Abstract
We analyse gender wage gaps in Italy in the mid-1990s and in the mid-2000s. In this period, important labour market developments took place and they could have had a gender asymmetric impact on wages. We identify the time trends of different components of the gender wage gap across all the wage distribution. Although the unconditional gender wage gap remained roughly constant over time, we find that the component of the gap due to different rewards of similar characteristics deteriorated women’s relative wage. We show that especially women at the centre-top of the wage distribution swam against the tide: while the trend in female qualifications slightly reduced the gender wage gap, the gender-relative trends in the wage structure significantly increased it.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In Italy the average GDP annual growth rate was around 1.5 % in this period (Eurostat).
These figures are available in Internet at http://epp.eurostat.ec.europa.eu.
More information about the ECHP and SILC is available in Internet at http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home and http://epp.eurostat.ec.europa.eu/portal/page/portal/microdata/eu_silc, respectively.
Considering that on average there are 4.345 weeks in a month, the hourly wage in the mid-1990s (mid-2000s) is computed as follows: \(w= \text{ PI111 }/(\#\,\text{ of } \text{ months } \text{ at } \text{ work }\times \text{ PE005 }\times 4.345) (w= \text{ PY010N }/(\#\,\text{ of } \text{ months } \text{ at } \text{ work }\times \text{ PL060 }\times 4.345)\)).
Employee earning is defined as the total remuneration payable by an employer to an employee in return for work done by the latter during the income reference period. The income reference period is the previous calendar or tax year. Information on earning at time \(t\) (current year) therefore refers to time \(t-1\) (income reference or previous year).
The definitions of both employment and not employment are based on self-declared economic status and therefore do not match the ILO criteria.
The deflator is the Consumer Price Index (CPI), provided by ISTAT.
For the overall sample, we also consider a dummy indicator for the presence of children younger than 12 years and the number of household components. Indeed, even if exclusion restrictions are not needed for model identification (for details see Picchio and Mussida 2011), these two variables will be exploited to explain the selection equation, but will not enter the specification of the wage distributions.
The conditional probability of full-time employment is therefore given by \(\Pr (y_t=1|z_t,\varepsilon )=\exp [-\exp (z_{t}^{\prime }\delta +\varepsilon )]\). As a consequence, an increase in a variable with a positive coefficient results in the decrease in the probability of full-time employment. The estimation results of the gompit selection equation into employment by gender and time period are reported in Table 10 and commented in Appendix A.
A piecewise constant function is constant within predefined intervals. We divided the wage support into \(J=72\) intervals \(I_j=[w_{j-1}, w_j)\), where \(j=1,\ldots ,J, w_0<w_1<\cdots <w_J, w_0=0\), and \(w_J=\infty \). \(w_1\) and \(w_{J-1}\) correspond to the first and last percentiles of the wage distribution in each time period. We chose the width of the other 70 wage baseline segments by dividing the wage support between \(w_1\) and \(w_{J-1}\) in 70 equally spaced intervals. Our choice of the number of the baseline segments is somewhat arbitrary. Nevertheless, it returns a narrow segment width (0.22€ and 0.28€ for the mid-1990s and the mid-2000s, respectively) and it is thereby suitable for flexibly approximating all possible wage hazard functions.
We tried to increase the number of support points. The decomposition analyses reported below are robust to this kind of sensitivity analysis.
Parameter estimates without sample selection are not reported in the paper, but available upon request.
As stressed by the job satisfaction literature (see e.g. Chevalier 2007), men and women might have different tastes and preferences, leading to different choices in the labour market.
The estimation results and the gender wage gap decompositions without sample selection are available upon request from the authors.
The unexplained component of the gender pay gap is often viewed as discrimination by the literature (Chevalier 2007). Discriminatory behaviours are nonetheless difficult to observe. Only very detailed information might help finding evidence of discrimination for the process under investigation, e.g. employment discrimination in hiring (Goldin and Rouse 2000).
As the reliability of these simulations depends on the capacity of our model to predict the realized wage distributions, we compute goodness-of-fit checks of the estimated model and report them in Appendix B.
Barbieri and Sestito (2008) find that in Italy temporary work increases future chances of having a “satisfactory employment” especially for women.
