Equal Pay for Equal Task: Assessing Heterogeneous Returns to Tasks across Genders

• Author: Claudia Pigini,Elizabeth J. Casabianca,Alessia Lo Turco
• Published date: 01 April 2020
• DOI: http://doi.org/10.1111/irel.12253
• Date: 01 April 2020

Equal Pay for Equal Task: Assessing

Heterogeneous Returns to Tasks across Genders

ELIZABETH J. CASABIANCA, ALESSIA LO TURCO and

CLAUDIA PIGINI*

We inspect the heterogeneous association between tasks and wages across genders

using individual-level data on U.S. workers. Our findings suggest that women receive a

higher wage premium when engaged in cognitive tasks and experience more contained

wage losses when performing manual activities. However, a wage penalty characterizes

women engaged in highly social intensive jobs. Further inspection reveals that this

result is especially driven by the teamwork component of social activities.

Introduction

A significant share of academic research has been devoted to the analysis of

the gender wage gap. At the outset, economists approached this issue by

applying the traditional human capital model, in which returns to labor are

linked to skills acquired through education and work experience. The docu-

mented decline of the gender wage gap in the United States in the 1980s was,

then, attributed to an increase in the share of high-skilled women entering the

labor market (Mulligan and Rubinstein 2008). Lately, an increasing number of

researchers in this field have started to employ the “task framework”(Autor,

Levy, and Murnane 2003) to overcome the limits of the traditional approach.

1

According to this view, occupations are bundles of tasks necessary to suc-

cessfully accomplish a job. Early work using this model has associated the

abovementioned narrowing of the gender wage gap to two factors: On the one

hand, to the higher occurrence of female workers in cognitive-intensive activi-

ties (Bacolod and Blum 2010), which have been increasingly valued by

employers, especially with the diffusion of information and communication

JEL codes: J24, J31.

*The authors’affiliations is Prometeia, Bologna, Italy. E-mail: elizabeth.casabianca@prometeia.com;

Universit

a Politecnica delle Marche (IT), Ancona, Italy. E-mail: a.loturco@univpm.it; and Universit

a Politec-

nica delle Marche (IT), Ancona, Italy. E-mail: c.pigini@univpm.it.

1

Among the main criticisms is the inability of the traditional human capital model of capturing the

demand side of the human capital market (Autor and Handel 2013) and thus, identify the skills required to

successfully perform a job.

INDUSTRIAL RELATIONS, DOI: 10.1111/irel.12253. Vol. 59, No. 2 (April 2020). ©2020 Regents of the

Universit y of Calif ornia. Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA,

and 9600 Garsington Road, Oxford, OX4 2DQ, UK.

197

technologies (Autor, Levy, and Murnane 2003; Black and Spitz-Oener 2010);

on the other, to the displacement of male workers by technological progress,

traditionally concentrated in motor-intensive occupations (Yamaguchi 2018).

Against this background, we contribute to the related literature first by

exploring whether men and women perform different tasks both between and

within narrowly defined occupations. Second, we inspect the heterogeneous

association between individual tasks and wages across genders.

Our work is motivated by the possible existence of within-occupation task

and wage disparities across genders. More specifically, as an occupation is

defined as the outcome of the combination of different job tasks to be per-

formed with heterogeneous intensities, men and women engaged in the same

occupation may sort into different job activities depending on their compara-

tive advantages/preferences over tasks. Existing evidence points to men show-

ing a higher propensity for brawn-intensive activities, while revealing that

women are more inclined to perform brain-intensive activities (Rendall 2010;

Welch 2000). Related work also finds that, among cognitive jobs, women pre-

fer jobs with better anticipated work–life balance and tend to be less identified

with the stereotypes of masculine jobs (Barbulescu and Bidwell 2013). In the

same line, further research has shown that women differently assess, with

respect to men, the importance of money and people in a job and that these

factors have different returns in the labor market (Fortin 2008). Similarly,

some have argued that women typically have lower negotiating skills com-

pared to their male counterparts, especially when it comes to asking for pro-

motions and pay raises (Babcock and Laschever 2003). Recent contributions,

though, have failed to find evidence supporting this argument (Artz, Goodall,

and Oswald 2018; Stevens and Whelan 2019). Women also tend to sort, rela-

tively more than men, into occupations that are more socially rather than mon-

etarily oriented (Grove, Hussey, and Jetter 2011; Krueger and Schkade 2008).

