Popular in Social Sciences

Political Science

V-Dem’s Conceptions of Democracy and Their Consequences

University of Notre Dame Michael Coppedge

Central Monetary Authorities

Macroeconomic modelling of CBDC: a critical review

European Central Bank Ulrich Bindseil

Social Identities and Divisions

A new explanation of antisemitism: Jew hatred as a civilisational phenomenon

Boston University Liah Greenfeld

Central Monetary Authorities

Financing Private Credit

Federal Reserve Bank of New York Nina Boyarchenko

Economics

Failing Banks

Massachusetts Institute of Technology Emil Verner

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

American Politics

Racial Peer Effects at Work

View more insights in Social Sciences - Economics - Labor Economics

View more figures from GIGA German Institute for Global and Area Studies

View the page of Katharina Fietz

Read the Study

Next in Economics

An early indicator for anomalous stock market performance

Thomas Gries | Universität Paderborn

International tax competition with rising intangible capital and financial globalization

Vincenzo Quadrini | University of Southern California

Katharina Fietz examines how working with same-race coworkers influences employee retention within firms. Using employer-employee administrative data from Brazil, the research leverages unexpected deaths of workers from different racial backgrounds as natural shocks to peer group composition.

Image courtesy of interviewee. October 7, 2024

Log-in or Sign-up to Faculti
Currently viewing this subject insight as a guest. You have insight(s) remaining for this month. Login to view 8000+ figures on the platform.

Generating Workspace...

Export as PDF

The study examines racial peer effects in Brazil and their influence on racial segregation in labor markets. It highlights significant disparities, revealing that 80% of white workers have white colleagues, while non-white workers have only about 50% white coworkers. This video “homophilic preference” for working with similar peers can lead non-white workers to leave jobs when the racial composition of their peer group changes, particularly after the unexpected death of a colleague.

The research focuses on what happens post-hire, analyzing how changes in peer group racial composition affect employee retention, drawing a distinction between voluntary quits and employer-initiated layoffs. Key findings indicate that non-white workers are more likely to quit after a non-white coworker dies, revealing that their response to changes in racial composition is more pronounced than that of white coworkers.

Moreover, the study considers various factors impacting these dynamics, including occupational roles and the presence of teamwork, which can mitigate negative retention effects by fostering closer interactions regardless of racial differences.

Overall, the results suggest that coworker composition plays a significant role in employee retention and labor market behavior, and the study highlights the necessity for policies promoting workplace diversity to consider not just hiring practices but also the interaction dynamics post-hiring to address racial inequalities effectively.

Language:

Export as PDF

Export as PDF
Here are five key points from the video:

1. **Racial Peer Effects and Job Retention**: The study investigates how changes in the racial composition of a worker's peer group affect their likelihood of remaining in their current job. It finds that non-white workers are more likely to leave their jobs after the death of a non-white colleague, indicating a preference for racially similar peers and suggesting that such dynamics contribute to racial segregation in the workplace.

2. **Focus on Post-Hiring Dynamics**: Unlike many studies that analyze racial bias in hiring practices, this research emphasizes the consequences of peer group composition after hiring. It differentiates between voluntary quits (employees choosing to leave) and involuntary layoffs (employer decisions), highlighting that the observed retention effects primarily result from worker-initiated quits.

3. **Methodology and Data Use**: The study employs a unique identification strategy involving the unexpected deaths of employees to assess the impact on peer group composition. It uses matched employer-employee records from Brazil, covering various sectors and representing a significant portion of the workforce, thus providing a broader understanding of peer effects in a middle-income country convideo.

4. **Findings on Racial Composition and Job Characteristics**: The results reveal that the death of a non-white coworker results in a significant decrease in the non-white share of the peer group and affects retention, particularly in white-collar jobs, while the effects are weaker in blue-collar occupations and jobs that require more teamwork.

5. **Implications for Diversity Policies**: The study contributes to understanding racial segregation in labor markets, emphasizing that policies promoting workplace diversity should address not only hiring discrimination but also post-hiring peer dynamics. It suggests future research should explore firm policies on staffing and promotions to mitigate the racial peer effects that perpetuate inequalities.

Export as PDF
Here are some key quotes and paraphrased key points from the video:

1. **Racial Peer Effects and Segregation**:
- "Our study extends racial peer effects in Brazil and how they might drive racial segregation."
- "In Brazil, 80% of white workers' colleagues are also white, while non-white workers have only about 50% of white colleagues..."

2. **Homophilic Preferences**:
- "Social science theories suggest that people have homophilic preferences... they prefer others who are like themselves."
- "Similar attraction can lead to preferences for working with demographically similar peers."

3. **Focus of Research**:
- "We focus on what happens after the hiring process... how employees react to changes in the racial composition of their peer group."

4. **Key Question**:
- "We posed at the outset... do workers prefer to work with coworkers who are racially similar to themselves?"

5. **Impact on Turnover**:
- "A higher share of white peers increases the turnover of black women."
- "Non-white workers are significantly more likely to leave their job after the death of a non-white coworker compared to when a white coworker dies."

