‘Lfiwhzm.

.u.

... 75K. ... 2,

. $9.”:

»

 

 

his
(9
3004
(2052"?307

This is to certify that the
dissertation entitled

THREE ESSAYS ON CAUSES OF SKILL, RACIAL AND
ETHNIC LABOR MARKET DIFFERENCES

presented by
LINDA ANDREA BAILEY

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in ECONOMICS

 

 

 

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

— —........-A-v-----I-u-O.-pu-o-o-o-n---»-o-.-o----o-c-a-u-c-o-n-u-on

c.-------c--.-.— .

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 cz/C-IfiC/DateDueJndd-pjs

 

THREE ESSAYS ON CAUSES OF SKILL, RACIAL AND ETHNIC
LABOR MARKET DIFFERENCES

By

Linda Andrea Bailey

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Economics

2004

ABSTRACT

THREE ESSAYS ON CAUSES OF SKILL, RACIAL AND ETHNIC LABOR
MARKET DIFFERENCES

By

Linda Andrea Bailey

Labor market differences by skill, race, and ethnicity persist in the US. This
dissertation investigates possible influences on labor markets of less-skilled workers and
racial and ethnic labor market gaps. First, it asks how major immigration reform affected
relative labor markets. Second, it asks how labor markets of blacks are affected by
relatively higher arrest rates. Finally, it asks how immigrant status affected displaced
workers’ labor market re—entry.

Chapter 1 examines the 1986 Immigration Reform and Control Act (IRCA)
impacted less-skilled native workers’ relative labor markets. I allow the Act’s effect to
vary by race. The Act allowed approximately three million illegal immigrants to apply
for legal residence in the US. Previous work examining the Act’s effects has not
allowed varying racial or skill impact. I find that IRCA had differential effects by skill
level and race. These effects were either short-lived or inconsequentially small in
magnitude.

Chapter 2 examines the relationship between higher relative arrest rates for blacks
and their relative labor market outcomes. I find that cities in which blacks have higher
relative arrest rates are those in which they have worse relative hourly wages. After

controlling for city-specific effects, there is no evidence that higher arrests lead to worse

relative labor markets for black males. After controlling for city specific effects, blacks
are still assigned to lower status occupations the higher their relative arrest rate.

Chapter 3 examines the effect of immigrant status on labor market re-entry for
displaced workers. Immigrant status positively affects the chances of being re-employed
and lowers earnings losses among workers who make a full-time to full-time job
transition. The lower earnings losses among immigrants appear to be due to their
concentration in the low end of the earnings distribution. Possible explanations for the
higher employment probabilities of immigrants are examined. Immigrants are more like
to switch occupations when re-employed. I do not find support for the hypothesis that
differential access to government income support or non-labor income induced
immigrants to search more intensely for work. Additional differences among immigrant

groups are examined.

To my parents, Francis and Pauline, and my siblings, Bernie and Stephen, who have

loved and supported me every step of the way to the fulfillment of this dream.

iv

ACKNOWLEDGEMENTS

I owe a great debt of gratitude to my chair, Jeff Biddle, for his inexhaustible
patience and guidance. Without his unwavering support throughout my tenure at
Michigan State this dissertation would not have been possible.

I would like to thank David Neumark, with whom I began this dissertation
journey for an invaluable learning experience. I must acknowledge for their time,
meetings, comments, and frankness, Jeff Wooldridge, Steven Haider, and Stephen
Woodbury. Their guidance, support and contributions as my committee have enabled me
to complete this journey and are greatly appreciated.

For friendship, late night company, discussions, repeated reading of this work at
different stages and much more, special thanks are due to Scott Adams, Gregory Dean,
Troy Hicks, Jongbyung Jung, Paula Kazi, Daiji Kawaguci, Lebohang Lijane, Elena
Nizalova, Elda Pema and Gifford Rankine. They read or viewed parts of this dissertation
in various forms (sometimes multiple times) and provided helpful feedback.

It is not possible to thank all the people who have positively impacted my life
while in East Lansing, but special mention must be made of the Gebes family who
provided a home away from home and much needed emotional support and
encouragement.

Finally and most of all I would like to thank God and my family, who have
always and who continue to support me in all my endeavours with their love, prayers,
encouragement and faith that with God’s help this dissertation and in fact all things are

possible.

TABLE OF CONTENTS

LIST OF TABLES.. ......... . .......... . .................................... . viii
LIST OF FIGURES... ....... .. ..... . ........... ................. x
CHAPTER I: DID THE IMMIGRATION REFORM AND CONTROL
ACT AFFECT RELATIVE LABOR MARKET OUTCOMES? ..... 1
1.1 Introduction ................................................................ 1
1.2 Potential Impact on the Native Labor Market. ....................... 3
1.3. Data and Empirical Strategy ............................................ 6
1.3.1 Data ................................................................... 6
1.3.2 Empirical Strategy ................................................. 11
1.3.3 Limitations of Data and Methodology ........................... 14
1.4 IRCA’s Effect on Labor Market Gaps ................................ 15
1.4.1 Wages ................................................................ 16

1.4.2 Annual Income, Usual Weekly Hours, Annual Weeks,
Probability of Non-Employment and Living in a Poor

Family ................................................................ 17
1.5 IRCA’s Effect by Race ................................................... 18
1.5.1 Wages ................................................................ 19

1.5.2 Annual Income, Usual Weekly Hours, Annual Weeks,
Probability of Non-employment and Living in a Poor

Family ............................................................... 20
1.6 Conclusion .................................................................. 22
References ........................................................................... 39
Appendix A: Practical Estimates of IRCA’s Impact ........................ 42

CHAPTER 2: CRIMINAL JUSTICE OUTCOMES AND RELATIVE
LABOR MARKETS OF BLACKS 44

2.1 Introduction ................................................................. 44
2.2 Theoretical Framework .................................................. 48
2.3 Data .......................................................................... 51
2.4 Empirical Methodology .................................................. 56

2.4.1 Related Issues ...................................................... 57
2.5 Results ....................................................................... 59
2.6 Conclusion .................................................................. 61
References ............................................................................ 65
Appendix A: Socio-Economic Index for 13 Occupation Categories.......68
Appendix B: Relative Arrest Rates by Year and City ....................... 68

vi

Appendix C: Comparison of arrest rates for blacks relative to whites with
those for 16-45 year old black males compared to similar white males..70

CHAPTER 3: THE IMPACT OF JOB DISPLACEMENT ON

3.1 Introduction ............................................................... 72
3.2 Conceptual Framework.................................. ................. 75
3.3 Data ........................................................................... 77
3.4 Results ..................................................................... 78
3.5 Why Might Immigrants Be Hurt Less by Displacement:
Exploring the Results ................................................... 80
3.5.1 Earnings on the Old Job ........................................... 81
' 3.5.2 Labor Market Flexibility of Immigrants ........................ 82
3.5.3 Government Income Support, Non-Labor Income
and Search Intensity .............................................. 82
3.6 Heterogeneity among Immigrants ...................................... 87
3.6.1 Citizenship ........................................................... 88
3.6.2 Region of Origin and Immigrant Type .......................... 88
3.6.3 Ethnic Networks .................................................... 89
3.7 Conclusion .................................................................. 92
References ............................................................................ 101
Appendix A: Weekly Earnings Change by
Immigrant Status: 1994-2002 ..................................................... 105
Appendix B: Participation Rates in Government Income Support ~
Programs and Access to Non-Labor Income ................................... 106

vii

1.1b

1.2

1.3

1.4

1.5

1.6

1.7

1.8

A1.l

2.1

2.2

2.3

A2.1

A22

A23

3.]

