The Gender Wage Gap Among College Graduates in Italy

Abstract

The paper investigates the gender wage gap among recently graduates, controlling for job and academic variables and for the field of study, as women lag behind in highly remunerative majors. The raw gender differential in hourly wages is 5.6%. Although including academic variables and the field of study—on top of job-related variables—slightly reduces the unexplained gap, the latter still accounts for most of the total difference. Using quantile decomposition, the paper shows that the unexplained gap increases along the wage distribution, indicating a glass ceiling effect. Heterogeneities arise among fields of study: the largest total gap emerges in Law, Political-Social sciences, and Economics-Statistics. In most disciplines, there is a significant unexplained gap—from 3.3% (Medicine), to 8.7% (Law), up to 9.6% (Agriculture)—which constitutes the largest share of the difference, confirming that most of the wage gap remains unexplained also by major. Finally, I use geographical differences to explore the influence of institutional and macro-economic variables, as well as of attitudes towards gender norms. The results indicate that childcare and part-time employment availability are correlated with lower gender wage gaps, while traditional gender norms are associated with higher differentials.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3

Notes

  1. 1.

    There are other additional differences: (i) I use wages per hour as a dependent variable, while Anelli and Peri (2015) have only information about income per year; (ii) I consider recently graduated individuals, while they have a pooled sample of people who graduated between five and fifteen years earlier; (iv) in the regressions, I include a much larger set of control variables that may reduce the gender gap—detailed in Sect. 4.

  2. 2.

    I may sometimes use the word ‘effect’ to simplify the exposition, but the reader should interpret the results as correlations.

  3. 3.

    The field of study control may be endogenous. For this reason, I include it in a subsequent specification, after having included only other academic variables.

  4. 4.

    Master degree includes both master degrees and single-cycle 5-year degrees, at the end of which the student directly obtains a master degree.

  5. 5.

    I.e. the student has not completed the degree within the set time period (usually 3 years for bachelor and 2 years for master degrees) (‘fuori corso’).

  6. 6.

    The variables that should affect labour supply but not wages.

  7. 7.

    See Vergolini and Vlach (2017) for a description of the Italian educational system.

  8. 8.

    Math-Sciences includes mathematics, physics, and computer science (distinguished from Natural Sciences).

  9. 9.

    Not shown, available upon request.

  10. 10.

    Controlling for all the independent variables, except sectors of employment (as column (f) in Table 6) yields the same results, reported in Table 17 in the Appendix.

  11. 11.

    As discussed above as regards the coefficient associated with the gender dummy, also coefficients associated with the field of study should not be interpreted in causal terms, but as partial correlations.

  12. 12.

    As robustness checks, I estimate the OLS regression and the O–B decomposition (full specification) also on three subgroups: (i) only bachelor students; (ii) only master students; (iii) only students graduated before age 30 (Table 18). In the first group, the raw gender gap is slightly larger (6.2%), whereas in the second and third case it is slightly smaller (4.8 and 4.5%), while the OLS female coefficients are very similar to the one estimated in the main analysis. Moreover, the main conclusions are unaffected: the gender gap is unaccounted for by observable characteristics.

  13. 13.

    I also estimated the O–B decomposition without any control variable not included in Piazzalunga and Di Tommaso (2016). Also in this case (available upon request), the unexplained gap among graduated students is smaller than in the full population.

  14. 14.

    Summary statistics are presented in Table 19, columns (c) to (g). Definitions of the variables and sources are presented in Table 22.

  15. 15.

    Summary statistics are presented in Table 20.

  16. 16.

    Checchi and Peragine (2010) show that the South of Italy is characterized by higher levels of inequality.

References

  1. Addabbo T, Favaro D (2011) Gender wage differentials by education in Italy. Appl Econ 43(29):4589–4605

    Article  Google Scholar 

  2. Anelli M, Peri G (2015) Gender gap in Italy: the role of college majors. In: Boeri T, Patacchini E, Peri G (eds) Unexplored dimensions of discrimination. Oxford University Press, London, pp 79–109

    Chapter  Google Scholar 

  3. Azmat G, Petrongolo B (2014) Gender and the labor market: What have we learned from field and lab experiments? Labour Econ 30:32–40

    Article  Google Scholar 

  4. Barone C (2011) Some things never change: gender segregation in higher education across eight nations and three decades. Sociol Educ 84(2):157–176

    Article  Google Scholar 

  5. Barone C, Schizzerotto A, Abbiati G M, Assirelli G (2017) Gender, information barriers and field of study choice: a field experiment. LIEPP Working Paper 63

