Unfair Tournaments

Abstract

This paper addresses the gender pay gap among Italian university graduates on entry to the labor market, and stresses the potential for gender stereotypes to impact subjective assessment of individual productivity. We build upon previous research about gender and wage inequality, introducing tournament theory as a framework for the gender pay gap analysis. We hypothesize that the effects of gender make occupational tournaments less fair in some arenas compared with others. As a consequence, men workers have higher probabilities of winning the wage competition, but this process is uneven. Our data show that in contexts where stereotypes are most likely to occur, tournaments appear to be less fair and the unexplained component of the gender pay gap is higher.

Keywords

labor market Italy gender pay gap stereotypes tournaments

In Italy, equal pay and gender discrimination legislation has been in place since 1991. New entrants to the labor market in the mid-2000s grew up in a society that encouraged them to take equal opportunities for granted. No employer would deny, in principle, employees’ rights to be evaluated according to their actual productivity. As a consequence, it is tempting to believe that discrimination is a thing of the past, currently carried out only by a small set of uninformed people (Magnusson 2010). On the contrary, a large body of research has reported evidence that women are systematically rated as performing worse than men in the workplace, even after controlling for ability and experience (Sackett, Dubois, and Noe 1991). We hypothesize that the assessment of productivity is strongly influenced by stereotypes,¹ beliefs, or expectations that stem from the distribution of men and women into social roles and affect our judgments of others (Eagly and Steffen 1984). Stereotypes and prejudice may preclude the fair assessment of individual performance and create workplace discrimination. Hence, even today, gender seems to shape individuals’ preferences and choices and lead to systematic underestimation of women’s productivity and rewards.

This paper explores gender pay disparities among Italian university graduates and focuses on wage discrimination variation we can observe in settings where gender stereotypes are more or less prevalent, making the rules of competition between genders more or less likely to be fair. We add to the psychological and sociological literature regarding women’s discrimination in the labor market by introducing tournament theory as a framework for the analysis of the gender pay gap. We argue that gender stereotypes may be reflected in unfair occupational tournaments, and associated with higher levels of gender discrimination. In particular, the unexplained component of the gender pay gap is a measure of wage discrimination. Tournaments are competitive mechanisms that rank players, that is, assess their relative performance. If tournaments are unfair based on the influence of gender stereotypes, the probabilities of winning differ by gender, all else being equal. The greater the influence of stereotypes, the lower women’s probability of winning, and the wider the unexplained gender pay gap.

In the next section, we summarize the main explanations for gender disparities in pay, separating those focusing on the choices made by the women (supply-side explanations) from the ones focusing on the choices made by employers (demand-side explanations). All of them play a role in tournament theory, influencing both women’s decisions to participate and women’s probability of winning the tournament. Supply-side choices reflect the influence of gender stereotypes on women’s decisions to participate the tournament. Niederle and Vesterlund (2007) establish that women enter in tournaments less than men because of gender differences in confidence and in attitudes towards competition. The gender gap in tournament entry is a way to explain the gender pay gap. Another way is to analyze the choices from the demand side, that is, the effects of stereotypes on the women’s performance appraisal (Valian 1998). As the usual methodology to analyze supply-side preferences consists of survey questions not included in our data set, we focus on demand-side discrimination arising in the gendered assessment of productivity. In particular, we identify specific environments in which the use of stereotypes is expected to be more or less likely to exert an influence on performance appraisal, and we hypothesize that the unexplained component of the gender pay gap increases or decreases in line with the expected influence of these stereotypes. We provide empirical support for our hypotheses using a unique data set provided by the Italian Institute of Statistics (ISTAT) in which we examine college-to-work transitions three years after graduation.

Traditional Explanations for Gender Disparities in Pay and the Role of Stereotypes

Social science scholars have analyzed the sources of the gender pay gap from different perspectives. In particular, labor economists have emphasized two broad sets of explanations: explanations focused on the supply side of the labor market, such as the effect of individual preferences and characteristics on pay, and explanations focused on the demand side, such as the effect of occupation and job characteristics on pay. Traditionally, the first set of explanations focuses on the choices made by women, while the second set of explanations focuses on job-related constraints faced by women. However, these two sets of explanations are not mutually exclusive; they both play a role in explaining the gender pay gap. In the next two sub-sections we examine the influence of gender stereotypes on both women’s (supply-side) and employers’ (demand-side) decisions.

Supply-Side Explanations

A preference-based explanation of women labor choices posits that gender differences in career path and earnings derive largely from gender differences in preferences (Hakim 2000). However, it is worth considering whether gender differences in preferences could be constrained by social norms too. Indeed, gender stereotypes may shape employment preferences and make men and women choose different types of jobs and different career paths. Thus, revealed preferences (including career aspirations) are shaped by individuals’ assessments of their likely success in a given activity, and individuals adjust their activities in light of the gendered expectations that others have of them (Correll 2001, 2004; Ridgeway 2009; Ridgeway and Correll 2004; Winslow 2010).

Human capital theory explains women’s lower wages with gender differences in personal characteristics that affect individuals’ productivity. Mincer and Polachek (1974) have argued that anticipated family responsibilities influence women’s decisions about both the number of years of schooling and the field of study, as well as the length of time devoted to market activities. Hence, the lower incentives of women to invest in human capital reduce their productivity and their earnings compared with men. However, over the past three decades women have made substantial progress in educational outcomes, and this trend is expected to widen further (Blau and Kahn 2000, 2007; Bobbitt-Zeher 2007). While in the past men typically had better access to university-level institutions, nowadays female graduates exceed the number of male graduates, and on average female students outperform male students in academic achievements in most OECD countries (OECD 2009). Unfortunately, despite women’s progress in higher education, the belief that women and men are suited for different fields of study and professions is still persistent and pervasive even among individuals participating in higher education. Women continue to prefer majors traditionally dominated by women, and men are not likely to choose faculties with high proportions of women in significant numbers (England and Li 2006; Reskin and Bielby 2005).

Demand-Side Explanations

Gender inequality in wages might also be due to differences in working conditions (England and Folbre 2005). If female-dominated occupations enjoy nonpecuniary benefits (as working part-time), which make it easier to combine work and family life, these benefits may result in lower wages. Solberg and Laughlin (1995) point out that any measure of earnings that excludes these benefits may produce misleading results as to the existence of gender discrimination.