LIST OF TABLES

PROP and Dependent Variable Means by
Time Period-Education—PROP Group Male Sample ............................... 28

PROP and Dependent Variable Means by
Time Period-Education-PROP Group Female Sample ............................ 29

IRCA's Effect on Log Hourly Wages: Coefficients Estimates
from Equation (1.1) ............................. , ........................................ 3O

IRCA Effect on Outcomes: Coefficients Estimates
from Equation (1.1) ................................................................... 31

IRCA Effect on Outcomes: Coefficients Estimates
from Equation (1.1) Using Later Post-IRCA Data. ................................ 32

IRCA's Effect on Probability of Living in a Poor Family:
Estimates from Equation (1.1) ......................................................... 33

IRCA's Effect on Log Hourly Wages by Race: Estimates from Model (2). ...34

IRCA Effect on Outcomes by Race: Estimates from Model (2) ................ 35
IRCA Effect on Outcomes by Race: Estimates from Model (2)

Using Later Post-IRCA Data ......................................................... 37
IRCA's Estimated Impact for selected MSA's and Outcomes .................. 43
Mean of Relative Independent and Dependent variables ........................ 62
OLS Estimates of the Effect of Relative Arrest Rates on Relative Labor
Market Outcomes of Blacks Compared With Whites ............................ 63
Fixed Effect Estimates of the Effect of Relative Arrest Rates on Relative
Labor Market Outcomes of Blacks Compared With Whites ..................... 64
Socio-Eoonomic Index for 13 Occupation Categories ............................ 68
Relative Arrest Rates of Blacks to Whites in Each Year .......................... 68
Mean of Relative Arrest Rates for Blacks Compared to Whites ................. 71
Mean of Selected Variables by Immigrant Status ................................. 96

viii

3.2

3.3

3.4

A3.1

A32

A33

Effect of Immigrant Status for Workers Displaced from 1993-2001 ......... 97

Coefficients from Specifications Exploring the Reasons for Better Immigrant

Outcomes ............................................................................... 99
Coefficients of Regressions Estimating Immigrant Heterogeneity ............ 100
Weekly Earnings Change by Immigrant Status: 1994-2002 .................... 105

Portion of Weighted Sample Receiving Unemployment Insurance
Benefits by Immigrant Status and Sex ............................................. 106

Coefficients fiom Estimates of Government Income Support Use,
Access to Non-Labor Income and Implied Search Intensity Rank ............ 107

ix

1.1

1.2

1.3

1.4

1.5

3.1

LIST OF FIGURES

IRCA’s Impact on the Unauthorized Labor Market........... 24

IRCA’s Impact on the Authorized Labor Market........... 24

Distribution for Sample of Men in 1991 and 1992 CPS ......................... 25
Status of LAW and SAW Amnesty Program Applicants ........................ 26
Spike over 1989 through 1993 Reflecting Increased Legal

Immigration due IRCA ............................................................... 27

Ethnic Population Share Distribution .............................................. 95

CHAPTER I

DID THE IMMIGRATION REFORM AND CONTROL ACT
AFFECT RELATIVE LABOR MARKET OUTCOMES?

1.1 Introduction

The Immigration Reform and Control Act (IRCA), passed in October 1986, aimed
to reduce the number of undocumented immigrants in the US. in two ways. First, IRCA
enacted penalties for employers who hired illegal aliens or who failed to check legal work
status before hiring, in order to deter prospective illegal immigration by removing the
lure of jobs. Second, IRCA provided amnesty for qualified illegal aliens and the
opportunity for them to obtain legal immigrant status. Amnesty was granted in
recognition of the commitment of these immigrants to residence in the US. and for some
their attachment to the US. labor market.

The general immigration literature has recognized that differences in skill level
and race matter when measuring the impact of immigration on domestic labor markets.1
The skill level of the immigrant group may have differential impact on the US. labor
market depending on the skill distribution of the natives in the market. Previous
literature exploring the IRCA’s impact has not allowed the effect to vary by skill, race
and legal status while using a measure directly related to IRCA. This paper adds to the
current literature by exploring a question which carmot be answered from previous work:
has the IRCA—which largely increased the legal labor supply with Hispanics of
relatively low education levels—worsened the relative position of disadvantaged

American?

 

' See Card (2001) and Jaeger (1996).

While many previous studies (see for example Crane et a1. 1990; Juffras 1991;
Phillips and Massey 1999) have investigated the impact of IRCA on various aspects of
the US. labor market, exploration of IRCA’s impact on native groups by skill and race is
incomplete. Sorensen and Bean (1994) and Bucci and Tenorio (1997) examine the
effects on native workers by looking at changes in wages before and afler IRCA’s
passage. Sorensen and Bean look at the log of real wages of males with Mexican
ancestry, as well as native and immigrant white males. They find that there were no
significant changes in the wages of white male natives or white male immigrants after the
IRCA. Bucci and Tenorio investigate changes in the immigrant-native wage gap
potentially caused by IRCA’s employer sanctions provisions. Discrimination against
immigrants is hypothesized to occur if employers view expected IRCA fines as a tax on
immigrant labor, but not on native labor. They find that the increase in the native-
immigrant wage gap is primarily due to increased nepotism towards natives rather than
increased discrimination against immigrants. Neither paper uses a direct measure of
either the amnesty or sanctions. Analyses such as these that rely solely on the passage of
time for their identification strategy are subject to the criticism that concurrent changes
offer alternative explanations for the changes occurring in the labor markets under study.

Therefore, the empirical strategy I propose here uses both wage changes and a
direct measure of the potential impact of the IRCA amnesty. Cobb-Clark, Shiells and
Lowell (1995) (hereafter CSL), and Fry, Lowell and Haghihat (1995) (hereafter FLH),
examine IRCA’s impact on average wages at the metropolitan area-industry level. Both
papers look at the impact of the employer sanctions and amnesty components of IRCA

separately, using regional measures of average monetary fines and the ratio of annual

amnesty applications to metropolitan area population, respectively. Both papers find that
higher relative amnesty populations are associated with higher wages.2 The industries
were chosen for their high immigrant participation rates and the metropolitan-level wage
data used averages across workers of different skill level, race and legal status.

Though the IRCA was passed in 1986, a full understanding of its impact remains
relevant in light of on-going support for further such amnesties by the business
commrmity and labor unions, which both benefit from immigration.3 Though the IRCA
amnesty was supposed to be a one-time occurrence, an additional 6 million people have
taken advantage of several amnesties after 1986 and more are likely (Briggs, Jr., 2003).4
Most recently, a policy similar in magnitude to the IRCA amnesty has been proposed to
regularize the estimated 5 million plus undocumented immigrants currently residing in

the US 5'6
1.2 Potential Impact on the Native Labor Market

Along with the introduction of penalties for employers of illegal labor, IRCA
introduced two programs under which illegal aliens could apply to regularize their status.
First, the legally authorized worker (LAW) program, also called the general amnesty,
allowed applications from illegal aliens who had resided in the US. since January 1982,

thereby showing a commitment to long-term residence. Second, the special agricultural

 

2 Both papers also instrument for the amnesty populations. The stated result is robust.

3 The AF L-CIO openly states its support for further amnesty of illegal workers (see
http://www.aflcio.org/aboutaflcio/ecouncil/ec0731a2001.cfin, site checked January 11, 2003).

4 In fact, Chau (2001) models the decision of whether or not a country should grant an amnesty at a point in
time and finds, under costly internal enforcement of immigration laws, intermittent arrmesties are optimal.

5 October 1996 estimates of the illegal population is between was between 4.6 and 5.4 million persons and
growing at a rate of over '/4 million persons per year (Source: “Illegal Alien Resident Population,” USCIS
accessible http://uscis.gov/graphics/shared/aboutus/statistics/illegalalien/).

6 The current President has proposed a legalized work visitor program reminiscent of the Bracero program

worker (SAW) program allowed illegal aliens who had worked in agriculture in seasons
prior to the IRCA to apply for amnesty.

The employer sanction and amnesty provisions targeted the undocumented
population. Nevertheless, natives could also be affected. The employer sanctions should
have reduced the demand for undocumented workers. The downward shift of the demand
curve for undocumented labor from Du to Du' is shown in Figure 1. I follow CSL and
view fines as a tax on unauthorized labor. Thus, Du' represents the demand curve net of
expected fines: Figure 2 shows the corresponding upward shift of the documented labor
demand curve for low-skilled labor, D3 to Da'. As hypothesized by CSL, if the SAW
program induces new undocumented migrants, this would shift the supply curve for
unauthorized workers further to the right, Su to S,,'.7

Based on this model, effective employer sanctions should increase demand for
less-educated authorized workers to fill new vacancies that would otherwise be filled by
illegal immigrants.8 In the extreme case that undocumented labor only occupied jobs that
natives did not want, sanctions should not affect native outcomes. Papademetriou et al.
(1991 p.42) use data from a survey of employers and note that firms that report regularly
hiring unauthorized workers were “9.5% more likely to increase wages to attract
authorized workers” than other firms. Hiring fines are imposed on firms that
“knowingly” hire illegal workers. FLH find significant positive wage effects related to

larger hiring fines in a metropolitan area. This casts doubt on the idea that illegal and

 

7 New illegal aliens, who otherwise may have remained in the home country, are thought to have entered
the U.S. in order to fraudulently apply for amnesty (see Designations of Temporary Protected Status and
Fraud in Prior Amnesty Programs: Hearing before the subcommittee on Immigration and Claims
http://commdocs.house.gov/committees/judiciary/hju5987 1 .OOO/hju5987 l_0f.htm).