  6. Becker GS (1957) The economics of discrimination. University of Chicago Press, Chicago

    Google Scholar 

  7. Bertrand M (2011) New perspectives on gender. In: Ashenfelter OC, Card D (eds) Handbook of labour economics, vol 4b. Elsevier, Amsterdam, pp 1545–1592

    Google Scholar 

  8. Bertrand M, Chugh D, Mullainathan S (2005) Implicit discrimination. Am Econ Rev 95(2):94–98

    Article  Google Scholar 

  9. Black DA, Haviland AM, Sanders SG, Taylor LJ (2008) Gender wage disparities among the highly educated. J Hum Resour 43(3):630–659

    Google Scholar 

  10. Blau FD, Kahn LM (2003) Understanding international differences in the gender pay gap. J Labor Econ 21(1):106–144

    Article  Google Scholar 

  11. Blau FD, Kahn LM (2017) The gender-wage gap: extent, trends, and explanations. J Econ Lit 55(3):789–865

    Article  Google Scholar 

  12. Blinder AS (1973) Wage discrimination: reduced form and structural estimates. J Hum Resour 8(4):436–455

    Article  Google Scholar 

  13. Brilli Y, Del Boca D, Pronzato C (2016) Does child care availability play a role in maternal employment and children’s development? Evidence from Italy. Rev Econ Household 14(1):27–51

    Article  Google Scholar 

  14. Campa P, Casarico A, Profeta A (2011) Gender culture and gender gap in employment. CESifo Econ Stud 57(1):156–182

    Article  Google Scholar 

  15. Ceci SJ, Williams WM (2010) Sex differences in math-intensive fields. Curr Dir Psychol Sci 19(5):275–279

    Article  Google Scholar 

  16. Charles M, Bradley K (2009) Indulging our gendered selves? Sex segregation by field of study in 44 countries. Am J Sociol 114(4):924–976

    Article  Google Scholar 

  17. Checchi D (2016) Rapporto sullo stato del sistema universitario e della ricerca. Agenzia Nazionale di Valutazione del Sistema Universitario e della Ricerca, Rome

    Google Scholar 

  18. Checchi D, Peragine V (2010) Inequality of opportunity in Italy. J Econ Inequal 8(4):429–450

    Article  Google Scholar 

  19. Chernozhukov V, Fernández-Val I, Melly B (2013) Inference on counterfactual distribution. Econometrica 81(6):2205–2268

    Article  Google Scholar 

  20. Christofides LN, Polycarpou A, Vrachimis K (2013) Gender wage gaps, ‘sticky floors’ and ‘glass ceilings’ in Europe. Labour Econ 21:86–102

    Article  Google Scholar 

  21. Cobb Clark D (2016) Biology and gender in the labor market. IZA Discussion paper 10386

  22. Contini D, Di Tommaso ML, Mendolia S (2017) The gender gap in mathematics achievements. Evidence from Italian data. Econ Educ Rev 58:32–42

    Article  Google Scholar 

  23. De Pascali P (2017) Capabilities, entitlements, and occupational gaps: three essays on gender discrimination in the South of Italy. PhD thesis, University of Salento

  24. Del Boca D (2002) The effect of child care and part time opportunities on participation and fertility decisions in Italy. J Popul Econ 15(3):549–573

    Article  Google Scholar 

  25. Duncan OD, Duncan B (1955) A methodological analysis of segregation indexes. Am Sociol Rev 20(2):210–217

    Article  Google Scholar 

  26. Eurostat (2017a) Gender equality datasets. http://ec.europa.eu/eurostat/web/equality/data/database

  27. Eurostat (2017b) Involuntary part-time employment as percentage of the total part-time employment, by sex and age. http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=lfsa_eppgai&lang=en

  28. Flabbi L (2011) Gender differences in education, career choices and labour market outcomes on a sample of OECD countries. Background Paper to World Development Report 2012: Gender Equality and Development, World Bank, Washington. D.C, World Development Report, p 2012

  29. Fortin N (2005) Gender role attitudes and the labour-market outcomes of women across OECD countries. Oxf Rev Econ Policy 21(3):416–438