An alternative explanation for the gender income differences is discrimination against women, that is, employers’ gender-biased decisions on the allocation of individuals across and within occupations. Mainstream economic theory describes two related aspects of the social division of labor that are important for gender discrimination. The first aspect is related to the horizontal division of labor (diversity of professions at the same hierarchical level). The second aspect is related to the vertical division of labor (assignment of heterogeneous agents to different hierarchical levels). In feminist economic literature, these two aspects are strictly related to the theory of the gender segregation (Bielby and Baron 1986; Hakim 2000). A substantial body of research shows that both the possibility of entering an occupation and access to promotion within occupations differs between men and women (Anker 1998). Especially notable is the case for statistical discrimination (Arrow 1972; Stiglitz 1973), which occurs when employers make hiring and promoting decisions based not on an individual’s personal characteristics but on the average productivity of the individual’s gender. For example, according to this view, although not all women have children or quit, employers may make their hiring decisions on the basis of a higher statistical probability for women to quit (England 1992). Even more troubling is the fact that female-dominated occupations have lower wages than male-dominated occupations despite comparable qualifications (England 2005; England 2010; England and Folbre 2005). Proponents of the comparable worth view show how skills, effort, responsibilities, and working conditions in female jobs have been overlooked in evaluation processes (England 1992; Steinberg 1990).

Tournament Theory

We introduce tournament theory (Lazear and Rosen 1981) in the feminist social science literature as a framework for the analysis of gender disparities in pay. Labor tournaments are competitive mechanisms that define the ranking of employees/players. They assign employees to a particular level in the hierarchy and award to them the remuneration that corresponds to it. Players are ranked according to their assessed relative performance. This performance, in turn, depends on both talent and effort. Effort being equal, the most talented individual will win, thereby ensuring that greater talent commands greater resources (Lazear 1998; Rosen 1982). Thus, tournament theory emphasizes the role of talent as the main determinant of the final rank, and regards competition among agents as an efficient tool to match individuals with jobs in hierarchical organizations (Lazear and Rosen 1981). It is worth noting, however, that the efficiency of this outcome depends on the symmetry of the tournament (O’Keeffe, Viscusi, and Zeckhauser 1984). Symmetric tournaments occur when agents are homogeneous and are treated equally by the rules of the competition (Schotter and Weigelt 1992). Asymmetric tournaments may be uneven or unfair: they are uneven when agents have different cost-of-effort functions; they are unfair when agents are identical, but yet the rules of the competition favor one of them and discriminate the other. In particular, in unfair tournaments the probabilities of winning the contest can differ by gender, all else equal.

We assume that because of the influence of gender stereotypes, the “rules” of the competition favor male contestants and discriminate against female participants. In settings where gender stereotypes are more prevalent, women will have a lower probability of winning the contest prize, and the gender pay gap will be wider. In particular, women with the same characteristics as men (same education, work experience, and so on) will not have the same chance of winning the job competition, so they are not matched to the position they deserve and their talent is underutilized.

Effect of Stereotypes on Labor Tournaments

In this section, we identify different types of employment/jobs to test tournament theory. In particular, we expect a lower unexplained component of the gender pay gap in occupational contexts where the tournament is fairer. However, in occupational contexts where stereotypes are more prominent, we expect a higher unexplained component of the gender pay gap.

The first environment we consider is one where there are simply no tournaments (self-employed workers) compared to a context in which tournaments are held (wage and salary workers). As the self-employed employ themselves, and they know their own productivity without any kind of assessment, there is neither performance evaluation nor any ranking of contestants, and therefore, stereotypes are less likely to exert an influence. As a consequence, we expect a lower unexplained gender pay gap among self-employed workers than among employees (Moore 1983).

Then, we identify low-profile clerical jobs, as a setting in which tournaments are more likely to be unfair because the assessment of individual productivity is usually inaccurate, thereby allowing stereotypes to exert an influence on performance appraisal. In particular, in lower-skilled occupations with basic professional knowledge and routine tasks, the perceiver is not motivated to make accurate judgments, and the criteria used to assess individual productivity remain often unspecified. In the absence of concrete criteria, expectations based on stereotypes tend to dominate the judgment process (Heilman 2001; Nieva and Gutek 1980). Temporary work represents another situation in which stereotypes are more likely to occur because the employer is less motivated to make accurate judgments. Firms may use fixed-term contracts as a probationary stage during which they can observe individual performance (Booth, Francesconi, and Frank 2002; Loh 1994; Wang and Weiss 1998). In this case, the evaluation of productivity, which is expensive, is less important because the contract will expire. Therefore, estimates of productivity are less accurate and more superficial, and leave room for the stereotype to exert an influence. In both these conditions, evaluators may “safely” rely on their stereotypes when deciding whom to hire or promote, or what an appropriate salary increase will be. As a consequence, we expect that the unexplained component of the gender pay gap is higher in low-level occupations than it is in intellectual professions and highly specialized occupations. We also expect that the unexplained component of the gender pay gap is greater among employees hired under fixed-term contracts than it is among employees hired under permanent contracts.

Finally, we analyze the gender pay gap when the recruitment takes place through open competition. In this case, tournaments are more likely to be fair because the assessors are required to provide justifications for their choices, and must use objective criteria and structured evaluation procedures, thereby increasing the incentive for a more accurate assessment of individual performance. The stereotyping literature actually indicates that ambiguity in evaluative criteria and a lack of structure in the evaluation process can create the conditions for gender stereotypes to flourish (Heilman 2001; Heilman and Okimoto 2008; Welle and Heilman 2007). Indeed, Tetlock and Kim (1987) and Dobbs and Crano (2001) find that people show greater accuracy in productivity assessments when they anticipate having to justify their ratings. As a consequence, where decision makers are required to justify their choices and describe the criteria they use to evaluate candidates, as in open competition, they are less likely to discriminate against women. In open competitions, the recruitment procedure is a combination of examinations, scrutiny of the curriculum and qualifications, in which the information cannot easily be distorted to fit the stereotypes. Also, signaling and labeling incontrovertibly their own excellence at school (Spence 1973) may reduce ambiguity in personnel evaluation and curtail the unfairness of the tournament. Thus, we expect that the unexplained component of the gender pay gap is lower among employees hired through open competition than it is among those hired without open competition. In order to cut down the conditions for gender stereotypes to flourish, some women may self-promote and make explicitly clear that they are exceptionally qualified candidates and top performers in their field. Sorting models of education suggest that education is often used to draw inferences about unobserved characteristics of individuals (Spence 1973). If the abilities that are correlated with educational performance positively affect productivity on the job, the degree score may be a good signal of a worker’s productivity. Therefore, we expect that the unexplained component of the gender pay gap drops among graduates achieving the maximum degree score.

To summarize, the above considerations lead to the following predictions:

Hypothesis 1: The unexplained component of the gender pay gap is lower when the assessment of productivity is unnecessary, that is, there are no tournaments.

Hypothesis 2: The unexplained component of the gender pay gap increases when tournaments are more likely to be unfair (as in competitions for low-status positions and fixed-term contracts) and decreases when tournaments are more likely to be fair (as in competitions for high-status positions and open competitions).

Hypothesis 3: The unexplained component of the gender pay gap is lower when the assessment of productivity may be roughly inferred from excellent educational performance.