8 IRCA’s employer sanctions did not affect existing illegal workers who stayed with the same employer.
Only new hires were affected by the law.

legal workers are in completely separate markets. Notably, FLH find that the lower
wages resulting from fines levied on employers for not checking work authorization
offset the effects of hiring fines.

The amnesty provision increased the supply of authorized workers by giving legal
status to millions of illegal aliens. However, undocumented workers may still confine
themselves to the lower segment of the labor market solely because of their legal status.
Thus, changes in legal status may spell increased competition for authorized workers as
amnesty recipients move out of marginalized jobs.9 In Figure 2, S. shifts to 83' after the
amnesty. As these persons were initially unauthorized, there. should be an initial
reduction in the supply of undocumented workers, Su'shifts to Su".

Depending on the relative size of the increased competition and demand shifts,
declining labor market outcomes for natives may or may not be evident. Since there are
opposing effects on wages, the direction of IRCA’s impact is an empirical question.

In addition to the aforementioned, IRCA could affect labor market outcomes of
native blacks particularly strongly. Since non-Hispanic blacks and Hispanics have
greater occupational similarity than non-Hispanic whites and Hispanics, increased
competition for low-skilled jobs from the newly-legalized residents could have a greater

‘0 Keeping in mind, however, blacks may have

negative impact on blacks than whites.
competed with undocumented workers prior to the amnesty; employers who now comply
with IRCA would hire only authorized workers, perhaps improving the labor market

opportunitiesof these legal minorities. Most likely, this would be apparent in the market

for less-educated labor as undocumented workers generally fall in this category.

 

9 Kossoudji and Cobb-Clark (2000) find evidence of upward occupational mobility of the newly legalized.
'° See Bansak and Raphael (2001) and Altonji and Card (1991).

Similarly, if Hispanics and non-Hispanic blacks are competing for the same jobs and
documented Hispanics experience IRCA-induced discrimination, blacks may experience
better labor market outcomes. Thus, the effect on blacks relative to other groups is

ambiguous and is a major focus of this study.

1.3 DATA and EMPIRICAL STRATEGY

1.3.1 Data

To convincingly examine the effects of the amnesty, a direct measure of its
potential impact to a locality provides, arguably, the best tool. The 1990 Legalization
Summary Tapes, created by the Immigration and Naturalization Service (INS), provide
one.11 These data contain the IRCA amnesty applicants’ county of intended residence.
Due to confidentiality requirements, data on the actual number of amnesty recipients in
an MSA is unavailable. Nevertheless, over 90% of applications were ultimately
approved.12 Hence, the Summary Tapes give a reasonable measure of the actual size of
the legalized population. I divide the number of MSA applicants under the general
amnesty program by the 1982 population in the corresponding MSA.l3 The resulting
variable becomes the proportion of a city’s 1982 population that applied for the amnesty
(hereafter PROP).

For data on native labor market outcomes and demographic features, I use the

1986, 1987, 1991, and 1992 Current Population Survey (CPS) March files for the main

 

" I am extremely grateful to Deborah Cobb-Clark for providing this data.

'2 See Chart 1.4 for denial rates. Although, standard processing of LAW applications was done regionally,
there is no reason to believe that denials systematically varied by MSA. Juffras (1991 pp. 58-60) discusses
the legalization process, including denial rates. He notes that initially restrictive document requirements
were challenged in court and overturned. Thus, the later processing allowed for liberal documentation
resulting in a denial rate of l in 10,000 applications in the second phase of the general legalization
ro

3 The 1982 population figures are based on interpolation between the 1980 and 1983 MSA populations
from the 1985 Statistical Abstract of the United States.

analysis. These files provide labor market information by skill and race groups fi'om the
previous year.'4 1985 and 1986 data identify the relative pre-amnesty position of
different skill and racial groups in cities with higher amnesty populations.15 The latter

6

years provide post-amnesty labor market data.1 Applications to take advantage of the

amnesty under the general legalization program were accepted from May 5, 1987 to May
4, 1988. The SAW program had an official deadline of November 30, 1988.17 Thus,
prior to 1987 there should be no legalization effects. Applicants obtained temporary
work status, which was revocable if amnesty was denied. Thus, they did not have the
same legal work status as permanent residents until later.

Chart 1.5 presents the inflow of legal immigrants from 1900-1996. The large
increase in legally admitted immigrants in 1990 and 1991 relative to trends in prior years
was due to the legalization applicants. Approximately 74% of LAW applicants and 5%
of SAW applicants had achieved permanent residency status by the end of the fiscal year
1990. By the end of the fiscal year 1991, over 2.4 million people had gained permanent
residence under both programs. This comprised approximately 90% of those who would
eventually be granted permanent residence.18 Hence, at this point, much of the amnesty
population had the same legal labor market standing as a native.

Ideally, this policy analysis should take into account the effect on all authorized
workers, but legal status is not available in the CPS data used here. As a result, I choose

to examine a subset of authorized workers, natives. Unfortunately, in the years of data

 

'4 These ‘before’ years are chosen as number of MSAs identified increased dramatically from the 1986
March files onward. In addition, some MSA definitions changed at the same time as the increase. Thus,
using data from earlier files would make comparisons with data from after the IRCA problematic.

'5 These data are in the 1986 and 1987 cps March files.

'6 CPS Outgoing Rotations did not identify as many cities as the CPS March files until 1989.

'7 Late filings and court challenges to restrictions on those who qualified for the amnesty allowed for
further applications.

'8 See 1991 INS Statistical Yearbook pp. 70-1.

used for this analysis, native status is also not available.19 I, therefore, use a proxy for
nativity, non-Hispanic black and white status. Thus, the sample excludes anyone
declaring Spanish ethnicity and retains those reporting race as black or white. This is
reasonable as tabulations from the 1990 PUMS 1% sample20 indicate that, nationally,
about 90% of non-Hispanic whites and non-Hispanic blacks report that they were born in
the U.S. Around 50% of those with Spanish ethnicity and approximately 18% of non-
Pacific Islander Asians report being U.S. born. This holds even in high immigrant cities.
In the New York and Los Angeles MSAs, approximately 90% of non-Hispanic whites
still report that they are U.S. born, while 81% of non-Hispanic blacks in New York and
88% of those in Los Angeles report that they are U.S. born. For those claiming Spanish
ancestry, those figures fall to 45% for New York and 30% for Los Angeles. The figures
for non-Pacific Islander Asians are below 20%.

Hispanics would confound the effect of the IRCA because they could experience
discrimination or penalties for illegal status after 1986 or if they are amnesty recipients,
they would earn a return to legal status.21 By limiting the sample to non-Hispanics, I
essentially avoid the problems associated with expected differences in the IRCA’s impact
by legal status. In addition to ethnicity and race restrictions, due to the interest in labor
market outcomes, my sample is limited to the working population, aged 18 to 64 years.

Only residents of MSAs identified in the CPS March files are included. I exclude states

 

'9 CPS Supplements in 1983, 1986, 1988 and 1989 used by Sorensen and Bean (1994) contain immigrant
status, but there are fewer MSAs identified in these supplements than in the March files until 1989.

2° This data set contains demographic information on a 1% sample from the 1990 U.S. Census.

2' Newly legalized workers have been documented to experience high wage growth, while those remaining
illegal have experienced wage penalties. See Kossoudji and Cobb-Clark (2000, 2002) and Phillips and
Massey (1999).

in New England.22 As New England counties are not strictly contained in one MSA, the
county level data in the Summary Tapes cannot be reliably allocated to a particular city.

The hourly wage provides one assessment of the labor market impact of the
IRCA. However, examination of changes to hourly wages does not adequately address
whether the native population is better or worse off. If, for instance, employers shift from
full time workers, who have expensive benefits, to more highly paid part-time workers
without such benefits, we may optimistically misinterpret the labor market conditions
after IRCA. We may see average hourly wages increase, but lower annual incomes, as a
result of the reduced hours. This work therefore utilizes the CPS data to examine a
broader range of outcomes.

The CPS March files provide data for the previous year on annual wage income,
usual hours worked per week, weeks worked per year, employment status, and family
poverty level. These outcomes are investigated here. With the exception of log hourly
wages, sample sizes utilize all those with information whose hourly wages do not exceed
$100 (1982-1984 dollars). Only positive log hourly wages are used for the wage
regressions. Hourly wages are calculated by dividing annual labor income by annual
weeks times usual weekly hours. In order to calculate log annual labor income, 0.00001
replaces zeros for those with no wage income reported. The U.S. Consumer Price Index
for all Urban Consumers (CPI-U) adjusts wage and income data to 2003 dollars. Of note,
some MSAs had no applicants; the Los Angeles-Long Beach metropolitan area had the
highest PROP with approximately 8 amnesty applicants out of every 100 residents in

1982.