6. **Labor Supply Dynamics**:
- "Observed effects are driven completely by employee-initiated quits rather than employer-initiated layoffs... suggesting a labor supply response instead of a labor demand effect."

7. **Racial Inequality**:
- "Racial inequality is substantial. Non-white workers earn on average 30 to 35% less than white workers."

8. **Study Convideo**:
- "Data comes from matched employer-employee records for formal dependent employees in Brazil, covering years from 2006 to 2021."

9. **Peer Group Changes**:
- "When a non-white worker dies, the non-white share of this peer group drops significantly by almost 20 percentage points."

10. **Teamwork and Diversity**:
- "Racial peer effects are more pronounced in higher-skilled occupations... teamwork can offset the negative effects of racial segregation."

11. **Gender Dimension**:
- "The groups more strongly affected are non-white females... retention is affected by both the racial and the gender composition of their coworkers."

12. **Policy Implications**:
- "Our findings highlight significant heterogeneities and the effect of racial peer group composition... underscoring the importance of post-hiring dynamics in shaping labor market outcomes for non-white and white workers."

These quotes capture the essence of the study, highlighting the connections between racial peer group composition, employee dynamics, turnover, and broader implications for workplace diversity and racial segregation.

X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:90000

1
00:00:00.065 --> 00:00:02.685
Our study extends racial peer effects in Brazil

2
00:00:02.825 --> 00:00:05.245
and how they might drive racial segregation.

3
00:00:06.185 --> 00:00:09.725
And racial segregation is evident in many labor markets,

4
00:00:10.115 --> 00:00:13.885
including Brazil, which serves as the context of our study.

5
00:00:14.505 --> 00:00:16.565
For example, in Brazil, 80%

6
00:00:16.625 --> 00:00:19.365
of white workers colleagues are also white.

7
00:00:19.815 --> 00:00:22.805
While non-white workers have only about 50%

8
00:00:22.805 --> 00:00:26.865
of white colleagues and several social science theories

9
00:00:26.865 --> 00:00:29.345
suggest that people have homophilic preferences.

10
00:00:30.205 --> 00:00:33.425
In other words, they prefer others who are like themselves.

11
00:00:34.245 --> 00:00:37.425
For example. Similar attraction can lead to preferences

12
00:00:37.525 --> 00:00:40.025
for working with demographically similar peers.

13
00:00:40.595 --> 00:00:44.305
While social categorization can also result in lower levels

14
00:00:44.365 --> 00:00:46.985
of cooperation and communication within groups

15
00:00:47.085 --> 00:00:48.625
who share fewer similarities,

16
00:00:49.815 --> 00:00:51.665
thereby a decrease in the presence

17
00:00:51.665 --> 00:00:55.145
of racial similar peers may lower job certification

18
00:00:55.565 --> 00:00:57.825
and make workers look for a different job.

19
00:00:59.195 --> 00:01:00.455
And large part of the

20
00:01:01.015 --> 00:01:03.695
existing literature primarily focus on the hiring process,

21
00:01:03.955 --> 00:01:07.095
asking questions like, are managers more likely

22
00:01:07.195 --> 00:01:10.815
to hire workers who are, who share characteristics

23
00:01:10.915 --> 00:01:12.495
as the manager who hires?

24
00:01:13.395 --> 00:01:16.775
And those questions mostly concern the labor demand side.

25
00:01:16.915 --> 00:01:18.495
So really what employers seek

26
00:01:18.555 --> 00:01:20.255
and demand when hiring a worker.

27
00:01:21.425 --> 00:01:25.245
In our study, however, my co-author Osher,

28
00:01:25.245 --> 00:01:28.485
from the University of Potsdam and I focus on what happens

29
00:01:28.695 --> 00:01:29.965
after the hiring process.

30
00:01:31.125 --> 00:01:34.325
Specifically, we examine how employees react

31
00:01:34.325 --> 00:01:37.325
to changes in the racial composition of their peer group,

32
00:01:37.775 --> 00:01:39.925
particular in terms of the likelihood

33
00:01:39.925 --> 00:01:41.485
of remaining in their current job.

34
00:01:42.705 --> 00:01:44.205
Um, this approach allows us

35
00:01:44.205 --> 00:01:47.485
to explore the labor supply side employees behavior

36
00:01:47.975 --> 00:01:51.485
after changes we can distinguish between quits,

37
00:01:51.695 --> 00:01:54.685
which reflect employees active decision to leave

38
00:01:54.865 --> 00:01:58.125
and layoffs which are initiated by the employer.

39
00:01:59.025 --> 00:02:02.125
So basically the key question we posed at the outset

40
00:02:02.125 --> 00:02:05.525
of this project was, do employees prefer to work

41
00:02:05.525 --> 00:02:08.445
with coworkers who are racially similar to themselves?

42
00:02:08.585 --> 00:02:09.765
We analyze the impact

43
00:02:09.765 --> 00:02:12.565
of an unexpected change in the racial composition

44
00:02:12.585 --> 00:02:14.525
of peer groups on workers' likelihood

45
00:02:14.525 --> 00:02:15.605
of staying at the firm.