Data, Measures, and Method

The data source of this paper is the Survey on the Early Career of College Graduates (SECCG, “Indagine sull’inserimento professionale dei laureati”) conducted by ISTAT every three years. In particular, we analyze the individuals who graduated in 2004 and were interviewed in 2007, which represents the last survey available when the article was written. The graduate population of 2004 consisted of 167,886 individuals (68,939 males and 98,947 females). In the first semester of the survey year, ISTAT extracts a random sample from the universe of the individuals graduating in that year, stratified on the basis of gender, faculty, and university. In the second semester, ISTAT interviews the sampled individuals by the CATI (computer-assisted telephone interview) interviewing technique, double-checking all answers with universities’ administrative records. The response rate was about 69.5 percent, yielding a data set containing information on 26,570 graduates.²

The surveys collect information on (1) University Career and High School Background, including kind of high school attended, high school grades, other education, university, subject, duration, degree score, accommodation, work during university, postgraduate studies; (2) Work Experience, including previous experience, experience in actual work, type of work, net monthly wage; (3) Search for Work, including kind of work desired, willingness to work abroad, preference over working hours, minimum net monthly wage required; (4) Family Information, including parents’ work, parents’ education level, brothers and/or sisters; and (5) Personal Characteristics, including date of birth, gender, marital status, children, country of domicile, country of birth, residence.

Among the Italian surveys currently available, the SECCG offers the most precise and detailed information on demographic characteristics, college attended, ability, family background, current employment, and income of recently graduated individuals. Our sample, consisting of 25,408 individuals (11,909 males and 13,499 females), was obtained after excluding individuals with missing values for gender, university career, and labor market outcomes. Table 1 presents descriptive statistics for the sample whereas Table A1 in the appendix contains details of each variable used in the empirical analysis.

Table 1:

Descriptive Statistics for Selected Variables used in the Analysis

	Women		Men
Variable	Mean	Standard Deviation	Mean	Standard Deviation
Number of observations	13,499		11,909
Wage	7.110	0.230	7.227	0.258
Hours worked	3.687	0.082	3.710	0.084
High school exam score	50.585	6.997	48.751	7.253
Educational performance	90.903	17.438	88.583	16.528
Experience	1.964	2.368	2.358	2.402
Experience squared	9.465	14.615	11.326	15.196
Degree subject
Sciences	0.023	0.149	0.043	0.200
Pharmacy	0.053	0.221	0.046	0.207
Natural sciences	0.048	0.210	0.039	0.191
Engineering	0.047	0.209	0.188	0.386
Architecture	0.050	0.214	0.057	0.228
Agricultural studies	0.025	0.155	0.029	0.165
Economics, business, and statistics	0.142	0.344	0.152	0.354
Political sciences and sociology	0.091	0.283	0.064	0.241
Law	0.123	0.324	0.102	0.298
Humanities	0.060	0.233	0.036	0.183
Foreign languages	0.052	0.219	0.014	0.115
Teachers Education	0.046	0.206	0.015	0.118
Psychology	0.023	0.147	0.020	0.137
Medicine	0.217	0.407	0.194	0.390
University of Northern Italy	0.454	0.498	0.471	0.499
University of Central Italy	0.259	0.438	0.275	0.446
Private university	0.062	0.241	0.050	0.219
General high school	0.696	0.460	0.655	0.475
Previously entered another degree course	0.121	0.327	0.151	0.358
Studied in hometown	0.365	0.481	0.383	0.486
Moved to attend university	0.225	0.417	0.225	0.418
Attended private lessons	0.025	0.156	0.030	0.171
Attended student exchange program	0.086	0.280	0.094	0.292
Working student	0.607	0.488	0.638	0.481
Married	0.318	0.466	0.210	0.407
Children	0.110	0.313	0.074	0.262
Father college degree	0.243	0.429	0.284	0.451
Father secondary degree	0.375	0.484	0.367	0.482
Mother college degree	0.199	0.399	0.216	0.412
Mother secondary degree	0.382	0.486	0.381	0.486
Father’s occupation
Father employed	0.956	0.206	0.960	0.196
Father self-employed	0.280	0.449	0.258	0.438
Father manager	0.087	0.282	0.106	0.307
Father in intermediate position (between manager and clerk)	0.115	0.319	0.140	0.347
Father white collar, clerk	0.197	0.398	0.193	0.394
Mother’s occupation
Mother employed	0.551	0.497	0.546	0.498
Mother self-employed	0.086	0.280	0.076	0.266
Mother manager	0.012	0.110	0.013	0.115
Mother in intermediate position (between manager and clerk)	0.110	0.312	0.124	0.330
Mother white collar, clerk	0.210	0.407	0.212	0.409
Employed	0.636	0.481	0.735	0.441
Firm size	0.069	0.254	0.039	0.194
Industrial sector	0.218	0.413	0.351	0.477
Paid training	0.027	0.163	0.026	0.159

The dependent variable, earnings, was measured as monthly earnings in 2007 in euros and net of taxes and social security contributions. Table 2 reports average monthly earnings and employment probability 3 years after graduation by gender and field of study. The average earnings were €1,299.00 and €1,081.00 per month for the male and the female subsample, respectively. The average earning difference between male and female students is statistically significant for all of the subjects studied, as indicated by the test in Table 2. Approximately 74 percent of men and 64 percent of women were employed 3 years after graduation. Therefore, on average, male graduates earned about 20 percent more than females and were more likely to have a job 3 years after graduation.

Table 2:

Average Earning and Employment Probability by Gender and Field of Study

	Average Monthly Earning (€)			Average Employment Probability
Field of Study	Men	Women	T Statistic	Men	Women
Sciences	1,252.36	1,065.03	−4.03***	0.69	0.66
Pharmacy	1,280.79	1,137.91	−4.35***	0.74	0.76
Natural sciences	1,232.25	1,062.48	−4.04***	0.65	0.59
Medicine	1,468.22	1,234.35	−4.24***	0.45	0.27
Engineering	1,391.70	1,287.06	−3.85***	0.92	0.83
Architecture	1,221.35	1,054.29	−4.15***	0.87	0.82
Agricultural studies	1,141.59	905.72	−4.06***	0.77	0.7
Economics, business, and statistics	1,349.92	1,169.86	−9.09***	0.83	0.77
Political science and sociology	1,300.48	1,096.71	−7.68***	0.78	0.82
Law	1,172.35	1,018.93	−4.63***	0.6	0.51
Humanities	1,107	948.09	−4.19***	0.69	0.75
Foreign languages	1,204.67	1,048.28	−3.3***	0.85	0.8
Education	1,062.94	961.70	−2.65***	0.81	0.79
Psychology	1,078.69	832.67	−4.88***	0.72	0.7
Health	1,098.13	882.75	−4.75***	0.78	0.74
Total	1,299.28	1,080.96	−23.34***	0.74	0.64

NOTE: The fourth column reports the values of the T statistic for the null hypothesis that the difference between the monthly earning is zero.