 

22 New England states are Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and
Vermont. City boundaries/groups rather than county groups define New England MSAs. There are 23
MSAs, which would constitute 8.6% of the sample, that are excluded because of this restriction.

Table 1.1 gives mean outcomes, for males and females, before and after IRCA for
the three education groups examined: high school dropouts (HDOs), high school
graduates without a four-year college degree (HGs) and college graduates (CGs). The
means are disaggregated by a dichotomous version of the variable used to identify the
IRCA’s effect: MSAs with relative low amnesty population shares (low-PROP MSAs)
and those with relatively high shares. Low-PROP areas are those with amnesty shares at
or below the median MSA share of 0.00046.

Comparing hourly wages before and after IRCA for low-PROP MSAs for the
male sample, we can see that all education groups had lower real wages after 1987. The
same is true for high-PROP MSAs. If we examine wages before and after IRCA to
determine its effect, we could be erroneously attributing negative changes to the policy.
On the other hand, a comparison of wages in low-PROP and high-PROP MSAs after
1987 reveals that the latter have higher average wages for all education groups.
Examination of pre-1987 figures, however, shows that high-PROP areas have historically
high wages. Thus, a comparison of low-PROP and high-PROP MSAs only might give
the impression that the policy increased hourly wages. To calculate the IRCA’s effect,
we should control for initial conditions. We can do this by comparing wage changes
between high and low-PROP MSAs before and after 1987.

However, if changes over time in low-PROP cities differed systematically from
those in high-PROP cities, we could once again make erroneous conclusions about the
IRCA’s effect. In addition, amnesty recipients belong mainly to the lowest education
category, HDOs. There could be competitive effects from IRCA on HDOs, who are the

most similar in human capital terms, while complementary effects could be exerted on

10

higher education groups. For these reasons, a third level of differencing to capture
differences in MSA time trends is desired.
1.3.2 Empirical Strategy

My approach is similar to CSL and FLH’s in that the identifying variable is at the
MSA level. The variable differs slightly as they use annual LAW applications divided by
the MSA population in the year of application, whereas PROP, the variable used here is
the total stock of applications under the program divided by the 1982 population. PROP
has no time dimension, varying only across cities. In order to examine the effects in the
years when most LAW applicants became permanent residents I cannot use their
measure. Data on admissions of amnesty recipients as legal permanent residents by year
and metropolitan area is not available. Additionally, the use of the later population
figures in the denominator also biases the results towards positive findings when
compared to the original measure. Cities with better labor markets may have growing
populations, decreasing PROP in areas with better labor markets.

Using a ‘difference-in-differences’ approach, I examine the aforementioned labor
market outcomes and identify the desired effect by interactions between lower education
group indicators for HDOs and HGs, PROP and a post-IRCA indicator. Along with
metropolitan area fixed effects, the interaction of PROP with the lower education groups
control for the relative position of HDOs and HGs in cities with larger amnesty
populations, prior to IRCA. These terms interacted with the post-IRCA indicator identify
the change in the relative position of less—skilled workers due to the IRCA. PROP
interacted with a post-RCA indicator captures any spurious correlation between changes

in cities with higher amnesty populations and labor market outcomes.

11

The following regression estimates the effect of the newly legalized on the change
in the relative labor market outcomes of the less-educated (hereafter Model 1):23
Y,,, =Xijt B] + AA, B; + HDOiJ-t [33 + HG”. [34 + HDOUC" PROP,- 135 + HGijt* PROP,- [36 +
HDOij-WAAI [37 + HG,,-.*AA. [38 + PROPfAA, Bg + HDO,-j.*PROPJ-*AA( 1310 +
HGijt*PROPJ-*AA(BH + BLACKithn + BLACK“, *AA, [313 + MSAj + 8,3, (1.1)
where, Yijt denotes various labor market outcomes (log hourly wage, log annual income,
hours worked per week, annual weeks, probability of non-employment, probability of
living in a poor family). Xijt includes indicators for marital status, age, age-squared; AA!
denotes the after the amnesty indicator; HDOiJ-t high school dropout indicator; HGiJ-t high
school graduate-incomplete college indicator; PROPj proportion of MSA amnesty
Applicants to 1982 MSA Population; BLACKijt black indicator; MSAJ- metropolitan area
fixed effects; e,,-, — random error with an expected value of zero and independent of the
above controls.

Metropolitan area fixed effects control for permanent differences in localities that
may be correlated with the size of the amnesty population, e.g., the historic attractiveness

of the city to immigrants.24

In terms of the regression, this requires omitting PROP in
level form, as it is perfectly collinear with the metropolitan area fixed effects.

The coefficient [39 captures local labor market changes that may be unrelated to
the amnesty, but which occurred in higher PROP cities after IRCA. Bio and I311 give the
relative effects of the legalized population on the change in the gaps between HDOs and

HGs and their more educated counterparts.25

 

23 Standard errors are robust to variation across metropolitan area pre- and post-IRCA observations.

2‘ The MSA fixed effects are estimated by a dummy variable regression.

2’ An alternate method would be to have fixed effects interacted with low education and post-Amnesty
indicators, followed by an examination of changes in the return to education based on different sizes of the

12

The estimated IRCA effects are not subject to the argument that the amnesty
population located in areas with better labor markets for such workers. Intuitively, the
general amnesty was directed at immigrants that were considered to have long-term
commitments in terms of residency, having lived in the US since 1982. So it is
reasonable to assume these applicants were already settled by the period of the labor
market data used here. Moreover, unlike SAW participants, they were not noted for
highly mobile behavior in response to job opportunities.26

The hypothesis being explored here is that the relative position of less-skilled
workers may be affected by the legalization of low-skilled illegal aliens. CGs would pick
up general changes in local labor market demand conditions across cities that may
confound the effect of IRCA.27 Additionally, CGs would capture any scale effects for the
additional paper work costs associated with IRCA sanctions. Although CGs are unlikely
to be illegal, IRCA requires all employers to have INS I-9 forms on record for staff.28
Thus, unless employers do not fill out I-9 forms for CGs, the new IRCA paperwork costs
are expected to affect all education groups similarly. CGs do not constitute a control
group as they, too, are expected to be affected by labor demand or supply shifts of less-
educated groups. Our primary interest here, however, is whether the approximately 3
million persons legalized by the IRCA contributed to the widening wage gap between

skilled and unskilled workers over the 19803.29

 

Amnesty population. This would be similar to method used by Reimers (1998) to study the effects of
immigration. The drawback to this method is that small sample sizes for some education and/or race
groups severely limit the number of cities that could be used.

6 SAW applicants moved from state to state depending on seasonal work (INS 1991 p.71).
27 Although, Topel (1994) finds “there is no evidence that different regional evolutions of wages are
demand-driven; the whole story is on the supply side.”
28 L9 forms require proof of work authorization.
2” See Katz and Murphy (1992).

13

If areas with higher amnesty populations experienced negative demand shocks,
relative to low-PROP cities, then studying only less-educated workers could yield
misleading results. So the third-level of differences better addresses the question of how
IRCA changed the relative labor outcomes of less-educated workers compared to CGs.

To address potential additional labor market competition for blacks, I further
interact the PROP terms with race-education indicators, allowing the effect of [RCA to
vary across race (hereafter called Model 2). Thus, the post-IRCA changes of each race-
education group relative to the post-IRCA changes of white CGs are estimated.

In addition to the main analysis, I explore CSL’s hypothesis of increased
competition as the legalized population acquires additional human capital. Using the
1993 and 1994 CPS March files, I re-estimate Models 1 and 2.30 Finally, separate
regressions are run for males and females.

1.3.3 Limitations of Data and Methodology

1 do not control for sanctions effects. However, as mentioned earlier, FLH find
that the effects of sanctions are small and offsetting. Furthermore, in the most recent
literature on the effects of IRCA, there appears to be a consensus that fiaudulent
documents, combined with inadequate enforcement, render employer sanctions

ineffective regarding its main intent—deterring the employment of illegal aliens.3 I

 

3° Besides allowing the investigation of later effects, this allows me to address a potential problem with
1991 and 1992 data. Recall that the March files provide data on the previous year’s labor market
outcomes. Since the period from July 1990 to March 1991 was an economic recession, (see Business Cycle
Dating Committee of NBER at http://www.nber.org/cycles.html) the relative labor market outcomes of
less-skilled workers could be negatively affected independent of the newly legalized. In contrast, March
1991 through March 2001 is recorded as a period of expansion. Thus for the later years of data, we can
expect increasingly tight labor markets, which are not expected to negatively affect low-skilled workers
compared to others.