46
00:02:16.245 --> 00:02:18.085
Probably the most relevant body of work

47
00:02:18.185 --> 00:02:21.285
for us is the literature on peer effects in the workplace.

48
00:02:22.145 --> 00:02:25.325
And most of the economic literature actually focus on

49
00:02:25.325 --> 00:02:27.765
focuses on skills and productivity of workers

50
00:02:28.065 --> 00:02:29.525
and their impact on peers.

51
00:02:30.185 --> 00:02:33.005
For example, how does working with a more productive

52
00:02:33.105 --> 00:02:35.845
or less productive colleague affect other employees?

53
00:02:36.945 --> 00:02:39.195
It's more the interdisciplinary research

54
00:02:39.265 --> 00:02:40.635
that looks at the impact

55
00:02:40.635 --> 00:02:44.675
of demographic similarities in peer groups, often finding

56
00:02:44.675 --> 00:02:46.075
that groups with higher levels

57
00:02:46.095 --> 00:02:48.835
of similarity experience less turnover.

58
00:02:50.095 --> 00:02:53.595
For our study, probably the most relevant work is from Lenos

59
00:02:53.595 --> 00:02:55.475
and co-authors from 2024

60
00:02:55.655 --> 00:02:58.715
who analyze random team assignment in a

61
00:02:58.715 --> 00:03:01.485
High wage US professional service firm

62
00:03:02.105 --> 00:03:04.365
and find that a higher share of white peers

63
00:03:05.005 --> 00:03:07.005
increases the turnover of black women.

64
00:03:08.525 --> 00:03:11.145
And most studies on demographic composition

65
00:03:11.245 --> 00:03:14.945
and its effect on coworkers, focus on single firm settings

66
00:03:15.125 --> 00:03:17.225
and mostly in high income countries.

67
00:03:18.005 --> 00:03:20.905
So we see our study introducing a novelty

68
00:03:21.125 --> 00:03:23.025
by examining the causal effect

69
00:03:23.045 --> 00:03:25.905
of peer group composition across a large set

70
00:03:25.905 --> 00:03:28.665
of firms in a middle income country, Brazil,

71
00:03:29.205 --> 00:03:31.425
and across different sectors and location.

72
00:03:32.605 --> 00:03:34.585
But we also see ourself contributed

73
00:03:34.605 --> 00:03:37.305
to the literature on cross race interaction

74
00:03:37.405 --> 00:03:40.305
and racial segregation in various social contexts.

75
00:03:40.725 --> 00:03:41.985
For example, the literature

76
00:03:42.135 --> 00:03:44.985
that tests the so-called white flag hypothesis.

77
00:03:45.565 --> 00:03:46.905
For example, how do housing

78
00:03:47.045 --> 00:03:50.465
and school choice decisions of whites are affected

79
00:03:50.525 --> 00:03:51.825
by the presence of blacks?

80
00:03:52.685 --> 00:03:55.875
While we extend the literature by focusing on the workplace,

81
00:03:56.175 --> 00:03:59.435
we also added by focusing not only on the majority group,

82
00:03:59.535 --> 00:04:03.155
but analyzing also how the minority group responds

83
00:04:03.155 --> 00:04:05.235
to changes in cross race exposure.

84
00:04:06.815 --> 00:04:09.435
And finally, we also see ourself contributing

85
00:04:09.435 --> 00:04:11.795
to the literature on racial racial preferences

86
00:04:11.855 --> 00:04:13.915
and discrimination in the labor market.

87
00:04:14.795 --> 00:04:15.875
I already mentioned

88
00:04:15.875 --> 00:04:19.115
that many scholars focus on discrimination in the hiring

89
00:04:19.145 --> 00:04:22.835
process of different, of different managers.

90
00:04:23.495 --> 00:04:26.995
And if different managers have preferences to hire white

91
00:04:27.095 --> 00:04:30.315
or non-white workers, if they also are white or non-white.

92
00:04:30.935 --> 00:04:32.785
We, however, in our study,

93
00:04:32.915 --> 00:04:37.905
focus on the post hiring stage looking at worker separation,

94
00:04:38.035 --> 00:04:39.185
especially quits

95
00:04:39.845 --> 00:04:42.905
and how those quits respond to respond

96
00:04:42.905 --> 00:04:45.385
to changes in racial composition of coworkers.

97
00:04:45.385 --> 00:04:47.265
Racial inequality is substantial.

98
00:04:47.265 --> 00:04:48.825
In Brazil, for instance,

99
00:04:48.825 --> 00:04:51.265
non-white workers earn on average 30

100
00:04:51.325 --> 00:04:53.625
to 35% less than white workers.

101
00:04:54.405 --> 00:04:55.945
And even after accounting

102
00:04:55.965 --> 00:04:58.425
for differences in education and occupation.

103
00:04:58.795 --> 00:05:00.785
Those disparities persist due

104
00:05:00.785 --> 00:05:03.825
to differences in wage settings, practices and hiring.