***

p < 0.01, **p < 0.05, *p < 0.10.

The first step of our empirical analysis consists of estimating the earnings equation for male and female samples. The presence of interval income data prevents us from working with the hourly wage rate and therefore we restrict the sample to full-time workers only. We define full-time workers as those who worked more than 30 hours per week. This restriction leads to a sample of 5,392 men and 4,503 women full-time workers. In the following, we examine the earnings differential between different groups of employed individuals: self-employed versus employees, permanent contracts versus fixed-term contracts, overeducated versus not overeducated, and so on. We defined a worker as being overeducated if her or his educational level exceeds the minimal required education to do his or her job. Table 3 shows that within each group the number of observations by gender is similar, reflecting the stratified sampling method followed by ISTAT. The sample size of every group considered is large enough so that accurate estimates of the earning equations can be performed.

Table 3:

Number of Observations by Groups of Employed Individuals

	Women	Men
Employees	3,274	3,709
Self-employed	699	1,233
Public sector	1,087	961
Private sector	2,156	2,470
Permanent contracts	2,176	2,789
Fixed-term contracts	1,629	1,373
Graduate degree required	1,344	1,434
Graduate degree not required	3,160	3,963
Recruitment through open competition	334	297
Recruitment without open competition	4,169	5,095
Intellectual professions, scientific and highly specialized occupations	1,816	2,604
Intermediate professions (technical)	1,730	2,110
Professions relating to low-profile clerical jobs	756	363

There has been much debate about what variables one should enter into the earnings functions used in studies of the gender wage differential (Altonji and Blank 1991). The standard Mincer equation including experience and schooling is typically augmented by factors such as human capital, employment, and personal and family background characteristics (Prokos and Padavic 2005). We consider as human capital variables the following: broad major, the score earned in the high school graduation exam, dummy variables for the kind of high school attended, degree subject, and private or public university. To ease the effect of potential endogeneity in the earning equation, we measure the unobservable ability by means of the degree performance. In particular, our measure of degree performance combines information on the degree score and the speed at which students complete their academic career. This variable is called Edperf and it is calculated as the product of the final degree score and the inverse of the degree completion time. Formally, [Edperf =(dscore)/(1 + 0.1 years)] where dscore is the degree score and years is the number of years used to get the degree. The degree scores have been normalized to take into account that different majors might employ different marking scales and have different durations. The upper bound limit of Edperf is set to 113, which corresponds to honors degree with no delay in completion.³ The employment variables include work experience as well as its square as an indicator of the diminishing marginal utility of the work experience, dummy variables for employment sector and job characteristics (public vs. private sector, firm size, dummy for the employment sector of the graduate) when appropriate, a dummy variable for the size of the firm where the respondent works, and the natural log of hours worked per week. Family background characteristics include mother’s and father’s education and employment status when the individual was 14 years of age. Personal characteristics include family status and gender (in models with both men and women). In order to capture the impact of differences in regional wages, we include dummies for the region where the current job is located in the earning equation and dummies for the region of residence in the selection equation.⁴ By considering close to 65 controls we expect to explain earnings as well as selection into employment and therefore we expect to capture the explained and the unexplained components of the difference in mean outcomes, that is, the gender wage gap.

In order to measure gender wage differentials, we estimate separately the wage equation for men and women and then we use these equations to decompose the gross differential in explained and unexplained components by means of the standard Oaxaca-Blinder decomposition. To analyze the wage equation, we need a model for selection bias. We estimate the sample selection model by means of the Heckman (1979) two-step procedure.

We first consistently estimate the selection equations, binary choice–type equations, where the binary variable simply indicates working or not working. The estimation is conducted by means of probit maximum likelihood. We then use the estimation results of the first stage to consistently estimate by ordinary least squares the linear earning equations:

w = X β + λ α + ε,

where w is the natural log of the net monthly wage and λ is the estimate of the inverse Mills ratio constructed from the first-stage probit model. With the inverse Mills ratio included, the coefficients on the X, β, represent consistent estimates of the parameters of the wage equation. Such estimation takes into account the possibility that individuals may select a particular employment status for themselves because they have a comparative advantage. Following much of the existing literature on the gender wage gap, we use a dummy variable recording whether the individual has at least one child as the identifying variable for the Heckman (1979) procedure. Finally, from the separate regression analyses by gender, we compute the gender difference in earnings and their Oaxaca-Blinder decomposition.

The procedure illustrated above (estimation and Oaxaca-Blinder decomposition) was then used for all the other groups considered.⁵ In particular, we estimate the sample selection model depending on the categories underlined. For instance, when we consider the wage gap between public and private sector (second line of Table 8), the selection equation indicates the choice of working people with respect to the activity sector, that is, working in the public sector or working in the private sector. Moreover, we adopt the same identifying variable, children, for the Heckman two-step procedure and we include the Heckman correction term only if there is a significant selection bias, that is, “lambda” is statistically significant.

Results

We present detailed estimation results only for the decomposition of the gender wage gap among the employees. Table 4 reports results from estimating gender-specific earnings equations controlled for self-selection, whereas the results of the first-stage probit model are presented in Table 5. Table 4 shows that the block of educational variables displays expected results. The educational performance has a positive impact on earnings, and the type of degree is statistically significant in explaining the earnings. We observe that the amount of experience and its square have no impact on wages. This may be due to the fact that we are considering a proxy of a first entry in the labor market and, therefore, the effect of work experience on earnings may be negligible. The significance of lambda in Table 4 confirms the selectivity bias for both men and women.

Table 4:

Ordinary Least Squares Estimation Results of the Earnings Equation for Employees

	Women		Men
Variable	Coefficient	t Value	Coefficient	t Value
Constant	5.640	30.484	5.810	31.007
Educational performance	0.001	4.122	0.001	5.115
Inverse Mills Ratio	0.384	4.929	0.185	2.937
Experience	−0.010	−1.011	0.012	1.076
Experience squared	0.001	1.235	−0.001	−0.759
Degree subject
Sciences	0.108	2.508	0.135	3.421
Pharmacy	0.227	5.629	0.197	5.184
Natural sciences	0.065	1.532	0.128	2.998
Engineering	0.288	6.752	0.243	6.278
Architecture	0.108	2.456	0.133	3.122
Agricultural studies	0.079	1.751	0.090	2.037
Economics, business, and statistics	0.169	4.409	0.186	5.102
Political sciences and sociology	0.119	2.967	0.100	2.595
Law	0.008	0.186	0.117	2.860
Humanities	0.047	1.128	−0.018	−0.404
Foreign languages	0.083	2.035	0.035	0.739
Education	0.031	0.716	0.065	1.265
Psychology	0.012	0.242	0.090	1.780
Medicine	−0.061	−0.935	0.168	3.514
University of Northern Italy	−0.092	−6.457	−0.028	−1.880
University of Central Italy	−0.027	−1.825	0.033	2.212
Private university	0.022	1.560	0.041	2.307
General high school	−0.009	−0.940	0.008	0.846
Previously entered another degree course	−0.009	−0.672	0.014	1.025
Studied in hometown	−0.001	−0.123	0.003	0.354
Moved to attend university	0.014	1.497	0.021	1.963
Attended private lessons	0.015	0.685	0.001	0.050
Student exchange program	0.032	2.748	0.050	3.842
Working student	0.092	6.478	0.047	4.048
Paid training	−0.086	−5.445	−0.061	−3.985
Married	0.003	0.416	0.075	6.354
Father college degree	0.030	2.210	0.015	1.042
Father secondary degree	0.022	2.239	0.004	0.411
Mother college degree	−0.001	−0.101	0.006	0.376
Mother secondary degree	−0.008	−0.772	0.009	0.875
Father’s occupation
Father in work	−0.002	−0.108	−0.005	−0.234
Father self-employed	0.022	2.165	0.021	1.953
Father manager	0.010	0.629	0.028	1.772
Father in intermediate position (between manager and clerk)	−0.006	−0.434	0.030	2.023
Father white collar, clerk	−0.009	−0.784	−0.002	−0.162
Mother’s occupation
Mother manager	0.014	0.968	−0.019	−1.252
Mother white collar, clerk	0.025	2.420	0.008	0.714
Firm size	0.101	5.873	0.084	4.053
Industrial sector	0.028	3.089	0.041	4.544
Hours worked	0.345	7.275	0.317	6.505
Regional dummies	X		X
Number of observations	3,274		3,709
Adjusted R²	0.185		0.164
F	12.807 (0.00)		12.566(0.00)
Average wage women (ln)	7.104
Average wage men (ln)	7.227

Table 5:

Probit Estimation Results of the Employment Probabilities for Employees

	Women		Men
Variable	Coefficient	t Value	Coefficient	t Value
Constant	0.643	3.860	0.425	2.710
Educational performance	−0.001	−0.430	0.001	1.600
Degree subject
Sciences	−0.191	−1.930	0.003	0.030
Pharmacy	0.004	0.040	0.316	3.560
Natural sciences	−0.309	−3.130	−0.214	−2.470
Engineering	0.736	8.380	0.551	5.830
Architecture	0.384	3.790	0.477	5.130
Agricultural studies	0.130	1.170	0.139	1.350
Economics, business, and statistics	0.250	2.940	0.230	2.980
Political sciences and sociology	0.128	1.360	0.398	4.840
Law	−0.289	−3.320	−0.386	−4.950
Humanities	−0.165	−1.610	0.227	2.640
Foreign languages	0.185	1.240	0.286	3.150
Education	0.415	2.640	0.426	4.570
Psychology	0.016	0.130	0.101	0.960
Medicine	−0.593	−7.250	−0.921	-12.260
University of Northern Italy	0.048	0.890	−0.042	−0.840
University of Central Italy	0.048	0.970	−0.042	−0.910
General high school	−0.061	−1.830	−0.117	−3.850
Attended private lessons	0.089	1.060	0.252	3.040
Student exchange program	−0.034	−0.700	0.013	0.270
Working graduate	0.360	12.070	0.337	12.680
Married	0.304	7.250	0.026	0.860
Children	0.207	3.020	−0.236	−5.370
Father college degree	−0.048	−0.950	0.034	0.760
Father secondary degree	0.034	0.860	0.058	1.700
Mother college degree	0.020	0.370	−0.012	−0.240
Mother secondary degree	0.031	0.800	−0.002	−0.050
Father’s occupation
Father in work	0.067	0.950	0.009	0.140
Father self-employed	−0.003	−0.090	0.091	2.770
Father manager	−0.007	−0.120	−0.013	−0.240
Father in intermediate position (between manager and clerk)	−0.057	−1.100	0.023	0.470
Father white collar, clerk	−0.017	−0.380	0.036	0.950
Mother’s occupation
Mother manager	−0.018	−0.350	−0.029	−0.600
Mother white collar, clerk	−0.019	−0.490	−0.022	−0.640
Regional dummies	X		X
Number of observations	13,499		11,909
Percentage correctly predicted	78.35		74.67
Madalla’s pseudo-R²	0.207		0.234

Then, from the separate regression analyses by gender, we compute the gender difference in earnings and their Oaxaca-Blinder decomposition.⁶ The first line of Table 6 reports the decomposition for the gender difference among employees. First of all, we observe that the raw gender pay gap among graduate employees at the beginning of their career is significant and amounts to about 11 percent. Moreover, only about 12 percent of the gender pay gap can be explained by differences in average observed characteristics. The remaining 88 percent is not explained by the large amount of observed individual characteristics we control for, and it is attributable to gender differences in economic returns to individual characteristics.⁷

Table 6:

Gender Pay Gap in Different Groups

	Percentage of Raw Pay Gap	T Statistic	Percentage of Gap Explained by Observed Characteristics	Percentage of Gap Unexplained by Observed Characteristics
Employees	11.05	20.66***	12.23	87.77
Self-employed	12.13	6.19***	27.83	72.16
Intellectual professions, scientific, and highly specialized occupations	11.44	11.07***	44.39	55.60
Professions relating to low-profile clerical jobs	9.08	7.06***	21.80	78.19
Permanent contracts	12.70	6.86***	33.06	66.94
Fixed-term contracts	7.18	21.52***	19.24	80.76
Recruitment through open competition	12.12	7.93***	39.64	60.36
Recruitment without open competition	12.09	21.89***	33.99	66.01
Graduates with excellent educational performance (edperf = 113)	14.32	4.37***	48.07	51.92
Total graduates	11.55	19.45***	27.03	72.98

NOTE: The third column reports the values of the T statistic for the null hypothesis that the difference between the monthly earning is zero.

***

p < 0.01, **p < 0.05, *p < 0.10.

Many empirical studies (see Brown and Corcoran 1997 and Lin 2010, among others) consider the choice of college majors as a relevant factor in explaining large gender disparities in pay at the beginning of the career. Results in Table 7, however, document large gender disparities in pay that persist even between individuals who studied the same field. Moreover, we observe that the unexplained component of the gender pay gap remains nevertheless high within each field of study.

Table 7:

Gender Pay Gap by College Majors

College Majors	Percentage of Raw Pay Gap	T Statistic	Percentage of Gap Explained by Observed Characteristics	Percentage of Gap Unexplained by Observed Characteristics
Humanities	8.86	3.02***	10.55	89.45
Economics, business, and statistics	9.25	8.22***	12.74	87.26
Political science and sociology	8.95	4.76***	56.77	43.23
Sciences	10.27	7.87***	1.83	98.17
Law	3.11	4.35***	53.68	46.32
Engineering	6.64	4.36***	16.4	83.6
Architecture	10.02	3.42***	13.98	86.02
Medicine	16.13	2.24***	46.72	53.28

NOTE: The third column reports the values of the T statistic for the null hypothesis that the difference between the monthly earning is zero.