3' Sec Legomsky (1998) for a detailed discussion of employer sanctions. Passel (1998 p. 198) also
discusses the lack of effectiveness of sanctions in achieving their deterrence aim

14

While the metropolitan area fixed effects included in the regression control for the
overall effect of immigration, they may not account for the impact of general immigration
on the relative position of less-educated workers. As the deviation in the legal
immigration trend is almost entirely due to the amnesty, there is a basis for not
accounting for other immigration in an examination of the amnesty effects.32

This work uses only applications under the general amnesty program. Of the
approximately 3 million IRCA amnesty applications, the SAW program had over 1.2
million, with over 1 million approved for temporary residency. Migration by SAW
applicants from agricultural to non-agriculture jobs should imply greater negative effects
than LAW applicants, most of who already worked and lived in cities. Arguably, any
analysis of competitive effects should account for their impact. Their inclusion is
problematic as SAWs may bias estimates towards more positive findings.33 Prevailing

labor market conditions may be important in the location decision for this group.
1.4 IRCA’s Effect on Labor Market Gaps

To address the question of what effects the IRCA had on less-educated workers
relative to college graduates, Tables 1.2 through 1.5 give the estimated coefficients from
equation (1.1) for variables measuring IRCA’s effect. Coefficients for less-educated
groups show IRCA-induced changes relative to CG changes. The coefficient on the
PROP-post—IRCA (AA) interaction term controls for general post-IRCA changes in
higher PROP MSAs. Negative coefficients on the interaction between the post-[RCA
indicator, the education group indicators (HDO and HG), and PROP would support the

hypothesis that the effect of the increased labor supply of low-skilled workers dominated

 

32 See again Chart 1.5.
33 The more detailed version of this paper includes results accounting for SAW applications. Those results
show less evidence of a negative impact as expected.

15

the increased demand due to employer sanctions. A positive coefficient would indicate
that higher demand for less-educated native workers outweighed the increase in
competition. In other words, it may reflect success in reducing employer demand for
illegal workers or discrimination against Hispanics. Significant effects are discussed
below. Long-term effects examined by using later post-IRCA data than the main analysis
are also discussed. See Appendix A Table A1.1 for calculations of implied effects for
selected outcomes in representative low-PROP, middle-PROP and hi gh-PROP MSAs.
1.4.1 Wages

First, I explore the estimated effects with the log hourly wage as the dependent
variable by asking the question, “Do the earnings of less-skilled workers exhibit a relative
decline in areas with larger newly legalized populations?” Table 1.2 columns (1) and (2)
show the estimated effects of IRCA on the log hourly wage gap between less-educated
workers and CGs under Model 1 assumptions. Under Model 1, the effect is constant
across race.

For both sexes, the post-IRCA effect of higher legalized populations on the
relative wages of HGs is positive and significant at the 10% level. These estimates imply
that if an additional person out of every 100 members of the city’s population changed
legal status, an increase of 0.91% in the average hourly wages of male HGs relative to
male CGs is expected. For female high school graduates, the increase is about .85% per
1/ 100 increase in the population share that changes status. Practically, these estimates

imply very small changes for two reasons. First, a 1/100 increase in the amnesty share is

16

quite large and therefore unlikely.34 Second, an increase of 1% in the average wage of
female HGs would only come to about 13 cents per hour for this sample.

Tables 1.2 columns (3) and (4) examine the hypothesis of CSL that acquisition of
skills by the newly legalized population may exert competitive effects in years
subsequent to their analysis35--the years immediately after IRCA passed. This analysis
with data from later post-IRCA years shows a significant decline in the relative log
hourly wage of male HDOs compared to male C63 in cities with higher legalized
populations. In addition, the later data shows that higher legalization rates are associated
with lower hourly wages in the MSA. This contrasts with the findings of the previous
literature which uses earlier data.

1.4.2 Annual Income, Usual Weekly Hours, Annual Weeks, Probability of Non-
Employment and Living in a Poor Family

As mentioned before, the hourly wage—while a good measure of the labor market
position of a group—may not give a full picture of IRCA’s impact. In Table 1.3, log
annual income, usual hours worked per week, annual weeks worked, and the probability
of non-employment are examined for both sexes. Again, IRCA’s effect is restricted to be
the same across race within education categories. The sample for these regressions uses
all persons satisfying the race, ethnicity and residence restrictions and not just those with
positive log hourly wages. IRCA did not significantly affect males, but it did
significantly lower weekly hours worked for female HGs relative to female CGs.

Long-terrn effects in Table 1.4, based on the later post-IRCA years, reveal general

declines for all less-educated males relative to CGs, with significant relative declines in

 

3‘ Only 35 of 224 MSAs used in this analysis had amnesty population shares in excess of 0.01.
3’ CSL use data up to 1989. These years immediately follow the IRCA, but do not encompass the years in
which most of arrmesty recipients became permanent residents.

17

usual weekly hours and annual weeks worked of male HGs in the post-IRCA period in
cities with higher legalized populations. For males, both HGs and HDOs see a significant
increase relative to CGs in the probability of non-employment in cities with higher
legalized populations after IRCA. In these later years, less-educated females do not
experience any adverse effects on their relative position after IRCA.

Table 1.5 columns (1) and (2) show that less-educated members of both sexes are
more likely than CGs to live in poor families, but the estimated effects are insignificant.
In the later post-IRCA years, Table 1.5 columns (3) and (4) show that IRCA significantly
increased the likelihood of living in a poor family for less-educated persons relative to
CGs. Again, these data are a couple of years after most of the amnesty applicants had
received permanent residency status. If CSL’s hypothesis is true, the competitive effects
should be stronger at this later stage.

It is of note that Tables 1.2 through 1.5 show significant negative effects of larger
legalized population shares on CGs. Again, this implies that cities with relatively higher
amnesty populations experienced general declines in their labor markets. These results
suggest that aggregation in the previous literature where higher legalized population
shares were associated with higher wages masked differing effects. Possibly, as demand
for native-appearing HGs increased in cities with large illegal populations, substitution

effects forced employers to reduce demand for CGs.

1.5 IRCA’s Effect by Race

To address the question of whether the IRCA effect differed by race-education
group, Tables 1.6 through 1.8 give the estimated coefficients from Model 2 which allows

the IRCA’s effect on the relative position of each race-education group relative to white

18

CGs to vary. Coefficients for all race-education groups give IRCA changes relative to
CGs’ changes. Significant effects of the IRCA on black-white labor market gaps for
similarly educated workers are discussed along with notable findings on changes to the
within-race return to education.36

1.5.1 Wages

Model 2 allows black education groups to be differently affected than whites.
Table 1.6 shows that IRCA induced a narrowing of the wage gap between white HGs and
white CGs. Columns (1) and (3) show no changes in the status of black males relative to
white male CGs. So the earlier estimate indicating that IRCA led to a narrowing of the
gap between HGs and CGs is driven by improvements in the position of white male HGs
relative to white CGs. Similarly, while estimated effects for HDOs of both races are
negative, it is only white male HDOs who see a significant decline in their position
relative to white male CGs. In later years, IRCA significantly reduces the wage gap
between female HGs and white female CGs. White female HDOs also saw their relative
wages increase compared to white CGs.

In their investigation of possible discrimination against Hispanics due to IRCA,
Bansak and Raphael (2001) use blacks and whites as comparison groups. Though they
use earlier years of data for their analysis, they found post-IRCA declines in Hispanic
wages relative to black wages, but no declines relative to white wages. Their sample is

limited to southwestern states, which had high numbers of amnesty recipients. The post-

 

” Tests for significant changes in the within black return to education or in return to education across race
are not shown.