105
00:05:05.095 --> 00:05:06.235
And probably we also need

106
00:05:06.235 --> 00:05:08.915
to mention a few words on the racial notion in Brazil.

107
00:05:09.715 --> 00:05:13.155
Brazil's official statistics classifies race into five

108
00:05:13.155 --> 00:05:14.235
different categories.

109
00:05:14.875 --> 00:05:18.555
Barranco for white, creto for black, pado for brown,

110
00:05:18.715 --> 00:05:21.595
Amarillo for yellow, and indigenous for indigenous.

111
00:05:22.725 --> 00:05:26.225
In our study, we combine creto and pado, so black

112
00:05:26.285 --> 00:05:29.145
and brown into one category called non-white.

113
00:05:29.765 --> 00:05:32.385
And together with the category wide, we

114
00:05:32.385 --> 00:05:35.185
therefore capture 99% of the population.

115
00:05:36.645 --> 00:05:39.345
Our data comes from matched employer employee records

116
00:05:39.605 --> 00:05:42.105
for formal dependent employees in Brazil,

117
00:05:42.565 --> 00:05:45.625
and we use the years from 2006 to 2021.

118
00:05:46.565 --> 00:05:48.615
This data set, which is called ice,

119
00:05:48.735 --> 00:05:52.135
includes information about workers' race, occupation,

120
00:05:52.655 --> 00:05:53.855
contract type, and

121
00:05:53.875 --> 00:05:56.615
how their contracts were terminated, for example, through

122
00:05:56.705 --> 00:05:58.020
Quits or through layoffs.

123
00:05:59.435 --> 00:06:00.535
And it is important

124
00:06:00.595 --> 00:06:02.935
to mention Brazil separation regulations.

125
00:06:03.615 --> 00:06:06.305
Workers that are laid off are eligible

126
00:06:06.325 --> 00:06:09.185
for certain benefits like unemployment, insurance

127
00:06:09.325 --> 00:06:11.265
and severance payment provided

128
00:06:11.265 --> 00:06:13.305
that they meet certain workers store,

129
00:06:13.305 --> 00:06:14.505
history requirements.

130
00:06:14.565 --> 00:06:18.515
Of course, however, these benefits are not available

131
00:06:18.655 --> 00:06:20.795
to workers who actively quit their job.

132
00:06:21.695 --> 00:06:24.505
This distinction is important in our setting

133
00:06:24.895 --> 00:06:27.665
because our study D differentiates

134
00:06:27.665 --> 00:06:29.145
between quits and layoffs.

135
00:06:29.685 --> 00:06:31.305
So really quits are the results

136
00:06:31.325 --> 00:06:34.865
of employees actively choosing to leave their job

137
00:06:35.625 --> 00:06:38.225
bearing the financial consequences of this decision.

138
00:06:40.015 --> 00:06:43.155
So now to test whether workers prefer to work with others

139
00:06:43.335 --> 00:06:45.195
who are more alike, we look at

140
00:06:45.255 --> 00:06:48.755
how workers react when their racial peer group composition

141
00:06:49.035 --> 00:06:52.075
suddenly changes and to exploit,

142
00:06:52.075 --> 00:06:54.635
exploit exogenous changes in the racial

143
00:06:54.745 --> 00:06:55.915
peer group composition.

144
00:06:56.575 --> 00:06:59.315
We focus on the unexpected death of workers.

145
00:07:00.455 --> 00:07:01.995
We compare two peer groups

146
00:07:02.105 --> 00:07:04.675
that have the same initial non-white share.

147
00:07:05.005 --> 00:07:06.075
Let's say for example,

148
00:07:06.265 --> 00:07:08.915
both peer groups have a non-white share of 30%,

149
00:07:09.615 --> 00:07:12.475
and in one peer group a non-white worker dies.

150
00:07:12.775 --> 00:07:15.195
And in another one, a white worker dies.

151
00:07:16.055 --> 00:07:17.715
By only comparing peer groups

152
00:07:17.785 --> 00:07:20.835
with the same initial non-white share, we see

153
00:07:20.835 --> 00:07:24.595
that the remaining in-car characteristics do not predict the

154
00:07:24.595 --> 00:07:25.915
race of the deceased worker.

155
00:07:26.795 --> 00:07:29.895
In other words, we have two groups in both peer groups.

156
00:07:29.955 --> 00:07:31.055
One worker dies

157
00:07:31.315 --> 00:07:34.695
and the race of the deceased worker can be taken as random.

158
00:07:35.795 --> 00:07:39.375
We therefore have a credible identification strategy to see

159
00:07:39.405 --> 00:07:42.815
what effect causes this change in the racial coworker

160
00:07:42.885 --> 00:07:44.735
composition on the retention

161
00:07:44.795 --> 00:07:46.375
of the remaining income and workers.

162
00:07:47.925 --> 00:07:51.065
And while we do several sample restrictions in order

163
00:07:51.085 --> 00:07:53.345
to ensure the credibility of our results,

164
00:07:53.585 --> 00:07:54.665
I want to highlight two.