***

p < 0.01, **p < 0.05, *p < 0.10.

Our first hypothesis stated that gender stereotypes in performance evaluation should not operate when the assessment of individual productivity is unnecessary, that is, there are no tournaments. We consider self-employment as a method of avoiding discrimination by employers. Hence, the unexplained gender pay gap among self-employed workers should be lower than among employees. Our data confirm the first hypothesis: Table 6 (first two rows) shows that the explained component of the gender pay gap among the self-employed is more than twice that of employees (27.8 compared to 12.2).

Results in Table 6 confirm Hypothesis 2; that is, the unexplained component of the gender pay gap increases when tournaments are more likely to be unfair and decreases when tournaments are more likely to be fair. In particular, we consider two environments where stereotypes are more likely to exert an influence because the employer is less or not motivated to make accurate judgments: low-profile clerical jobs and temporary work. Our data show that the unexplained component of the gender pay gap in low-skilled jobs is higher than in high-skilled jobs. Similarly, the unexplained component of the gender pay gap is greater among employees hired under fixed-term contracts than it is among employees hired under permanent wage contracts (80.8 compared to 66.9). On the contrary, when employees are recruited through open competition, performance appraisal should be more objective, more structured and less ambiguous, thereby reducing the conditions for gender stereotypes to flourish. We observe that when employees are recruited through open competition, the unexplained component of the gender pay gap is slightly lower (60.3 compared to 66.0).

Moreover, we want to test whether a signaling activity through the educational performance can reduce the unexplained component of the gender pay gap. Our assumption is that excelling in a university career reduces ambiguity in personnel evaluation and help counteracting the effect of stereotyping. Our data confirm this assumption by showing that achieving the maximum degree score significantly reduces the unexplained component of the gender pay gap (from 73.0 to 51.9). This finding provides support for Hypothesis 3.

Lastly, we check the adequacy of our data to explain differences in wages other than the gender pay gap. Our results show that the same type of wage decomposition can capture most of the differences in productivity that explain the pay gap between groups other than gender. For example, available information on individuals and jobs can explain more than 65 percent of the difference in pay between self-employed and employees. The comparison between several types of wage differentials shows that the gender pay gap is by far the most unexplained among the considered groups (Table 8).

Table 8:

Gender Pay Gap versus Other Differences in Pay between Groups

	Percentage of Raw Pay Gap	T Statistic	Percentage of Gap Explained by Observed Characteristics	Percentage of Gap Unexplained by Observed Characteristics
Gender pay gap	11.05	20.66***	12.23	87.77
Public sector vs. private sector	0.74	0.98	80.32	19.48
Self-employed vs. employees	6.34	8.74***	65.02	34.98
Permanent contracts vs. fixed-term contracts	13.18	25.95***	40.44	59.56
Graduate degree required vs. not required (overeducation)	9.44	14.10***	48.4	51.6
Recruitment through open competition vs. without open competition	1.37	1.34	66.31	33.69

NOTE: The third column reports the values of the T statistic for the null hypothesis that the difference between the monthly earning is zero.

***

p < 0.01, **p < 0.05, *p < 0.10.

Conclusion

Even in societies where gender discrimination is normatively proscribed, cultural beliefs and gender stereotypes may influence performance evaluation of workers, generating a gender pay gap. We build upon previous research about gender wage inequality, introducing tournament theory as a framework for the analysis of gender disparities in pay. We hypothesize that gender stereotypes make occupational tournaments unfair, favoring men and lowering the probability of equivalently performing women of winning the prize. Our findings establish that wage discrimination is more obvious operational in arenas with more ambiguity and room for favoritism. Thus we argue this confirms the results of a growing empirical literature documenting the effects of gender prejudices on the assessment of individual productivity (Kobrynowicz and Biernat 1997; Olian, Schwab, and Haberfeld 1988; Schein 1973, 2001).

Our perspective suggests that the less fair the tournament is, the higher the unexplained component of the gender pay gap should be. We find that the gender pay gap is clear; assessments of women’s productivity appear to be biased when they enter the labor market. This result is important because it establishes a theoretically consistent and empirically robust relation between gender stereotyping and wage discrimination.

It is worth noting, however, that our data only allow us to infer discrimination from the wage gap that remains unexplained after including a wide range of proxies for productivity. Such inference is the most widely used approach to test for wage discrimination, but it is open to the criticism that the proxies do not adequately capture all differences in productivity. For example, our result could also contain the effect of some unobservable variable such as the aggressiveness that would make the participants different from each other, and therefore contributes to the explanation of the pay gap. This suggests that our data might be limited in explaining wage differences arising among heterogeneous individuals. Hence, we perform a robustness check on the suitability of our data to explain the differences in earnings among groups other than gender and we find that our wage decomposition captures most of the variability in pay. However, future research might add new evidence of gender discrimination in pay using panel data from personnel archives of large companies. The use of personnel data might substantially reduce the possible bias from unobserved heterogeneity by partially controlling for heterogeneity on the demand side.

This paper adds to the economic literature on gender stereotyping and wage discrimination focusing on tournaments’ unfairness. Both gender stereotyping and wage discrimination have been well documented by a substantial body of empirical research (Blau, Ferber, and Winkler 2010). However, as noted by Glick, Zion, and Nelson (1988), there is a lack of research showing that stereotyping is clearly related to wage discrimination. Our article helps establish a consistent relationship between jobs in which stereotyping may be more prominent and wage discrimination. We hypothesize that gender biases in the assessment of individual productivity makes occupational tournaments less fair in some arenas compared with others. Our analyses quantify how discrimination increases or decreases in different contexts in which stereotypes are more or less likely to come into play.

Footnotes

Appendix

Table A1.

Variable description.