2

Social activities also characterize managerial job positions in which, however,

the gender wage gap has been declining much more slowly with respect to the

middle and left tail of the wage distribution (Blau and Kahn 2017).

3

2

Experimental evidence further stresses the higher pro-sociality of female individuals (Eckel and Gross-

man 2008; Khachatryan et al. 2015).

3

The determinants of wage differentials in highly paid occupations have mainly been identified with the

greater career discontinuity and shorter work hours faced by women because of motherhood, in the tradi-

tional family division of labor by gender (Bertrand, Goldin, and Katz 2010; Wood, Corcoran, and Courant

1993). In this respect, Goldin (2014) has further emphasized the role of working long and specific hours in

explaining the gender wage gap in highly paid occupations. Specifically, she argues that penalties for a

lower flexibility concern jobs with specific occupational characteristics, among which is the worker’s ability

to establish and maintain relationships with others. Moreover, it has been estimated that women may face a

wage gap up to 45 percent in managerial positions because they run smaller businesses (Bertrand and Hal-

lock 2001; Mu~

noz-Bull

on 2010).

198 / ELIZABETH J. CASABIANCA,ALESSIA LOTURCO,AND CLAUDIA PIGINI

For the purpose of our work, we collect individual information on job tasks

available through the Princeton Data Improvement Initiative (PDII) survey.

These data allow us to build composite measures of job tasks at the person

level across a variety of task dimensions. In this work, we assume that an

occupation is a bundle of cognitive, manual, and social tasks. We characterize

cognitive tasks according to the proportion of time they are involved in prob-

lem solving, in the use of mathematics, and of the typical length of the docu-

ment workers read on the job. We build a measure of manual task based on

the amount of time workers spend performing motor activities. For the defini-

tion of social tasks, we select information on the proportion of time workers

spend managing or supervising other workers and working in teams.

4

We cor-

roborate the validity of our individual task indicators by comparing them to

occupation-level measures typically used in empirical work. We follow extant

literature and build similar indicators at the occupation level using the O*NET

database, such as the motor and cognitive task indicators proposed by Peri and

Sparber (2009) and Bacolod and Blum (2010) and the definition of social tasks

put forth by Deming (2017). A comparison between our individual task mea-

sures and the O*NET occupation level ones reveals that they are highly

similar.

Then we move on to explore whether men and women differ in terms of

task sorting behavior across and within narrowly defined occupations. To this

end, we make use of descriptive statistics and test the emerging evidence by

estimating a series of linear regression models for the task levels to describe to

what extent individual, industry and occupation heterogeneity can drive differ-

ences in task levels between men and women. Our second goal is to assess

whether individual tasks differently predict wages of men and women. We

estimate a wage equation in which we allow wages to differ between genders,

and tasks to be heterogeneously associated to men’s and women’s wages.

While doing so, we control for both occupation and industry fixed effects. In

the final part of the article, we assess to what extent our estimates could be

affected by the endogenous task sorting of workers. More specifically, individ-

uals are likely to self-select into those tasks that provide them the highest

return according to their comparative advantage. We tackle this issue by fol-

lowing the approach designed by Klein and Vella (2010) which exploits

heteroskedasticity as a source of additional exogenous variation.

Our work is close to the stream of research measuring tasks at the occupa-

tional level and applying the task approach to explain the evolution of the

4

Autor and Handel (2013) use the same data set but, differently from them, we identify from the survey

those questions that inform on the pure manual, cognitive, and social content of a job, rather than on routine

and abstract tasks, and inspect differences by gender.

Equal Pay for Equal Task / 199