165
00:07:55.675 --> 00:07:59.105
First, we only exclude death that are not related

166
00:07:59.165 --> 00:08:00.665
to workplace accidents.

167
00:08:01.315 --> 00:08:04.825
Since such incidents could influence coworkers feelings

168
00:08:04.825 --> 00:08:06.225
of job safety, for example.

169
00:08:07.205 --> 00:08:11.145
And second, we only focus on small peer groups that have no,

170
00:08:11.285 --> 00:08:12.945
not more than 30 workers.

171
00:08:13.885 --> 00:08:17.305
As the loss of only one worker in a larger group would not

172
00:08:17.305 --> 00:08:20.185
create a noticeable change in a peer group composition,

173
00:08:21.335 --> 00:08:23.595
we then use cross-sectional regressions

174
00:08:23.615 --> 00:08:26.395
to analyze retention effects during the year of the death

175
00:08:26.455 --> 00:08:27.875
and the three following years

176
00:08:28.015 --> 00:08:30.435
before we come to the main effect,

177
00:08:30.435 --> 00:08:32.475
which will be the effect on retention.

178
00:08:33.235 --> 00:08:36.315
I want to talk about the effect on the non-white share.

179
00:08:36.535 --> 00:08:40.115
So we assess whether the death of a non-white worker leads

180
00:08:40.135 --> 00:08:43.435
to lasting change in the racial composition of a peer group.

181
00:08:44.135 --> 00:08:46.875
We find that when a non-white worker dies,

182
00:08:47.415 --> 00:08:50.915
the non-white share of this peer group dropped significantly

183
00:08:51.575 --> 00:08:54.795
by almost 20 percentage points in the year of the death.

184
00:08:55.745 --> 00:08:57.725
And also over the following three years,

185
00:08:58.025 --> 00:09:01.765
the non-white share only recovers slowly remaining about 11

186
00:09:01.775 --> 00:09:03.725
percentage points lower than before.

187
00:09:05.025 --> 00:09:09.035
With that in mind, coming to our key findings that are

188
00:09:09.065 --> 00:09:10.355
that non-white income

189
00:09:10.355 --> 00:09:14.515
and workers are significantly more likely to leave their job

190
00:09:14.845 --> 00:09:18.875
after the death, death of a non-white coworker compared

191
00:09:18.875 --> 00:09:20.595
to when a white coworker dies.

192
00:09:21.505 --> 00:09:24.575
White coworkers on the other hand do not have similar

193
00:09:25.135 --> 00:09:27.175
response to changes in the racial composition.

194
00:09:28.095 --> 00:09:29.475
So those results highlight

195
00:09:29.475 --> 00:09:32.755
that peer effects are not symmetric across racial groups

196
00:09:33.495 --> 00:09:36.275
and that non-white workers are more strongly affected

197
00:09:36.295 --> 00:09:39.155
by the racial composition of the peers than white workers.

198
00:09:40.555 --> 00:09:41.975
And then in the paper, we try

199
00:09:41.975 --> 00:09:43.535
to shed light on the mechanisms.

200
00:09:44.455 --> 00:09:45.525
First, we find

201
00:09:45.525 --> 00:09:48.365
that the observed effects are driven completely

202
00:09:48.505 --> 00:09:52.765
by employee initiated quits rather than employer initiated

203
00:09:52.835 --> 00:09:57.005
layoffs suggesting a labor supply response instead

204
00:09:57.005 --> 00:09:58.365
of a labor demand effect.

205
00:09:59.545 --> 00:10:01.245
Second, we track workers

206
00:10:01.655 --> 00:10:05.525
after they separate from the initial firms for the subset

207
00:10:05.665 --> 00:10:08.845
of workers who start a new final job in the same year

208
00:10:08.905 --> 00:10:09.965
or the year following.

209
00:10:09.965 --> 00:10:13.485
After the separation, we compare the characteristics

210
00:10:13.505 --> 00:10:15.285
of the old and the new job.

211
00:10:16.255 --> 00:10:18.515
We find that those non-white workers

212
00:10:18.665 --> 00:10:21.995
that experience the death of a non-white peer move

213
00:10:21.995 --> 00:10:24.755
to a new job that feature a higher non-white

214
00:10:24.755 --> 00:10:25.875
share than the old job.

215
00:10:26.755 --> 00:10:29.855
So overall, those results suggest are a consistent

216
00:10:29.855 --> 00:10:32.775
with the idea of homophilic preferences for working

217
00:10:32.775 --> 00:10:35.135
with more same race peer,

218
00:10:35.165 --> 00:10:37.855
that drive the initial separation of non-white workers.

219
00:10:39.315 --> 00:10:41.855
And finally, we also look at different heterogeneities.

220
00:10:42.955 --> 00:10:44.145
First, we find

221
00:10:44.175 --> 00:10:47.505
that the effects are only present in white color occupations

222
00:10:47.645 --> 00:10:49.665
and not in blue color occupations patients.

223
00:10:50.525 --> 00:10:54.145
And in addition, we see that the effects are weaker in jobs

224
00:10:54.175 --> 00:10:55.905
that involve more teamwork.