For example, Cockx and Picchio (2012) find that for Belgian youth, short-lived jobs are stepping stones to long-lasting jobs especially for more disadvantaged individuals, e.g. the lower educated and those living in areas where the unemployment rate is higher.
Overeducated workers have a job requiring lower qualifications than their educational level.
Under the Gompertz assumption on the distribution of the error term \(u_t\) in Eq. (1), the probability of workforce participation is indeed given by \(\exp \{-\exp (z^\prime _t \delta +\varepsilon )\}\).
References
Addabbo T, Favaro D (2011) Gender wage differentials by education in Italy. Appl Econ 43(29):4589–4605
Addabbo T, Favaro D, Magrini S (2012) Gender differences in productivity rewards: the role of human capital. Int Rev Econ 59(1):81–110
Albrecht J, van Vuuren A, Vroman S (2009) Counterfactual distributions with sample selection adjustments: econometric theory and an application to the Netherlands. Labour Econ 16(4):383–396
Barbieri P, Cutuli G (2009) Equal job, unequal pay-fixed term contracts and wage differentials in the Italian labour market. Working Paper No. 45, Department of Sociology and Social Reseach, University of Trento, Trento
Barbieri P, Sestito P (2008) Temporary workers in Italy: who are they and where they end up. Labour 22(1):127–166
Blundell R, Gosling A, Ichimura H, Meghir C (2007) Changes in the distribution of male and female wages accounting for employment composition using bounds. Econometrica 75(2):323–363
Chevalier A (2007) Education, occupation and career expectations: determinants of the gender pay gap for UK graduates. Oxf Bull Econ Stat 69(6):819–842
Cockx B, Picchio M (2012) Are short-lived jobs stepping stones to long-lasting jobs? Oxf Bull Econ Stat 74(5):646–675
Donald S, Green D, Paarsch H (2000) Differences in wage distributions between Canada and the United States: an application of a flexible estimator of distribution functions in the presence of covariates. Rev Econ Stud 67(4):609–633
European Commission (2005) The continuity of indicators during the transition between ECHP and EU-SILC. Office for the Official Publications of the European Communities, Luxembourg. http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-CC-05-006/EN/KS-CC-05-006-EN.PDF
Goldin C, Rouse C (2000) Orchestrating impartiality: the impact of “blind” auditions on female musicians. Am Econ Rev 90(4):715–741
Heckman J, Singer B (1984) A method for minimizing the impact of distributional assumptions in econometric models for duration data. Econometrica 52(2):271–320
ISTAT (2005) Rapporto annuale. La situazione del paese nel 2005. ISTAT, Rome
ISTAT (2009) Rapporto annuale. La situazione del paese nel 2009. ISTAT, Rome
Kugler A, Pica G (2008) Effects of employment protection on worker and job flows: evidence from the 1990 Italian reform. Labour Econ 15(1):78–95
Lazear E (1990) Job security provisions and employment. Q J Econ 105(3):699–726
Mulligan C, Rubinstein Y (2008) Selection, investment, and women’s relative wages over time. Q J Econ 123(3):1061–1110
Niederle M, Vesterlund L (2007) Do women shy away from competition? Do men compete too much? Q J Econ 122(3):1067–1101
Olivetti C, Petrongolo B (2008) Unequal pay or unequal employment? A cross-country analysis of gender gaps. J Labor Econ 26(4):621–654
Picchio M, Mussida C (2011) Gender wage gap: a semi-parametric approach with sample selection correction. Labour Econ 18(4):564–578
Rustichelli E (2005) I differenziali retributivi di genere. Quaderni Spinn, Ministero del Lavoro, pp 79–106
Schindler M (2009) The Italian labor market: recent trends. Institutions and reform options, IMF Working Paper No. 47, International Monetary Fund
Sciulli D (2006a) Labour market transitions in Italy: the effects of the Treu law on flows from an atypical contract. CEIS Working Paper No. 232, Tor Vergata University, Rome
Sciulli D (2006b) Making the Italian labour market more flexible: an evaluation of the Treu reform. Economics Working Paper No. 34, Department of Economics, University Carlos III, Madrid
Author information
Authors and Affiliations
Corresponding author
Additional information
Data from the European Community Household Panel Survey 1994–1997 are used with the permission of Eurostat (contract ECHP/2010/16). The results and conclusions are those of the authors and not those of Eurostat, the European Commission or any of the authorities whose data have been used. The authors wish to thank the participants to the XXVI National Conference in Labour Economics in Milan (09/2011), to the EALE conference in Cyprus (09/2011), and to the seminars at Tinbergen Institute (11/2011) and Ghent University (12/2011) for their comments and suggestions. Matteo Picchio acknowledges the financial support by the Research Foundation-Flanders (FWO), Belgium, and by Stichting Instituut GAK, through Reflect, the Research Institute for Flexicurity, Labour Market Dynamics and Social Cohesion at Tilburg University.