    Article  Google Scholar 

  30. Heckman JJ (1979) Sample selection bias as a specification error. Econometrica 47(1):153–161

    Article  Google Scholar 

  31. Istat (2009) L’inserimento professionale dei laureati. Anno 2007

  32. Istat (2012) I laureati e il lavoro. Indagine 2011 sui laureati del 2007

  33. Istat (2016) Formazione universitaria—laureati. http://dati.istat.it/

  34. Juhn C, McCue K (2017) Specialization then and now; marriage, children, and the gender earnings gaps across cohorts. J Econ Perspect 31(1):183–204

    Article  Google Scholar 

  35. Koenker R, Bassett G (1978) Regression quantiles. Econometrica 46(1):33–50

    Article  Google Scholar 

  36. Mann A, DiPrete TA (2013) Trends in gender segregation in the choice of science and engineering majors. Soc Sci Res 42(6):1519–1541

    Article  Google Scholar 

  37. Mussida C, Picchio M (2014a) The gender wage gap by education in Italy. J Econ Inequal 12(1):117–147

    Article  Google Scholar 

  38. Mussida C, Picchio M (2014b) The trend over time of the gender wage gap in Italy. Empir Econ 46(3):1081–1110

    Article  Google Scholar 

  39. Oaxaca R (1973) Male–female wage differentials in urban labor markets. Int Econ Rev 14(3):693–709

    Article  Google Scholar 

  40. Olivetti C, Petrongolo B (2008) Unequal pay or unequal employment? A cross-country analysis of gender gaps. J Labor Econ 26(4):621–654

    Article  Google Scholar 

  41. Phelps ES (1972) The statistical theory of racism and sexism. Am Econ Rev 62(4):659–661

    Google Scholar 

  42. Piazzalunga D, Di Tommaso ML (2016) The increase of the gender wage gap in Italy during the 2008–2012 economic crisis. IZA Discussion Paper 9931

  43. Reuben E, Wiswall M, Zafar B (2017) Preferences and biases in educational choices and labour market expectations: shrinking the black box of gender. Econ J 127(604):2153–2186

    Article  Google Scholar 

  44. Triventi M (2013) The gender wage gap and its institutional context: a comparative analysis of European graduates. Work Employ Soc 27(4):563–580

    Article  Google Scholar 

  45. Vergolini L, Vlach E (2017) Family background and educational path of Italian graduates. High Educ 73(2):245–259

    Article  Google Scholar 

  46. Waldfogel J (1997) The effect of children on women’s wages. Am Sociol Rev 62(2):209–217

    Article  Google Scholar 

Download references

Acknowledgements

This article has benefited from the comments provided by Daniela Del Boca, Maria Laura Di Tommaso, Enrico Rettore, and my former colleagues at FBK-IRVAPP. I would like to thank the editor and the referees for their time and valuable remarks.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Daniela Piazzalunga.

Appendices

Appendix A

See Tables 15, 16, 17, 18, 19, 20, 21 and 22

Table 15 Summary statistics for employed and non-employed, by gender, 2011
Table 16 Sample selection, 2011
Table 17 OLS regression: academic and high school related coefficients (full spec. w/o sectors), 2011
Table 18 Robustness checks: OLS regression and O–B decomposition on subpopulations, 2011
Table 19 Institutional and macro variables, by region, 2011
Table 20 Attitudes towards gender equality and family roles, by region, 2008
Table 21 Oaxaca–Blinder decomposition by region, 2011
Table 22 Regional variables and social norms: definition and sources

Appendix B

This Appendix complements the description of selection into employment (Sect. 4.1 of the paper).

In the paper, I present descriptive evidence that the differences between employed and not employed individuals are similar among men and women, with few exceptions. Moreover, there are distinct types of selection taking place. Individuals do not work because of:

  1. (i)

    further investment in human capital: individuals who continue to study or are engaged in some form of paid training;

  2. (ii)

    unemployment: individuals who cannot find a job;

  3. (iii)

    general inactivity: individuals who do not work for personal reasons or other reasons.

Distinct types of selection have different—sometimes opposite—effects on the characteristics of employed individuals. Employed people are more likely to have studied subjects with a high degree of employability (such as Engineering and Economics-Statistics), but are less likely to have top grades, to come from academic high-school tracks, and from advantaged socio-economic backgrounds.

Overall, these channels operate similarly for men and women, but some differences emerge. However, such differences are likely to be due to the different sorting of men and women into different fields of study. To provide additional evidence on these aspects, I complement the evidence presented in the main text of the paper with additional analysis.