225
00:10:56.825 --> 00:10:58.565
We draw two lessons from that.

226
00:10:59.215 --> 00:11:00.405
First, it seems

227
00:11:00.405 --> 00:11:03.045
that racial peer effects are more pronounced in higher

228
00:11:03.105 --> 00:11:06.805
income settings, which is by the way, in line with the fact

229
00:11:06.805 --> 00:11:09.685
that our average effects are smaller than the average

230
00:11:09.715 --> 00:11:11.925
effects found in high income settings,

231
00:11:11.925 --> 00:11:13.085
for example, in the us.

232
00:11:14.235 --> 00:11:17.255
And second, the teamwork Heterogeneities suggests

233
00:11:17.255 --> 00:11:20.095
that teamwork can kind of offset the negative effects

234
00:11:20.095 --> 00:11:22.295
of neuro retention, meaning

235
00:11:22.295 --> 00:11:24.535
that if workers have closer interaction,

236
00:11:24.915 --> 00:11:28.455
the racial coworker composition becomes less important.

237
00:11:29.355 --> 00:11:32.415
And this finding is consistent with the contact hypothesis

238
00:11:32.415 --> 00:11:35.215
that predicts lower racial biases when group

239
00:11:35.215 --> 00:11:36.935
members frequently interact.

240
00:11:38.825 --> 00:11:41.305
Additionally, we also look at, differences

241
00:11:41.365 --> 00:11:42.785
by the time the DCS

242
00:11:42.785 --> 00:11:46.425
and the income workers have worked together, we find

243
00:11:46.425 --> 00:11:49.025
that the negative retention effects are largest

244
00:11:49.325 --> 00:11:50.465
and only significant

245
00:11:50.955 --> 00:11:54.165
where the income has less tenure alongside the

246
00:11:54.365 --> 00:11:55.485
deceased worker in the firm.

247
00:11:56.575 --> 00:11:58.005
Those results are consistent

248
00:11:58.005 --> 00:12:00.805
with the previous mentioned heterogeneity on

249
00:12:00.815 --> 00:12:02.205
occupation and teamwork.

250
00:12:02.975 --> 00:12:05.355
As workers interact more with each other,

251
00:12:05.565 --> 00:12:08.435
their ties become less racially segregated

252
00:12:08.575 --> 00:12:11.595
and thereby they, thereby they're less likely to respond

253
00:12:11.795 --> 00:12:14.075
to whether the departing workers of the same

254
00:12:14.295 --> 00:12:15.395
or of a different race.

255
00:12:17.035 --> 00:12:20.215
So all of the results above suggest that people tend

256
00:12:20.215 --> 00:12:22.095
to have preferences to work with peers

257
00:12:22.125 --> 00:12:23.695
that are more alike like them.

258
00:12:24.205 --> 00:12:26.735
However, that those effects can be reduced

259
00:12:26.755 --> 00:12:30.915
to a higher degree of interaction with teams and the paper.

260
00:12:31.105 --> 00:12:33.315
Then we also look at the gender dimension

261
00:12:33.495 --> 00:12:36.635
and how the loss of a female versus a male worker affects

262
00:12:36.915 --> 00:12:38.195
retention as well.

263
00:12:38.215 --> 00:12:41.515
How the loss of a non-white female versus a non-white male

264
00:12:41.585 --> 00:12:42.595
affects retention.

265
00:12:43.305 --> 00:12:46.275
Without going into too much detail here, we see

266
00:12:46.275 --> 00:12:48.515
that the groups more strongly affected

267
00:12:48.815 --> 00:12:50.195
are non-white females.

268
00:12:50.805 --> 00:12:53.275
Their retention is affected by both the racial

269
00:12:53.695 --> 00:12:56.035
and the gender composition of their coworkers.

270
00:12:56.255 --> 00:12:59.435
Our study shows that when experiencing a decline in the

271
00:12:59.675 --> 00:13:01.275
presence of same race colleagues,

272
00:13:01.665 --> 00:13:04.355
non-white workers are more likely to quit their job.

273
00:13:04.975 --> 00:13:06.675
And this effect appears to be driven

274
00:13:06.735 --> 00:13:09.995
by non-white workers homophilic preferences for working

275
00:13:09.995 --> 00:13:11.115
with same race peers.

276
00:13:12.335 --> 00:13:14.015
Together our results highlight

277
00:13:14.075 --> 00:13:17.415
how coworker preferences can generate post hiring peer

278
00:13:17.415 --> 00:13:21.335
dynamics that amplify racial segregation across workplaces.

279
00:13:22.585 --> 00:13:25.885
Our findings also highlight significant heterogeneities

280
00:13:25.905 --> 00:13:28.685
and the effect of racial peer group composition.

281
00:13:29.425 --> 00:13:32.005
So studying only a small subset of firms

282
00:13:32.185 --> 00:13:33.445
or bundling different race

283
00:13:33.505 --> 00:13:36.365
and gender groups together may not provide a full picture

284
00:13:36.385 --> 00:13:38.005
of peer effects in the workplace.

