  MATH  Google Scholar 

  33. Kohler, H., Billari, F., & Ortega, J. (2002). The emergence of lowest-low fertility in Europe during the 1990s. Population and Development Review, 28(4), 641–660.

    Article  Google Scholar 

  34. Kravdal, Ø., & Rindfuss, R. R. (2008). Changing relationships between education and fertility: A study of women and men born 1940 to 1964. American Sociological Review, 73(5), 854–873.

    Article  Google Scholar 

  35. Kreyenfeld, M. (2010). Uncertainties in female employment careers and the postponement of parenthood in Germany. European Sociological Review, 26(3), 51–366.

    Article  Google Scholar 

  36. Long, J. S. (2009). Group comparisons in logit and probit using predicted probabilities. Working paper draft 2009-06-25. Retrieved from:

  37. Marcenaro-Gutiérrez, O., & Navarro, L. (2005). Nueva evidencia sobre el rendimiento del capital humano en España. Revista de Economía Aplicada, 13(37), 69–88.

    Google Scholar 

  38. Martín-García, T., & Baizán, P. (2006). The impact of the type of education and of educational enrolment on first births. European Sociological Review, 22(3), 259–275.

    Article  Google Scholar 

  39. Miller, A. R. (2011). The effect of motherhood timing on career path. Journal of Population Economics, 24(3), 1071–1100.

    Article  Google Scholar 

  40. Mills, M. R., Rindfuss, R., McDonald, P., & te Velde, E. (2011). Why do people postpone parenthood? Reasons and social policy incentives. Human Reproduction Update, 17(6), 848–860.

    PubMed Central  Article  PubMed  Google Scholar 

  41. Moffit, R. A. (1984). Profiles of fertility, labor supply, and wages of married woman: A complete life-cycle model. Review of Economic Studies, 52(2), 765–799.

    Google Scholar 

  42. Molina, J. A., & Montuenga, V. (2009). The motherhood wage penalty in Spain. Journal of Family and Economic Issues, 30(3), 237–251.

    Article  Google Scholar 

  43. Mulder, C. H. (2006). Home-ownership and family formation. Journal of Housing and Built Environment, 21(3), 281–298.

    Article  Google Scholar 

  44. Nicoletti, C., & Tanturri, M. L. (2008). Differences in delaying motherhood across European countries: Empirical evidence from the ECHP. European Journal of Population, 24(2), 157–183.

    Article  Google Scholar 

  45. O’Donoghue, C., Meredith, D., & O’Shea, E. (2011). Postponing maternity in Ireland. Cambridge Journal of Economics, 35(1), 59–84.

    Article  Google Scholar 

  46. Pronzato, C. D. (2009). Return to work after childbirth: Does parental leave matter in Europe? Review of Economics of the Household, 7(4), 341–360.

    Article  Google Scholar 

  47. Sheran, M. (2007). The career and family choices of women: A dynamic analysis of labor force participation, schooling, marriage, and fertility decisions. Review of Economic Dynamics, 10(3), 367–399.

    Article  Google Scholar 

  48. Suárez, M. J. (2013). Working mothers’ decisions on child care: The case of Spain. Review of Economics of the Household,. doi:10.1007/s11150-013-9189-6.

    Google Scholar 

  49. Van Bavel, J. (2010). Choice of study discipline and the postponement of motherhood in Europe: The impact of expected earnings. Gender Composition and Family Attitudes. Demography, 47(2), 439–458.

    PubMed  Google Scholar 

  50. Zhang, L. (2011). The influence of cohabitation on male and female fertility. In Male fertility patterns and determinants. The Springer Series on Demographic methods and population analysis, 27 (Part 3), (pp. 143–176). Springer, New York.

Download references


The authors wish to thank Dr. Maarten Buis for his help and advice on the application of his STATA program ldecomp while Maria A. Davia was visiting WZB (Wissenschaftszentrum Berlin für Sozialforschung), which hospitality is greatly acknowledged. The usual disclaimer applies.

Author information



Corresponding author

Correspondence to María A. Davia.



See Tables 3, 4 and 5.

Table 3 Mean values of the variables used in the models of first maternity (categorical variables as relative frequencies, in  %)
Table 4 Changes in values across cohorts and educational attainments (as  % in each birth cohort and educational group)
Table 5 Robustness checks: relative size of the indirect effect of the intervening variables in different specifications

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Davia, M.A., Legazpe, N. Educational attainment and maternity in Spain: not only “when” but also “how”. Rev Econ Household 13, 871–900 (2015).

Download citation


  • Fertility
  • Human capital
  • Intervening variables
  • Duration analysis
  • Decomposition

JEL Classification

  • J13
  • J16
  • J24