Variable	Description
Wage	Natural log of the net monthly wage. Monthly wages are in euros and net of taxes and social security contributions.
Edperf	Educational Performance. See the definition on page 11.
Lambda	Heckman inverse mills ratio obtained at the first stage
Experience	Years of work experience
Experience²	Experience squared
Sciences	1 if the graduate studied physical or mathematical science; 0 otherwise
Pharmacy	1 if the graduate studied pharmacy; 0 otherwise
Natural sciences	1 if the graduate studied a natural science; 0 otherwise
Engineering	1 if the graduate studied an engineering course; 0 otherwise
Architecture	1 if the graduate studied architecture or related subject; 0 otherwise
Agricultural studies	1 if the graduate studied agriculture or related subject; 0 otherwise
Economics, business and statistics	1 if the graduate studied business or economics or statistics; 0 otherwise
Political sciences and sociology	1 if the graduate studied political science or sociology; 0 otherwise
Law	1 if the graduate studied law; 0 otherwise
Humanities	1 if the graduate studied a humanities subject including Italian literature; 0 otherwise
Foreign languages	1 if the graduate studied foreign languages; 0 otherwise
Teachers college	1 if the graduate studied an education course, e.g., teacher training; 0 otherwise
Psychology	1 if the graduate studied psychology; 0 otherwise
Medicine	1 if the graduate studied medicine; 0 otherwise
Hours worked	Natural log of the number of hours worked in a week
University of north	1 if graduate studied in a university in the north; 0 otherwise
University of center	1 if graduate studied in a university in the center; 0 otherwise
University of south	1 if graduate studied in a university in the south; 0 otherwise
Private university	1 if graduate studied in a private university; 0 otherwise
Liceo	1 if graduate studied in a general high school (“Liceo”); 0 otherwise. In the Italian system, general high schools are academic oriented as opposed to technical and teaching schools.
Previously entered another degree course	1 if graduate previously attended a different degree course; 0 otherwise
Studied in the hometown	1 if graduate studied in the hometown; 0 otherwise
Moved to attend university	1 if graduate moved to a different town to attend university; 0 otherwise
Attended private lessons	1 if graduate attended private lessons; 0 otherwise
Student exchange program	1 if graduate attended the Erasmus program; 0 otherwise
Working student	1 if part time student; 0 otherwise
Paid training	1 if graduate received employer-paid training; 0 otherwise
Married	1 if graduate was married when interviewed; 0 otherwise
Children	1 if graduate has children when interviewed; 0 otherwise
Father college degree	1 if graduate’s father possesses a university degree; 0 otherwise
Father secondary degree	1 if graduate’s father possesses a high school degree; 0 otherwise
Mother college degree	1 if graduate’s mother possesses a university degree; 0 otherwise
Mother secondary degree	1 if graduate’s mother possesses a high school degree; 0 otherwise
Father in work	1 if graduate’s father was working when graduate was 14; 0 otherwise
Father self-employed	1 if graduate’s father was working as self-employed when student was 14; 0 otherwise
Father manager	1 if graduate’s father was a manager when the student was 14; 0 otherwise
Father in intermediate position (between manager and clerk)	1 if graduate’s father was an executive cadre when the student was 14; 0 otherwise
Father white collar, clerk	1 if graduate’s father was a white collar when the student was 14; 0 otherwise
Mother employed	1 if graduate’s mother was working when the student was 14; 0 otherwise
Mother self-employed	1 if graduate’s mother was working as self-employed when the student was 14; 0 otherwise
Mother manager	1 if graduate’s mother was a manager when the student was 14; 0 otherwise
Mother in intermediate position (between manager and clerk)	1 if graduate’s mother was an executive cadre when the student was 14; 0 otherwise
Mother white collar, clerk	1 if graduate’s mother was a white collar when the student was 14; 0 otherwise
Work	1 if graduate works; 0 otherwise
Firm size	1 if graduate works in a firm with more than 50 employees; 0 otherwise
Industrial sector	1 if graduate works in the industrial sector; 0 otherwise
Regional dummies	19 regional dummies for the region where the current job is located, used as controls in the earning equation; and19 regional dummies for the region of residence, used as controls in the selection equation

Notes

Carolina Castagnetti is assistant professor in the Department of Economics and Business at Pavia University. Her research focuses on Applied Econometrics and Econometrics for Finance and Education and has been published in Economics of Education Review; Small Business Economics; Journal of Applied Econometrics; Applied Financial Economics.

Luisa Rosti is full professor in the Department of Economics and Business at Pavia University. Her research on gender discrimination, education, and self-employment has been published in Economics of Education Review, Small Business Economics, International Journal of Manpower, Education + Training. She is author of Femina Oeconomica (Ediesse, Rome, 1996).

References

Altonji

Joseph G.

Blank

Rebecca M.

1999. Race and gender in the labor market. In Handbook of labor economics, edited by Ashenfelter

Orley C.

Card

David

. Princeton, NJ: Princeton University Press.

Anker

Richard

. 1998. Gender and jobs: Sex segregation of occupations in the world. Geneva: International Labour Office.

Arrow

Kenneth J.

1972. Models of job discrimination. In Racial discrimination in economic life, edited by Pascal

Anthony H.

Lexington, MA: Lexington Books.

Bielby

William T.

Baron

James N.

1986. Men and women at work: Sex segregation and statistical discrimination. American Journal of Sociology 91:759-99.

Blau

Francine D.

Kahn

Lawrence M.

2000. Gender differences in pay. Journal of Economic Perspectives 14:75-99.

Blau

Francine D.

Kahn

Lawrence M.

2007. Changes in the labor supply behavior of married women: 1980-2000. Journal of Labor Economics 25:393-438.

Blau

Francine D.

Ferber

Marianne A.

Winkler

Anne E.

2010. The economics of women, men and work, 6th edition. Upper Saddle River, NJ: Prentice Hall.

Bobbitt-Zeher

Donna

. 2007. The gender income gap and the role of education. Sociology of Education 80:1-22.

Booth

Alison

Francesconi

Marco

Frank

Jeff

. 2002. Temporary jobs: stepping stones or dead ends? The Economic Journal 112:189-213.

10.

Brown

Charles

Corcoran

Mary

. 1997. Sex-based differences in school content and the male-female wage gap. Journal of Labor Economics 15:431-65.

11.

Castagnetti

Carolina

Rosti

Luisa

. 2009. Effort allocation in tournaments: The effect of gender on Italian graduate performance. Economics of Education Review 28:357-69.

12.

Castagnetti

Carolina

Rosti

Luisa

. 2010. The gender gap in academic achievements of Italian graduates. In Gender gap: Causes, experiences and effects, edited by Davies

Samuel A.

New York: Nova Science.

13.

Correll

Shelley J.

2001. Gender and the career choice process: The role of biased self-assessments. American Journal of Sociology 106:1691-730.

14.

Correll

Shelley J.

2004. Constraints into preferences: Gender, status and emerging career aspirations. American Sociological Review 69:93-113.

15.

Dobbs

Michael

Crano

William D.

2001. Outgroup accountability in the minimal group paradigm: Implications for aversive racism and social identity theory. Personality and Social Psychology Bulletin 27:355-64.

16.

Eagly

Alice H.

Steffen

Valerie J.

1984. Gender stereotypes stem from the distribution of women and men into social roles. Journal of Personality and Social Psychology 46:735-54.

17.

England

Paula

. 1992. Comparable worth: Theories and evidence. New York: Aldine de Gruyter.

18.

England

Paula

. 2005. Gender inequality in labor markets: The role of motherhood and segregation. Social Politics 12:264-88.