285
00:13:39.185 --> 00:13:41.365
The results therefore underscore the importance

286
00:13:41.365 --> 00:13:44.485
of post hiring dynamics in shaping labor market outcomes

287
00:13:44.505 --> 00:13:46.205
for non-white and white workers.

288
00:13:47.305 --> 00:13:50.445
So policy aimed at promoting diversity, for example,

289
00:13:50.505 --> 00:13:53.725
should not only focus on eliminating discrimination in the

290
00:13:53.725 --> 00:13:57.005
hiring, but also considering coworker interaction

291
00:13:57.055 --> 00:14:01.405
after hiring and future research could probably examine the

292
00:14:01.405 --> 00:14:04.165
role of firm staffing and promotion policies

293
00:14:04.225 --> 00:14:05.685
and reducing peer effects

294
00:14:05.715 --> 00:14:07.845
that accelerate racial inequalities.

295
00:14:08.705 --> 00:14:10.485
Our findings, for example, suggest

296
00:14:10.485 --> 00:14:13.125
that forcing stronger entrepreneur interpersonal

297
00:14:13.355 --> 00:14:17.085
relationships within teams could reduce racial biases in the

298
00:14:17.085 --> 00:14:17.405
workplace.

Racial Peer Effects and Workplace Segregation in Brazil

00:00:00 - Our study extends racial peer effects in Brazil and how they might drive racial segregation. And racial segregation is evident in many labor markets, including Brazil, which serves as the context of our study. For example, in Brazil, 80% of white workers colleagues are also white. While non-white workers have only about 50% of white colleagues and several social science theories suggest that people have homophilic preferences. In other words, they prefer others who are like themselves. For example. Similar attraction can lead to preferences for working with demographically similar peers.

00:00:44 - of cooperation and communication within groups who share fewer similarities, thereby a decrease in the presence of racial similar peers may lower job certification and make workers look for a different job. And large part of the existing literature primarily focus on the hiring process, asking questions like, are managers more likely to hire workers who are, who share characteristics as the manager who hires? And those questions mostly concern the labor demand side. So really what employers seek and demand when hiring a worker.

00:01:25 - from the University of Potsdam and I focus on what happens after the hiring process. Specifically, we examine how employees react to changes in the racial composition of their peer group, particular in terms of the likelihood of remaining in their current job. this approach allows us to explore the labor supply side employees behavior after changes we can distinguish between quits, which reflect employees active decision to leave and layoffs which are initiated by the employer. So basically the key question we posed at the outset

00:02:05 - with coworkers who are racially similar to themselves? We analyze the impact of an unexpected change in the racial composition of peer groups on workers' likelihood of staying at the firm. Probably the most relevant body of work for us is the literature on peer effects in the workplace. And most of the economic literature actually focus on focuses on skills and productivity of workers and their impact on peers. For example, how does working with a more productive or less productive colleague affect other employees? It's more the interdisciplinary research that looks at the impact of demographic similarities in peer groups, often finding

00:02:46 - of similarity experience less turnover. For our study, probably the most relevant work is from Lenos and co-authors from 2024 who analyze random team assignment in a High wage US professional service firm and find that a higher share of white peers increases the turnover of black women. And most studies on demographic composition and its effect on coworkers, focus on single firm settings and mostly in high income countries. So we see our study introducing a novelty by examining the causal effect of peer group composition across a large set

00:03:29 - and across different sectors and location. But we also see ourself contributed to the literature on cross race interaction and racial segregation in various social contexts. For example, the literature that tests the so-called white flag hypothesis. For example, how do housing and school choice decisions of whites are affected by the presence of blacks? While we extend the literature by focusing on the workplace, we also added by focusing not only on the majority group, but analyzing also how the minority group responds to changes in cross race exposure.

Post-Hiring Racial Dynamics in Brazilian Labor Markets

00:04:09 - to the literature on racial racial preferences and discrimination in the labor market. I already mentioned that many scholars focus on discrimination in the hiring process of different, of different managers. And if different managers have preferences to hire white or non-white workers, if they also are white or non-white. We, however, in our study, focus on the post hiring stage looking at worker separation, especially quits and how those quits respond to respond to changes in racial composition of coworkers. Racial inequality is substantial.

00:04:48 - non-white workers earn on average 30 to 35% less than white workers. And even after accounting for differences in education and occupation. Those disparities persist due to differences in wage settings, practices and hiring. And probably we also need to mention a few words on the racial notion in Brazil. Brazil's official statistics classifies race into five different categories. Barranco for white, creto for black, pado for brown, Amarillo for yellow, and indigenous for indigenous. In our study, we combine creto and pado, so black and brown into one category called non-white.

