The Impact of Social Segregation on the Labor Market Outcomes of Low‐Skilled Workers*

• DOI: http://doi.org/10.1111/sjoe.12324
• Date: 01 January 2020
• Published date: 01 January 2020
• Author: Gergely Horváth

Scand. J. of Economics 122(1), 3–37, 2020

DOI: 10.1111/sjoe.12324

The Impact of Social Segregation on the

Labor Market Outcomes of Low-Skilled

Workers *

Gergely Horv´ath

Xi’an Jiaotong Liverpool University,215123 Suzhou, China

horvathgergely@gmail.com

Abstract

Westudy the impact of network homophily on labor market outcomes in a search-and-matching

model with two job search channels: the formal market and social contacts.There are two worker

types: low-skilled and high-skilled workers.The homophily level determines whether the referral

networks of the two types are mixedor segregated from each other. We showthat there exists an

intermediate homophily levelthat minimizes the unemployment rate and maximizes the wages of

low-skilled workers.Complete integration does not maximize the welfare of low-skilled workers,

unless it improves their productivity. Weargue that our model can explain the empirical findings

on the labor market effects of the Moving-to-Opportunity experiment and the integration of

immigrants.

Keywords: Homophily; integration of immigrants; Moving-to-Opportunity experiment; social

networks; social segregation

JEL classification:D85; J31; J64; J71

I. Introduction

There is a large body of literature showing that social networks are a

frequently used job search method (Granovetter, 1995; Bayer et al., 2008;

Bentolila et al., 2010; Glitz, 2017), and it is believed that employed

social contacts are more useful for obtaining a job than unemployed social

contacts. If workers with high unemployment probability, such as ethnic

minorities and low-skilled immigrants, are isolated in the social network

from others with better labor market outcomes, their disadvantages are

exacerbated by the lack of social capital (Calv´o-Armengol and Jackson,

2004, 2007). The social integration of these worker groups is believed to

decrease their unemployment rates and increase their earnings.

In this paper, we revisit these ideas and show that changing the

social network position of disadvantaged workers might not be sufficient

*I wouldlike to thank Matthew Jackson, Kiss Hubert Janos, Marco van der Leij, three anonymous

referees, and participants at the Summer School on “Social interactions and urban segregation”

(Rennes), the Annual Meeting of the SOLE (Raleigh), and the Annual Meeting of the EALE (St

Gallen) for their comments on this draft. All remaining errors are mine.

C

The editors of The Scandinavian Journal of Economics 2018.

4On the impact of social segregation

to improve their labor market outcomes, as social integration must be

accompanied by spillover effects that increase the productivity of workers.

We build a search-and-matching model with two worker types: low-skilled

workers with a low productivity level and job opportunities only in the

low-skill sector of the economy, and high-skilled workers with a high

productivity level and job opportunities in both the high- and low-skill

sector of the economy. Workers can find a job either in the formal

market or through their social contacts where it is assumed that only

employed contacts are useful for job searching. The two worker groups

differ in the characteristics of their social networks, more specifically,

in the number and types of neighbors. We parametrize the degree of

interconnectedness between the two groups, which represents the degree

of social segregation/integration or the network homophily level. When the

homophily level is high, low-skilled workers are mostly connected to other

low-skilled workers. When the homophily level is low, the two groups mix

with each other.

First, we study the impact of the homophily level on the labor market

outcomes when segregation affects only the type of social contacts that

workers have and use for job search. We find that the relationship between

the homophily level and the job-finding rate of low-skilled workers is non-

monotonic. There exists an intermediate homophily level that maximizes

the job-finding rate of low-skilled workers. The intuition for this result is

as follows. When the homophily level is small, low-skilled workers will

mostly be connected to high-skilled workers, while high-skilled workers

will be connected to a pool of low- and high-skilled workers. This implies

that the unemployment rate among the neighbors of high-skilled workers

will be larger than the unemployment rate among the neighbors of low-

skilled workers. Consequently, there will be more competition among the

neighbors of high-skilled workers for the job information they possess. We

show that, for this reason, a high-skilled worker is less likely to transmit

job information to a given social contact of theirs than a low-skilled worker,

despite the fact that a high-skilled worker is more likely to be employed.

Low-skilled workers can thus improve their chances of finding a job by

severing links to high-skilled workers and establishing links to other low-

skilled workers. In contrast, when the homophily level is high, low-skilled

workers are mostly connected to each other, so they are less likely to

be employed and to possess job information. Low-skilled workers can

improve their job-finding chances through social integration. There exists

an intermediate homophily level that maximizes the job-finding chances of

low-skilled workers.

We solve the model for plausible parameter values and we show

that the unemployment rate of low-skilled workers follows a U-shaped

pattern in the homophily level and that it is minimal for an intermediate

C

The editors of The Scandinavian Journal of Economics 2018.

G. Horv ´ath 5

homophily level. The unemployment rate of low-skilled workers can be

the same in a highly segregated society and in an integrated society.

Moreover, wages earned by low-skilled workers follow an inverse U-shaped

pattern in the homophily level, and they are maximized by an intermediate

level of network homophily. This is because wages are determined by

Nash bargaining; they are influenced by the outside options of workers,

which are best when job-finding chances are highest. In addition, we find

that high-skilled workers prefer a highly segregated society where their

unemployment rate is low and wages are high. Desegregation does not

make either of the two groups better off.

Our findings substantially change when we assume that integration

not only affects the job-finding opportunities through social contacts but

also increases the productivity of low-skilled workers. In this case, the

unemployment rate of low-skilled workers decreases with integration and

their wages monotonically increase. We also find that high-skilled workers

do not benefit from integration in this case, either. As the productivity

level of low-skilled workers increases, firms post more vacancies in the

low-skilled sector of the economy, which increases the mismatch rate of

employed high-skilled workers. This negatively affects their welfare, despite

decreasing their unemployment rate.

We argue that the model explains various findings in the empirical

literature about the labor market effects of social integration. The first

application of the model concerns the Moving-to-Opportunity (MTO)

experiment. Between 1994 and 1997, in five US cities, the MTO experiment

gave residents of high-poverty neighborhoods the opportunity to relocate to

low-poverty neighborhoods. Households eligible for public housing were

randomly assigned to one of three groups: the first group received housing

vouchers that could be used in census tracts with a less than 10 percent

poverty rate; the second group received housing vouchers that could be

used without restrictions; and a control group received no assistance.

The neighborhood-effects literature (see Topa, 2001; Bayer et al., 2008)

shows that individuals living in the same neighborhood have correlated

labor market outcomes due to the social network effects of finding a job.

This would imply that MTO participants should be able to improve their

labor market outcomes by building connections to residents of low-poverty

neighborhoods. Kling et al. (2007), Ludwig et al. (2012), and Chetty et al.

(2016) evaluate the experiment’s impact on labor market outcomes and find

no significant improvement in the wages of those who moved to low-poverty

neighborhoods as adults.1In contrast, Chetty et al. (2016) find substantial

1Note that these papers found significant treatment effects of the MTO experiment along other

dimensions, such as improvements in mental health, decreased crime rates, improvedschooling,

(see Kling et al., 2005; Ludwig et al., 2011).

C

The editors of The Scandinavian Journal of Economics 2018.