19.

England

Paula

. 2010. The gender revolution: Uneven and stalled. Gender & Society 24:149-66.

20.

England

Paula

Folbre

Nancy

. 2005. Gender and economic sociology. In Handbook of economic sociology, edited by Smelser

Neil J.

Swedberg

Richard

. Princeton: Princeton University Press.

21.

England

Paula

. 2006. Desegregation stalled: The changing gender composition of college majors, 1971-2002. Gender & Society 20:657-77.

22.

Glick

Peter

Zion

Cari

Nelson

Cynthia

. 1988. What mediates sex discrimination in hiring decisions? Journal of Personality and Social Psychology 55:178-86.

23.

Hakim

Catherine

. 2000. Work-lifestyle choices in the 21st century: Preference theory. Oxford: Oxford University Press.

24.

Hamilton

David L.

Trolier

Tina K.

1986. Stereotypes and stereotyping: An overview of the cognitive approach. In Prejudice, discrimination and racism, edited by Dovidio

John F.

Gaertner

Samuel L.

Orlando, FL: Academic Press.

25.

Heckman

James J.

1979. Sample selection bias as a specification error. Econometrica 47:153-61.

26.

Heilman

Madeline E.

2001. Description and prescription: How gender stereotypes prevent women’s ascent up the organizational ladder. Journal of Social Issues 57:657-74.

27.

Heilman

Madeline E.

Okimoto

Tyler G.

2008. Motherhood: A potential source of bias in employment decisions. Journal of Applied Psychology 93:189-98.

28.

ISTAT 2007. Inserimento professionale dei laureati Indagine 2007. Manuale utente e tracciato record www.istat.it.

29.

Kobrynowicz

Diane

Biernat

Monica

. 1997. Decoding subjective evaluations: How stereotypes provide shifting standards. Journal of Experimental Social Psychology 6:579-601.

30.

Lazear

Edward P.

1998. Personnel economics for managers. New York: John Wiley.

31.

Lazear

Edward P.

Rosen

Sherwin

. 1981. Rank-order tournaments as optimum labor contracts. Journal of Political Economy 89:841-64.

32.

Lin

Eric S.

2010. Gender wage gaps by college major in Taiwan: Empirical evidence from the 1997-2003 Manpower Utilization Survey. Economics of Education Review 29:156-64.

33.

Loh

Eng S.

1994. Employment probation as a sorting mechanism. Industrial and Labor Relations Review 47:471-86.

34.

Magnusson

Charlotta

. 2010. Mind the gap: Essays on explanations of gender wage inequality. Swedish Institute for Social Research Dissertation Series No 78.

35.

Mincer

Jacob

Polachek

Solomon

. 1974. Family investments in human capital. Journal of Political Economy 82:S76-108.

36.

Moore

Robert L.

1983. Employer discrimination: Evidence from self-employed workers. Review of Economics and Statistics 65:496-501.

37.

Niederle

Muriel

Vesterlund

Lise

. 2007. Do women shy away from competition? Do men compete too much? Quarterly Journal of Economics 3:1067-101.

38.

Nieva

Veronica F.

Gutek

Barbara A.

1980. Sex effects on evaluation. Academy of Management Review 5:267-76.

39.

OECD. 2009. Education at a glance. OECD Indicators.

40.

Olian

Judy D.

Schwab

Donald P.

Haberfeld

Yitchak

. 1988. The impact of applicant gender compared to qualifications on hiring recommendations: A meta-analysis of experimental studies. Organizational Behavior and Human Decision Processes 2:180-95.

41.

O’Keeffe

Mary

Viscusi

Kip W.

Zeckhauser

Richard J.

1984. Economic contests: Comparative reward schemes. Journal of Labor Economics 2:27-56.

42.

Prokos

Anastasia

Padavic

Irene

. 2005. An examination of competing explanations for the pay gap among scientists and engineers. Gender & Society 19:523-43.

43.

Reskin

Barbara F.

Bielby

Denise D.

2005. A sociological perspective on gender and career outcomes. Journal of Economic Perspectives 19:71-86.

44.

Ridgeway

Cecilia L.

2009. Framed before we know it: How gender shapes social relations. Gender & Society 23:145-60.

45.

Ridgeway

Cecilia L.

Correll

Shelley J.

2004. Unpacking the gender system: A theoretical perspective on cultural beliefs in social relations. Gender & Society 18:510-31.

46.

Rosen

Sherwin

. 1982. Authority, control, and the distribution of earnings. Bell Journal of Economics 13:311-23.

47.

Rustichelli

Emiliano

. 2010. Rompere il cristallo: i risultati di un’indagine Isfol sui differenziali retributivi di genere in Italia. Roma: ISFOL.

48.

Sackett

Paul R.

Dubois

Cathy L. Z.

Noe

Ann Wiggins

. 1991. Tokenism in performance evaluation: The effects of work group representation on male-female and white-black differences in performance ratings. Journal of Applied Psychology 76:263-67.

49.

Schein

Virginia E.

1973. The relationship between sex-role stereotypes and requisite management characteristics. Journal of Applied Psychology 2:95-100.

50.

Schein

Virginia E.

2001. A global look at psychological barriers to women’s progress in management. Journal of Social Issues 4:675-88.

51.

Schotter

Andrew

Weigelt

Keith

. 1992. Asymmetric tournaments, equal opportunity laws, and affirmative actions: Some experimental results. Quarterly Journal of Economics 107:511-39.

52.

Solberg

Eric

Laughlin

Teresa

. 1995. The gender pay gap, fringe benefits, and occupational crowding. Industrial and Labor Relations Review 48:692-708.

53.

Spence

Michael

. 1973. Job market signaling. Quarterly Journal of Economics 87:355-74.

54.

Steinberg

Ronnie

. 1990. Social construction of skill: Gender, power, and comparable worth. Work and Occupations 17:449-82.

55.

Stiglitz

Joseph E.

1973. Approaches to the economics of discrimination. The American Economic Review 63:287-95.

56.

Tetlock

Philip E.

Kim

Jae I.

1987. Accountability and judgment processes in a personality prediction task. Journal of Personality and Social Psychology 52:700-9.

57.

Valian

Virginia

. 1998. Why so slow? The advancement of women. Cambridge: MIT Press.

58.

Wang

Ruqu

Weiss

Andrew

. 1998. Probation, layoffs, and wage-tenure profiles: A sorting explanation. Labour Economics 5:359-83.

59.

Welle

Brian

Heilman

Madeline E.

2007. Formal and informal discrimination against women at work. In Managing social and ethical issues in organizations, edited by Steiner

Dirk D.

Skarlicki

Daniel P.

Gilliland

Stephen W.

Greenwich: Information Age.

60.

Winslow

Sarah

. 2010. Gender inequality and time allocations among academic faculty. Gender & Society 24:769-93.