00:05:32 - therefore capture 99% of the population. Our data comes from matched employer employee records for formal dependent employees in Brazil, and we use the years from 2006 to 2021. This data set, which is called ice, includes information about workers' race, occupation, contract type, and how their contracts were terminated, for example, through Quits or through layoffs. And it is important to mention Brazil separation regulations. Workers that are laid off are eligible for certain benefits like unemployment, insurance and severance payment provided

00:06:13 - history requirements. Of course, however, these benefits are not available to workers who actively quit their job. This distinction is important in our setting because our study D differentiates between quits and layoffs. So really quits are the results of employees actively choosing to leave their job bearing the financial consequences of this decision. So now to test whether workers prefer to work with others who are more alike, we look at how workers react when their racial peer group composition suddenly changes and to exploit,

00:06:54 - peer group composition. We focus on the unexpected death of workers. We compare two peer groups that have the same initial non-white share. Let's say for example, both peer groups have a non-white share of 30%, and in one peer group a non-white worker dies. And in another one, a white worker dies. By only comparing peer groups with the same initial non-white share, we see that the remaining in-car characteristics do not predict the race of the deceased worker. In other words, we have two groups in both peer groups. One worker dies and the race of the deceased worker can be taken as random.

00:07:39 - what effect causes this change in the racial coworker composition on the retention of the remaining income and workers. And while we do several sample restrictions in order to ensure the credibility of our results, I want to highlight two. First, we only exclude death that are not related to workplace accidents. Since such incidents could influence coworkers feelings of job safety, for example. And second, we only focus on small peer groups that have no, not more than 30 workers. As the loss of only one worker in a larger group would not

Impact of Non-White Peer Death on Racial Composition and Worker Retention

00:08:21 - we then use cross-sectional regressions to analyze retention effects during the year of the death and the three following years before we come to the main effect, which will be the effect on retention. I want to talk about the effect on the non-white share. So we assess whether the death of a non-white worker leads to lasting change in the racial composition of a peer group. We find that when a non-white worker dies, the non-white share of this peer group dropped significantly by almost 20 percentage points in the year of the death. And also over the following three years,

00:09:01 - percentage points lower than before. With that in mind, coming to our key findings that are that non-white income and workers are significantly more likely to leave their job after the death, death of a non-white coworker compared to when a white coworker dies. White coworkers on the other hand do not have similar response to changes in the racial composition. So those results highlight that peer effects are not symmetric across racial groups and that non-white workers are more strongly affected by the racial composition of the peers than white workers.

00:09:41 - to shed light on the mechanisms. First, we find that the observed effects are driven completely by employee initiated quits rather than employer initiated layoffs suggesting a labor supply response instead of a labor demand effect. Second, we track workers after they separate from the initial firms for the subset of workers who start a new final job in the same year or the year following. After the separation, we compare the characteristics of the old and the new job. We find that those non-white workers that experience the death of a non-white peer move

00:10:24 - share than the old job. So overall, those results suggest are a consistent with the idea of homophilic preferences for working with more same race peer, that drive the initial separation of non-white workers. And finally, we also look at different heterogeneities. First, we find that the effects are only present in white color occupations and not in blue color occupations patients. And in addition, we see that the effects are weaker in jobs that involve more teamwork. We draw two lessons from that. First, it seems that racial peer effects are more pronounced in higher

00:11:06 - that our average effects are smaller than the average effects found in high income settings, for example, in the us. And second, the teamwork Heterogeneities suggests that teamwork can kind of offset the negative effects of neuro retention, meaning that if workers have closer interaction, the racial coworker composition becomes less important. And this finding is consistent with the contact hypothesis that predicts lower racial biases when group members frequently interact. Additionally, we also look at, differences by the time the DCS and the income workers have worked together, we find

00:11:49 - and only significant where the income has less tenure alongside the deceased worker in the firm. Those results are consistent with the previous mentioned heterogeneity on occupation and teamwork. As workers interact more with each other, their ties become less racially segregated and thereby they, thereby they're less likely to respond to whether the departing workers of the same or of a different race. So all of the results above suggest that people tend to have preferences to work with peers that are more alike like them. However, that those effects can be reduced

Gender and Racial Dynamics in Worker Retention

00:12:31 - Then we also look at the gender dimension and how the loss of a female versus a male worker affects retention as well. How the loss of a non-white female versus a non-white male affects retention. Without going into too much detail here, we see that the groups more strongly affected are non-white females. Their retention is affected by both the racial and the gender composition of their coworkers. Our study shows that when experiencing a decline in the presence of same race colleagues, non-white workers are more likely to quit their job. And this effect appears to be driven by non-white workers homophilic preferences for working

00:13:12 - Together our results highlight how coworker preferences can generate post hiring peer dynamics that amplify racial segregation across workplaces. Our findings also highlight significant heterogeneities and the effect of racial peer group composition. So studying only a small subset of firms or bundling different race and gender groups together may not provide a full picture of peer effects in the workplace. The results therefore underscore the importance of post hiring dynamics in shaping labor market outcomes for non-white and white workers. So policy aimed at promoting diversity, for example,

00:14:17 - hiring, but also considering coworker interaction after hiring and future research could probably examine the role of firm staffing and promotion policies and reducing peer effects that accelerate racial inequalities. Our findings, for example, suggest that forcing stronger entrepreneur interpersonal relationships within teams could reduce racial biases in the workplace.

Faculti AI can make mistakes

Research

Policy

Analysis