19

IRCA changes in the black-white hourly wage gap (not shown) related to the size of the
amnesty population offers a possible explanation for that result.37

1.5.2 Annual Income, Usual Weekly Hours, Annual Weeks, Probability of Non-
Employment and Living in a Poor Family

Table 1.7, which again allows for IRCA’s effect to vary by race, shows that the
annual incomes of white male HDOs increase relative to white male CGs in cities with
higher amnesty populations. Black male HDOs experience declines in their annual
income when compared to white male CGs, but the change is insignificant. However, as
the gap between white CGs and I-IDOs narrowed, while the gap between white CGs and
black HDOs widened, the annual income gap between white HDOs and black I-IDOs
increased significantly due to higher legalized populations, post-IRCA. The decline is
significant at the 1% level. An increase in the amnesty applicant population share of a
city of l/ 1000 reduced average black male HDOs’ annual income, relative to average
white HDOs, by about $1077.38

While there are significant positive effects of IRCA on black male CGs’ position
relative to white male CGs in weekly hours, annual weeks and the probability of
employment, these do not translate into significant declines in black-white log annual
income gaps. Possibly, black male CGs and white male CGs are equally qualified, but
blacks have lower reservation wages in cities with high amnesty populations. If there
were general labor market downturns in cities with higher amnesty populations after

1987, cost-conscious employers may have substituted more expensive white CGs with

 

37 Model 2 with black-PROP interaction, but no education-PROP interactions show an insignificant
narrowing of the black-white wage gap in cities with higher legalized populations.

3’ I estimated wage effects based on an increase in the amnesty population share of 1/ 100 as the concern
about immigration is about high levels of irnrrrigration. An increase of 1/ 1000 may be more germane as out
of 224 MSAs in the sanrple on 35 had a PROP value greater than 0.01.

20

black CGs. Another possibility is that the IRCA increased demand for natives at lower
levels of the occupation structure, causing increases in relative wages of HGs. This may
have had an indirect effect on firm demand for high-skilled workers. Unexpectedly, black
male CGs are the only group significantly more likely than white male CGs to live in
poor families, post-IRCA, in cities with higher legalized populations. Possibly, the
distribution of the gains to blacks was highly skewed.

Table 1.7 shows that, for the female sample, the significant decline in relative
hours worked by HGs is driven by a decline in weekly hours of white HGs relative to
white CGs in higher PROP cities. Black CGs show relative as well as absolute
improvements in annual income and weeks worked in cities with higher legalized
populations when compared to white CGs in such cities. These improvements led to
widening annual income and weeks worked gaps in cities with higher legalized
populations between black CGs and black HGs. Less-educated black females were
significantly less likely to live in poor families before 1987 than their white counterparts.
The only group that became significantly more likely than white CGs to live in poor
families in cities with higher legalized populations is black female HGs.

Despite generally negative effects on annual income, IRCA lowered the
likelihood of living in poor families for less-educated workers. As the poverty index is
based on total family income and not just labor income, the results may reflect
differences in the importance of non-labor income.

From Table 1.8, using later post-IRCA years, all significant results indicate a
negative impact on labor market outcomes of less-skilled males relative to male CGs due

to IRCA. These changes do not lead to significant declines in log annual income for less-

21

educated workers when compared to CGs in cities with higher arnnestied populations.
For females, the gap between the annual incomes of black HGs and white CGs narrowed
in cities with higher legalized populations.

1.6 Conclusion

The 1986 Immigration Reform and Control Act was the culmination of years of
efforts to address the problem of illegal immigration. It was a two-pronged approach
aimed at reducing both the incentive to migrate for potential immigrants and figuratively
“wiping the slate clean” for persons already entrenched in the U.S. This study examines
a hitherto unasked question of how less-educated workers—those most similar in human
capital characteristics to the amnesty recipients (mainly high school dropouts)—fared
after IRCA relative to more highly-educated workers in cities with higher legalized
populations. It also investigates whether blacks are differently affected given the greater
similarity of their occupational distribution to that of Hispanics, who constitute the
majority of amnesty applicants.

In sum, I find positive post-IRCA effects of higher legalized populations on
wages of high school graduates relative to college graduates of both sexes during the
years when the greatest number of LAW applicants gained permanent residence. When
the IRCA’s effect is allowed to vary across race, white male and black female high
school graduates were the only less-educated group that had significant wage gains on
college graduates due to the amnesty in later post-IRCA years. Black female high school
graduates narrowed their annual income gap with white female college graduates (though

white female high school graduates did not).

22

Black male high school dropouts’ annual incomes were negatively affected
relative to white high school dropouts in the years that most amnesty recipients obtained
permanent resident status. In later years, there are significant negative effects on the
hourly wages of male high school dropouts and a general worsening of outcomes for
black male high school graduates relative to white male high school graduates. This did
not translate into lower annual income effects though. Negative post-IRCA effects of
higher legalized populations on the relative position of less-educated blacks compared to
black college graduates were also found. Thus, concern about the impact of fiu'ther
amnesties on low-skilled blacks is warranted.

There are positive IRCA effects of higher legalized populations on relative labor
market outcomes of high school graduates, who comprise a large part of the labor force.
These findings suggest that—when legalizing undocumented persons through amnesty—
there are deleterious effects on black males and male high school dropouts generally. In
the absence of a credible effort to enforce employer sanctions, further amnesties may
increase competition for existing authorized workers without any of the benefits, such as

increased demand for low-skilled workers, that came with the 1986 IRCA.

23

Figure 1.1 IRCA’s Impact on the Unauthorized Labor Market
it

Wage 8 u" Su'

 

 

tr

Units of unauthorized labor

Figure 1.2 IRCA’s Impact on the Authorized Labor Market

4

Wage $

 

 

Units of low-skilled authorized labor

24

Figure 1.3 Distribution for Sample of Men in 1991 and 1992 CPS

 

 

 

.6 T
.4 ‘
Fraction of
Sample
.2 “‘
_, E - _ I- g
0 I l l I l
0 .08

.04 . .06
Legalized Pop. Share in MSA

25

x0350.» Eosmsfim m7: 3% ”ooenom

......szzau;

 

 

 

 

\..\.\.....\\\\
. KlKKK.

 

 

 

1.42:9? B

2.32:... be 1252 B

”3:59.. I

 

L’\\\'\\\\
I.“

1H'\

 

r.“.VI“‘.‘V.\V.V
.-

O.

H::\'\’\\

.\.
\
.\
.\_
...\
..\
.\
.\
.\
\
..\
..\
...\
.\
..\

I

..:.‘\>\

'\ :\

.\.\\\\.\_\\\\\\.

o e . . e e e e . .

.\\.\\\.\.\\...\\\\.
.\\\\\\\.....\\\.\

\\.\\......\...\\\\...\\

\IEH...\\\\....\\\\\

.\
\...................\\\\\\K\\\
\

....\.\\....\\.\\.\

\.s.\.\\\\\\\\\

\\.\\\\\\\\\\

,.\........\\\x\\\\..\.\.x.\

..\ ..\ .\ ..\. .\ .\ \\ ..\ -\ .\ .\.

u.\ n.‘ a.‘ a.\ I\ n\ 0‘ a.‘ a‘ -\ o.‘ a
..\ ..\ .\ .\ .\ .\ .x \ .\ ..\ ..\.

..\ \ \ ..\ ..\ .\ \ .\ .\ \ .\ .A
..\ .\ .\ .\ ..s .\ \ .\ .\ .\ .\.
.\ ..\ ..\ ..\ ..\ \ ..\ .\ ..\ \ \\.
..\ .\ \ ..\ \. \ .\ \ .\ \ ...\
.\ ... .e .x .t .x \K \ .\.
..\ -.\ ..\ \\ ..\ .\ ..\ ~\ ..\ -.\ \ .

\\.\\\\\\\\\\

bllebb-L.LLLK

 

 

 

 

f .23.

1 .xx...

.27...

 

£5.”
.2325

.3533? same; reuse. 3% e5 33 he seam E :25

26

202:3» Ecumsfim mg 32 ”8.58m

0..— g g g , .g g .0— 88 ‘9 Oua- ‘—

  

 

{Own 8 2:.
graham—55 3on 63.3.85 wage—mom mam“ £9.85 33 .82. exam m4 EEO

27

.3130 05 .3 mooN 3 cognate can mun—ED Am oo>o meme? bozo: £33 owes“ «0 29:3 5:55 e m_ min. _

 

 

 

owed aged Nod mod o.\tv adv embdm wwSN OD
392 o“ ~od eod o—d Ndv fiwm Nomdm ems.— OE
memd 2 dd 3 .o mNd wdm edm moN.: ewdfi OQE
mom.— .3:
Amendmen—
webd mm m od Nod ved vdv w. 3 www.mm deN 00
3%.: Va ed mod mod 0. 3 Nfim meJ m v53 03
weed oodd 3 .o NNd Ndm wdm 513 no.3 OQI
ROMA—
..wE $3.2m
mood Noood mod mod ofiv ode o :uov fiedm OD
Vmodfi Noood cod wed N. mv vdm emvdm . wed“ OE
god. Noood 2 .o emd h. ~ m flow oomfio vwdfi ODE
mOmh
..Sot— boo—lunch
ohwd Noeod mod vod ofiv Mame. we _ .Nm Eda OD
mow.o_ Noood eed wed m. :V 0.5m veNdm 3a.: 03
wmmd Noood Ed vmd Ndm odN oeo.: 3.: OD:
mome—
tBot— beauty—m
onm mOMm 338m poem echo—WEN.— mxoog modem Ge 5 9.9.0 A—Oy—m
295% n 5 wE>= So: 353 >333 08005 Wows? 1:323: cm—

:omtoaem :omtooem 3553.. 3.303

oEEew 232

 

macaw mOmx—taomuaosemteotom 05:. .3 mane—Z o_aa_..a> Eco—Soon can mOmm .amé £th

28

.DALU 0:. kn meN o. “Bash“... 0.8 mum—~03 . .m .25 new?» £30: 5.3 08:. mo 2ng 3.38m m m. mmah .