First, I estimate the following linear probability model, where the dependent variable is a dummy variable which takes value 1 if the individual is employed, 0 otherwise (E):

$$\begin{aligned} E=F\delta _1 +D\delta _2 +B\delta _3 +S\delta _4 +BA\delta _5 +FOS\delta _6 +BA*{FOS}\delta _7 +A\delta _8 +e\nonumber \\ \end{aligned}$$
(B.1)

where F is a dummy variable equal to 1 if the individual is a woman, D is a vector of demographic control variables, B is a vector of individual background variables, S is a vector of family socio-economic background, BA is a dummy variable equal to 1 if the individual has completed a bachelor, FOS is a vector of field of study variables, A is a vector of the remaining academic controls, and e is the random error. \(\delta \ldots \delta _8 \) are the parameters to be estimated with OLS. Equation B.1 is estimated for the pooled sample and for men and women separately.

The results are presented in Table 23 (selected coefficients). Women have a lower probability than men of being employed, even when including several control variables. Among people who completed a master (or single-cycle) degree, graduates in Engineering enjoy the highest employment probability, while the lowest is for graduates in Law, Humanities, Modern Languages, but also Natural Sciences. For women, there is a significantly lower probability also in Agriculture and Psychology. The negative result for Medicine is discussed hereafter.

The total effects of having a bachelor degree—instead of a master—by field of study are reported in Table 24, to simplify the interpretation of the coefficients. In all fields, having a bachelor reduces the probability of being employed, with the largest drop in Pharmacy, Engineering, Architecture, Law, and Psychology. The opposite is true for Medicine: a bachelor has a positive effect on the probability of being employed compared to a master in the same field of study. Bearing in mind that Medicine includes all health-related subjects, this effect is driven by people with a degree in medical professions, such as nurses, midwifes, speech-therapists, physiotherapist, etc..., where women tend to be over-represented. On the other hand, the negative effect of a master/single-cycle degree is driven by people with a degree in Medicine strictly speaking, which requires a formal and paid speciality training after graduation.

Table 23 Employment (linear probability model): Academic and high school related coefficients, 2011
Table 24 Employment (linear probability model): Total effect of having a bachelor degree instead of a master degree, by field of study, 2011

Due to the different possible post-degree paths of not employed individual, one would like to understand whether individual characteristics explain the different outcomes, and not only the probability of being employed. To address this question, the post-degree condition can be modelled as a discrete unordered-choice model, where the choice depends on individual characteristics. The model is estimated with a multinomial logit (MNL) model.

After obtaining a degree, an individual can be in the following conditions (C):

$$\begin{aligned} C=\left\{ {\begin{array}{ll} 1 &{} \hbox {Keep on studying} \\ 2&{} \hbox {In paid training or starting a job} \\ 3&{} \hbox {Can't find a job (Unemployment)} \\ 4&{} \hbox {Personal reasons or other} \\ 5&{} \hbox {Employed [Reference]} \\ \end{array} }\right. \end{aligned}$$

The MNL model is given by:

$$\begin{aligned} ln\frac{P\left( {C=j} \right) }{P\left( {C=5} \right) }= & {} F\lambda _{j1} +D\lambda _{j2} +B\lambda _{j3} +S\lambda _{j4} +BA\lambda _{j5}\nonumber \\&+FOS\lambda _{j6} +A\lambda _{j7} \quad { for } \quad j=1,\ldots 4 \end{aligned}$$
(B.2)

where the controls are the same characteristics mentioned above. The MNL model is estimated by maximum likelihood.

Table 25 reports the results, in terms of marginal effects, for selected characteristics. Being a woman increases the likelihood of being unemployed or inactive, while it lowers the likelihood of being employed. Having a bachelor degree increases the likelihood of studying and lowers the probability of being in paid training or employed. Graduating with top grades increases the probability of being in paid training, and decreases the probability of being unemployed.

Different fields of study have a different impact on the possible outcomes. For instance, Natural Sciences, Political Sciences, Humanities, Modern Languages, and Psychology increase the unemployment probability. On the other hand, Natural Sciences, Medicine, and Law increase the likelihood of being in paid training.

Table 25 Post-degree condition: selected marginal effects from a multinomial logit model, 2011

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Piazzalunga, D. The Gender Wage Gap Among College Graduates in Italy. Ital Econ J 4, 33–90 (2018). https://doi.org/10.1007/s40797-017-0069-8

Download citation

Keywords

  • Gender wage gap
  • Oaxaca–Blinder decomposition
  • College graduates
  • Quantile decomposition
  • Field of study
  • Regional differences

JEL Classification

  • J16
  • J31
  • J71