Appendices
Appendix A: Nonrandom selection into employment
Table 9 displays summary statistics of the covariates used to estimate the model for the probability of employment. Table 10 reports the estimation results of the discrete mixture gompit model for the probability of employment by gender and time periods. Due to the gompit specification, regressors with positive estimated coefficients have a negative impact on the probability of being in the workforce.Footnote 23 The results are in line with the expectations. The employment probability decreases with age, but increases with potential experience. Higher educated people are more likely to be at work. While in the mid-1990s the employment probability was the highest in the north-east, followed by the north-west and the centre, and the lowest in the south, in the mid-2000s the north and the centre shared more similar employment probabilities. Family structure and married status have opposite effects on work participation between men and women, strongly consistent with the male breadwinner system: married (wo)men have a (lower) higher probability of being employed; for (wo)men work participation is higher (lower) if there are kids in the household. Finally, the number of household members and a bad/fair health condition reduce the employment probability both for men and women.
Appendix B: Goodness of fit
In this Appendix, we check the goodness of fit of the model by contrasting empirical aspects of the data with those predicted by model simulations. Our econometric model is characterized by a mixture of parametric and non-parametric assumptions and the counterfactual exercises in Sect. 3 are based on simulations. It is therefore important to assess the ability of the model to provide quantitative predictions of the statistics of primary interest, i.e. the gender wage gap in the mid-1990s, the gender wage gap in the mid-2000s and its change over time. Predictions of the wage distributions per each time period and each gender are computed by implementing the simulation algorithm in Picchio and Mussida (2011) and are exploited to derive the predicted gender wage gaps and their change over time. We verify the goodness of fit by checking whether the actual gender pay gaps and their variation over time lie within the confidence intervals of the simulated ones.
The top panel of Table 11 and the top graph of Fig. 2 report the goodness of fit of the model in predicting the gender wage gap in the mid-1990s using ECHP data. The panel in the middle of Table 11 and the graph in the middle of Fig. 2 focus on the goodness of fit of the model in predicting the gender wage gap in the mid-2000s using SILC data. Finally, at the bottom of Table 11 and Fig. 2, we contrast the actual variation and the predicted variation over time of the gender wage gap. We check thereby the goodness of the model in fitting the change over time in the gender wage gap, i.e. the statistic of primary interest in this study.
Goodness of fit: the gender wage gap and its variation over time. The grey areas are Monte Carlo 95 % confidence intervals, computed by 999 replications
The model perfectly fits the gender wage gaps observed in the mid-2000. Nevertheless, it shows some problems in fitting the gender wage gaps in the mid-1990s: the fit is fine until the \(75\mathrm{th}\) percentile of the wage distribution; thereafter the model somewhat systematically overpredicts the gender wage gap, especially at the \(95\mathrm{th}\) percentile. However, the poor ability of the model in fitting gender wage gaps in the mid-1990s is limited to a small segment of the wage support. Finally, the bottom panel of Table 11 and the bottom graph of Fig. 2 show that the model predicts very well the change over time in the gender wage gap. Only at the very top of the wage distribution, the actual change lies outside the confidence interval of the simulated one: as a consequence of the overprediction of the gender wage gap at the top of the distribution in the mid-1990s, the variation in the gender wage gap is underpredicted in that region.
Rights and permissions
About this article
Cite this article
Mussida, C., Picchio, M. The trend over time of the gender wage gap in Italy. Empir Econ 46, 1081–1110 (2014). https://doi.org/10.1007/s00181-013-0710-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-013-0710-9