 

 

dqmfi vm.e.e med 2d edm N.mm thdm oodm OD

owed. e. .ed oed vmd ..vm 9%.. 3:6. emd. OI

evflm ...ed .md ovd ...N K..... befih Ned CD:

.5... gm...

. bwouuumoh

owed hN_ed med m—d ..om ..Nm oovdm th~ 00

N.m..m hmged ood mmd m.Nm edN .emfi. mod. 03

moms“. 5. .ed hNd wvd odN wfi. .ewd ewd ODE

mOflm

..m... 53.9...

Swim Neeed med 2d hdv 9mm wmodm mud. 00

26.2 Neeed .d mmd mém @SN mhmd— vw.: OZ

0.0.N Neeed .md de md. ed— .ww.m :d 003

mOM—n—

spon— bmouémom

moo.m Neeed med m—d mdm m. . m Nce€N mod. 00

mmmd. Neeed :d emd e.Nm wdm em..m— he... 03

mend Neeed and Ed ed. m6. .wwd mwd ODE

A—Ofim

$54 boo—no.5

0N5 mOME bmgm ..oom tome—mam 3303 @503 a. .8 9.9.0 how—A—

oiEwm a E wEZ. .0: Enac< >333 0:505 .mowmg $535035—
:omtoaokm comtomokm .m..:..< 3.503

 

035mm 2..th

 

9:20 mom..-:o=3=..m-.§.8m 2...... .3 2.82 «Eat!» .5323: ...... no... ...... «3a...

29

...EI... ”gm ...... :32 :2 2.. s Eoficwmm...

$2 Ea ~92 88. a: 5 Ba 5 223 .32 as. 82 ea. 3% a: 3 Ba 3 2:360
.momofico.ao E a... 33.2. boo—$0.. was -0... on. E 3%). $80.... meanest... o. 6.50. 28.6 Enocfim

 

 

mm... .2 a... £3 83%....
$2: $3. .83. fit... 8232030
...:5.8 .38 .808 :ote
a... 8... a... a... o: 23 8 5233 96
an: :5: an: at:
a... R..- who :3 on: Ba 8 8223 96
...:ave name 608 62:
a... who- who- one- 8
Barnum 2m: 0389* 222
cc 5 S 3

 

.3.5 .8333— ...o... 3......3mm 3.83530 ”mama? .228: m3 .... 80.5— 933: .NA 933—.

30

59:: 9% ...... :32 :2 2: a 9505:»?

.885539 5 2m 62.59 53.369 can -29 05 E m<m2 awoken moo—58¢? 8 9958 28.5 6.3ma%

 

 

 

 

 

 

NS N; m; N; 92295
«ANNA: vmm,mo~ :33 va.mo_ 3232030
398 83: :99; c 89.3
5o 8; 7 men- 93. 6: Ba 8 5253 96
698 $15 $28 $3:
mmd 3.3 3.8- no; ODE 28 DU 5953 95
28.8 3:32 3 8:5 160$
wmd Sdm- wwd- wde 00
2083—95 8.33 3:53 9:53 2:35
.5: mo b:5mno£ >28? .95 D 32:79 mod
oEEmm 2.25m
25 w; :.o 3 83:9,-”—
Sm€o $36 $36 $36 3233030
§ .8 623 $35 83¢
mmd 2: T No.5- 86 CI new 00 5953 95
39.8 833 $23 $2:
$6 moNN _m 03 on: 98 00 59,33 95
ES $98.3 .653 31.359
cod KAN- 3.3- 2.3- 00
9583—95 8.33 3:52 $303 2:35
.5: «G 3:53on >23? 3sz 3:53 we:
298mm 232

 

.A—Av act—«sam— Ecb 8353mm 3:32:30 "moEcouac .8 Baum <U~= .m.— 039—.

31

.913... maxi ...... :26. $2 05 “a Emu—MEET,

.momoficosn 5 2a 32qu $2.82. c5 -05 2.: E 392 $88 moocobbmc 8 6.58 38.5 235%

 

 

 

 

 

 

:5 2 .o 2 .o 2 .o 83:65
”2.2: 332 ”2.2: ”2.2: 2.23530
3&8 8mm: 63 3 83V
:5 3w- 2:. 3 cm as oo 5052. 95
30.8 8E5 828 G3:
”3 5:- 2.2- 3- on: as 8 5933 96
22.8 :5}: *3: : 123.8
33 G. _ m- :2- 3:- co
Suzie—95 8.33 3:52 9:5: 2:85 3:53 wag
-co: mo hum—Empohm 3x02,» 3sz
038mm 2.“th
m; w; 2 .c S 8333
218 2mg 2&8 218 823530
...:amé ...:Asawv 1.5.8 5.3
$6 3.3.. $3- :3- or Ba 00 5232. has
:3on $28 823 3::
M: 3m- 9%.. 3.2- on: Ba 8 5233 95
...: fl .3 ...:C. fl .3 :Lmofiv ......Abodv
moo 3m- :3. 8.2- 8
2083—95 8.83 3:52. was: 2:85 .352 we;
..co: mo hum—Bangs 3x003 :25:
038mm 252

 

.53. <U~=éoém .323 Win: 2.: nataaam SP... magma—3mm m.:o_qucU ”$5330 .5 Bob”.— <U~= .v.— 035.

32

$7....- mxm ...... :26. $8 2: a “308%?
.88 Ba 88 so... as A8 23 5 223 .38 Ba 88 ea... 5% 8: 3 Ba 3 8:530
.385323 E 0.8 32th baa-«mom can -29 05 E 3.32 328 moocob-GE 9 “mane Echo 23:me

 

 

w 8 .o 85 :6 So Bang-m
£38 at? 3.38 $35 823530
...-.38 ...-58 E8 58
£5 ”no N; 85 o: as 8 8233 96
...-at8 .58 A88 58
MS 35 Ed 25 on: Ba 8 5233 95
A _ _.8 ...-:88 £28 ...-.88
:3- 85 Mac mm... 8
2mth 2m: Baa-5..“— 2m:
3 5 3 E

 

.38 8:25 .55 3352mm “Esau .2... a 5.. was: ... Essen .8 .85 365 .3 2...;

33

.88.:- €% .... :22 £8 2: a 288%?
.82 Ba 88 63 as A8 Ea A8 223 .88 Ba 38 so; 9% a: 3 ea :8 2828
.3355th E 2w muotoa gal-mom EB .05 of E 352 $83 moocoho-u% 8 $38 Echo 688.8%

 

 

 

 

 

 

 

So one mg 38 8.23.8
32: m3? mmmfi £3: 228530
22.: $28 so: so:

84 Rd. N2 38 oz
and £08 5.8 8&8
8.8 2.- :4- 88 com
...GN: 8m. 8 am: ...-am:
EN 35 8. _- ”mm 8
oEEom 2&2 0:“th 232
mOU SE? 8 038—0.. axon—m co Bums;
...-998 :38 $8 ...-5.8
on; 23 N2 8.“ 0: Ba 8 8253 95
*5: ...SN. 8 am: 38
:2 SN- m _ .8 So on: as oo 5253 95

...-Amn8 23.8 :98 :38
3.7 :3- 88. 35. co
25:0,...— o—nz 25th 2me

3:55 no 63::
A8 5 S 3

 

AG .252 5...: 8353mm "35— .3 mung? 228: u?— =e 30...:— m.<0~= 6A 03:3.

