Affirmative Action and Intersectionality at the Top: Evidence from South Africa

• DOI: http://doi.org/10.1111/irel.12272
• Author: Stephan Klasen,Anna Minasyan
• Published date: 01 January 2021
• Date: 01 January 2021

Aff‌irmative Action and Intersectionality at the

Top: Evidence from South Africa

STEPHAN KLASEN and ANNA MINASYAN

Previous research shows that gender-based board quotas alone do not increase

women’s share in senior management positions. We study the effects of an aff‌ir-

mative action policy in South Africa, which stipulates group-based targets for

senior management roles, beyond boards, focusing on representing intersectional

identities, sex, and race. Our f‌indings show that the policy led to a sizable

increase in employment probability in top positions for Black women relative to

their employment in these positions in the counterfactual scenario of no policy.

We extend our analysis and estimate policy spillovers in education, wages, and

self-employment.

Introduction

Aff‌irmative action policies, such as group-based quotas and targets, f‌irst

introduced in India and the United States in the 1950s, have become one of

the most controversial state policy interventions for increasing the representa-

tion of previously discriminated population groups in politics (Hughes 2011;

O’Brien and Rickne 2016), labor markets (Kurtulus 2012; Miller 2017; Peck

2017), and educational institutions (Backes 2012; Bagde, Epple, and Taylor

2016; Francis-Tan and Tannuri-Pianto 2018). There is a substantive body of

interdisciplinary literature studying the direct and indirect effects of group-

based aff‌irmative action policies on various outcomes (Holzer and Neumark

2000; Kirsch 2018; Leonard 1990).

The overarching aim of gender-based quotas is to accelerate the process of

achieving a proportionate representation of women in decision-making bodies.

Recent studies in economics explore spillovers of gender-based board quotas

and do not f‌ind any causal evidence of trickle-down effects in senior manage-

ment positions in Norway (Bertrand et al. 2019) or Italy (Maida and Weber

†

The authors’aff‌iliations are, respectively, University of Goettingen, Plats der Goettinger Sieben 3, 37073

Goettingen, Germany. University of Groningen, Nettlebosje 2, 9747 AE Groningen, The Netherlands.

E-mail: a.minasyan@rug.nl.

INDUSTRIAL RELATIONS, DOI: 10.1111/irel.12272. Vol. 60, No. 1 (January 2021). ©202 1 The Authors.

Industrial Relations published by Wiley Periodicals LLC on behalf of Regents of the University of California (RUC).

350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford, OX4 2DQ, UK.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs

License, which permits use and distribution in any medium, provided the original work is properly cited, the

use is non-commercial and no modif‌ications or adaptations are made.

3

2019). Research on gender wage gaps shows that the differences in pay

between men and women continue to persist in the presence of board quotas

(Blau and Kahn 2017; Reb´

erioux and Roudaut 2019). Yet, non-quota studies

f‌ind positive spillovers from higher representation of women on company

boards to top managerial positions in the United States and Norway over a

long period (Kunze and Miller 2017; Matsa and Miller 2011). Therefore,

beyond the board quotas, a more comprehensive policy, which incorporates

group-based targets for top management positions, may help accelerate gender

parity at the top. However, there is a lack of evidence if such full-f‌ledged

labor-market policies work, as these are relatively rare in practice.

In this article, we present a unique evidence from a South African aff‌irma-

tive action policy, the Broad-Based Black Economic Empowerment Act of

2003 (BEE), which stipulates not only group-based targets for board represen-

tation but also for top management positions. The BEE policy sets targets for

top management positions for Black women equal to 40 percent of all Black

senior managers. The policy target for all Blacks is 60 percent as the share of

all senior managers in South Africa. In the BEE Act and also throughout this

article, the word Black(s) is used as a generic term to denote non-White popu-

lation groups in South Africa (SA); that is, African Blacks, Colored, and

Indian/Asians. Subsequently, we use the word White to denote White people

of European descent in South Africa. Blacks are the majority population group

in South Africa, and have been explicitly discriminated against and geographi-

cally segregated by the minority Whites under the South African apartheid

government. Black people are the benef‌iciaries of the BEE policy, whereas

White people are not.

We use annual data from the Post-Apartheid Labour Market Series

(PALMS) and adopt a triple difference-in-differences estimation method to

estimate the effects of BEE on Black women’s employment in top positions.

In our f‌irst model, we estimate the change in the probability of employment in

top positions for Black women and compare it to the changes for Black men,

White women, and White men from the pre-BEE to the post-BEE period.

However, we acknowledge that despite the comprehensive controls, f‌ixed

effects, and linear trends included in the analysis, the pre-trends in the top

employment for Blacks and Whites may not be comparable due to the apart-

heid legacy in South Africa. We address this issue by exploiting a unique

BEE policy feature that designates only large f‌irms as compliers. Therefore, in

our second estimation model, we identify the BEE effects for Black women’s

employment in top positions in large f‌irms (employers) by differencing out the

total effects for Black men and the effects for Black women working for small

employers. We also estimate the BEE effects for White women using the same

model for the sample of White people, and do not expect any positive effects.

4/ S

TEPHAN KLASEN AND ANNA MINASYAN

The results from the f‌irst estimation model show that the probability of

employment in top positions for Black women increased by 0.5 percentage

points from the pre-BEE to the post-BEE period, which is larger than the

change for Black men and White women in that period. The second estima-

tion model, which identif‌ies the BEE effects, shows a more than 1 percent-

age point increase for Black women working for affected (large) employers.

The BEE effects are the largest for highly educated Black women, whose

chances for top management positions are twice more relative to their

employment in these positions in the counterfactual scenario of working for

non-affected, small employers. The chances of Black men and White women

for employment in top positions decreased slightly from the pre-BEE to the

post-BEE period.

In absolute terms, White men in South Africa, on average, have the highest

probability of employment at the top relative to all other groups. But in rela-

tive terms, the percentage change for Black women from the pre-BEE to the

post-BEE period is larger than the percentage change for White men, White

women, and Black men. This relatively large increase for Black women helped

the country almost reach the policy target of 40 percent as the share of all

Black senior managers by 2015.

We test the robustness of the main f‌indings by estimating the changes

related to an earlier policy (the Employment Equity Act in 1998) and the BEE

policy amendment of Codes of Good Practice in 2007. Additionally, we run a

placebo-like test by estimating the post-BEE changes for the informal sector,

which is quite prevalent in South Africa.

Finally, we explore the spillovers of the BEE policy in education, earn-

ings, and self-employment for Black and White women and men. The results

show that Blacks increased their schooling by more than a year from the

pre-BEE to the post-BEE period, reducing the racial education gap from 4 to

3 years. In terms of wages at the senior management level, we f‌ind that the

earnings for White men increased by 30 percent from the pre-BEE to the

post-BEE period, while the wages for Black women, Black men, and White

women remained unchanged in this period. We also detect the highest

increase in the probability of being self-employed among the Whites in this

period.

In the remainder of the article, we proceed as follows. We describe the

background of the BEE policy in South Africa in the next section, followed

by a presentation of the data and descriptive statistics based on the PALMS

database. We then lay out our two main estimation models then present our

main f‌indings. After reporting the results from robustness tests and spillover

analysis, we conclude with policy implications in the f‌inal section.

Aff‌irmative action and women in top management /5