34

3:: 2% ...... m.32 :8 05 a 288:9?
.momofiaoaa E 08 $2qu 32.52. 28 -oa 05 E 362 amp—om moocouobfi 8 #458 £25 3355

 

 

 

 

8.0 88 88 :8 3 Beam-m
Sag Sn: Sag 33$ $38 2232030
388 $8 $38 5.5 $8

5o 88. 3.2. 3.8. 8.» cm
5. c 5.: $38 $38 $38

8.8 m2 $.8- 8._ 3.2- cam
.288 $308 ...:GENV ...:am.~8 $5.:

$8 a. _- 8.8 838 m3- 8

900 333 3 033—2 flow—m :0 Hamma—
58 ~88 65.8 and: 2 m8

:3 2 m2- :3- :3 cm 2a 8 5223 96
58 A88 838 80.3 :23 3

So 88 8.8- 8.8 3.3 on: 2a 8 52:3 96

:88 58 ...-.258 $688 ...-:38
v2 85 3.3. 38- $.8- 8
Damn boom a “5:28.95 383 $30: 0802:

s 32% £5805 8&0 £385 .85. £83 33: .252 m3

 

3:55 so 83::

 

038mm 222

 

AS .255 8?: 83523— "35— .3 8.5830 .8 Batm— <U~= SA 933—.

35

8:... 8m ...... :32 :8 2: a 288.85...
$805538 5 03 $2.58 52.508 28 -28 05 E 392 $88 menses-amt 8 8.58 88.5 FEE—Sm

 

 

 

 

 

85 8d 8d 35 85
$38 $38 $.38 $38 «8.8:
1.618 $8 § .8 5.88 A83:

go mg. 5.0 8; 3

so: 6...: A858 A888 5.88
3 Rd. 88- 8.2 38
88 ...-Anne .338 5.88 1&3:
:.o 92. :fi ”3 6%
mva 0:55 9 03:22 332m :0 «owns:
a _ .8 A38 @988 .2888 £8
88. 8d 5.8 2.3. $5
:38 :38 $38 €38 A888

8.8 8d 8% 5.8- 3
1.: _.ov 8.1883 .....szmA 3 ANYNC ...:AmeuV

am... who , a. :1 8.x- v.8-

DEE-m “com a “conic—9:0 383 3:02 2:85
5 8%: mo $5305 .8: .8 $5305 .352 £83 .25 ESE was
$255 so H895

 

295m 2858

Bang-m
2232030

0:
OQE

DU

03 was 00 52,33 95
on: Ba 8 5233 95

OD

 

.5 .252 :3: magazmm 33m E 85830 .5 .85 <8: .8288 S 2...:-

36

557...... 55m ...... :96— 5\53 05 8 Enoufiwfi...
5805558 5 85 32qu 52.589 95 .2m 2: E m<m2 $28 $058ch 2 $38 $88 Emcefim

 

 

 

 

 

 

S5 25 5 _ 5 :5 :5 8353-5
2555 2555 . 25.8 2555 2555 823.530
3555 :555 £28.53 2.235: 355

Q5 55 M55- 55¢. 53. o:
AS. 3 an: €553 $553 625

:5- $5 on. _- NZ? 3. can
2 we @555 5.5 A2555 8::

$5 25- $55 52.5 55.3 8

500 SE? 8 255—2 3055 :o .0595
£255 iame 58555 .55: 5.3

$5 555 $55- 55.2- 32. c: 55 8 =8an 58
.555 £355 8558 A358 855:

52 2 $5- 3.x- 3.2- on: 55 oo 5233 58

**Ao_.ov in? _ .ov ***A©N.C :vadv 3.2."?ng
£5 25 55. :1 85m- :57 8
$83 “com 5 5583380 33>» 3:53 5:53 2:85
5 was: 55 $5385 -8: 55 $5385 £83 Ea: 3:52 53
3:55 so 8595
295m 252

 

5.5 «Gm—.53.— ..851— 9.7.3 ANV .352 :8...— moua—Eumm— "35— .3 52:32.0 .8 “ovum <95 .m.— 035,—.

37

3:: 3% ...... 52,2 .52 25 3 5305655...

5535:0an E 25 32qu 53.552. 5:5 -85. 05 E m<m2 3805 82.20556 8 338 £85 25.55%

 

 

 

 

 

£5 :5 25 m _ 5 N _ 5 85:35
55.52 5252 5252 5252 5252 82.3850
5.55 .2355 25.8 :55: .3555
:5 555- 5555 255 55.: on
85. a :5. c 55: . E58 35.3
:5 555 555- 85 53 cam
8555 3555 A55. _ a 255$ :55: _
555- $5- Em $5. 2 cu
v.00 BE? 3 255—8 8655 :o 35mg:—
.........9 :5 @555 A253 5.5: @555
:5 555 857 3.2- m: cm Ba 8 8052 58
@555 65.8 $553 £553 £33
£5 85. 85 525. 555- on: 2a 8 5232. 58
2 2: 5555 ......A 35: 35.5: 2.355
555 $5 5.2- 55.2- 85.- 8
385m boom 5 2883—95 8.83 _5:::< 3:0: 2:85
a 553:5 $5325 -855 £55325 £83 .25: 5352 53
$533 so 8595

 

295$ 258$

58 55-2.... .53 men: 5 .25: as. 8.55555 335 B 858:5 .5 .85 <8: 53:85 5.. 2.3.

38

References

AF L-CIO Executive Council Statement. 2001 . “Immigration.”
http://www.aflcio.org/aboutaflcio/ecouncil/ec0731a2001.cfrn checked January 11, 2003.

Altonji, Joseph and David Card. 1991. “The Effects of Immigration on the Labor
Market Outcomes of Less-Skilled Natives.” Pp. 201-234 in John Abowd and Richard
Freeman eds., Immigration, Trade, and the Labor Market. Chicago: University of
Chicago Press.

Bansak, Cynthia and Steven Raphael. 2001. “Immigration Reform and the Earnings of
Latino Workers: Do Employer Sanctions Cause Discrimination?” Industrial and Labor
Relations Review 54 (2): 275-295.

Briggs, Jr., Vernon M., “Immigration and Poverty Reduction: Policymaking on a Squirrel
Wheel.” Presentation at the Allied Social Sciences Association, January 4, 2003,
Washington, DC.

Bucci, Gabriella A. and Rafael Tenorio. 1997. “Immigrant—Native Wage Differentials
and Immigration Reform.” Review of Development Economics 1 (3): 305-23.

Business Cycle Dating Committee of the National Bureau of Economic Research. 2001.
Business Cycles Expansions and Contractions. Cambridge, MA.
http://www.nber.org/cycles.html.

Card, David. 1990. “The Impact of the Marie] Boatlift on the Miami Labor Market.”
Industrial and Labor Relations Review 43 (2): 245-257.

Card, David. 2001. “Immigrant Inflows, Native Outflows, and the Local Market Impacts
of Higher Immigration.” Journal of Labor Economics 19 (1): 22-64.

Chau, Nancy H. 2001. “Strategic Amnesty and Credible Immigration Reform.” Journal
of Labor Economics 19 (3): 604-634.

Cobb-Clark, Deborah A., Shiells, Clinton R., and B. Lindsay Lowell. 1995.
“Immigration Reform: The Effects of Employer Sanctions and Legalization on Wages.”
Journal of Labor Economics 13 (3): 472-98.

Committee on the Judiciary. 1999. “Designations of Temporary Protected Status and
Fraud in Prior Amnesty Programs: Hearing before the subcommittee on Immigration and
Claims.” Washington DC: GPO.
http://commdocs.house.gov/committees/judiciary/hju59871.000/hju59871__0f.htm Site
checked October 6, 2003.

39

Crane, Keith, Asch, Beth J ., Heilbrunn, Joanna Zorn, and Danielle C. Cullinane. 1990.
The Eflect of Employer Sanctions on the Flow of Undocumented Immigrants to the
United States. Santa Monica: Rand Corporation and Washington DC: Urban Institute.

Donato, Katherine M., Durand, Jorge, and Douglas S. Massey. 1992. “Stemming the
Tide? Assessing the Deterrent Effects of the Immigration Reform and Control Act.”
Demography 29 (2): 139-157.

Fry, Richard, Lowell, B. Lindsay, and Elhum Haghighat. 1995. “The Impact of
Employer Sanctions on Metropolitan Wage Rates.” Industrial Relations 34 (3): 464-84.

Jaeger, David. 1996. “Skill Differences and the Effects of Immigrants on the Wages of
Natives.” Bureau of Labor Statistics Economic Working Paper, number 273.

Juffras, Jason. 1991. Impact of the Immigration Reform and Control Act on the
Immigration and Naturalization Service. Santa Monica: Rand; Washington, DC: Urban
Institute.

Katz, Lawerence F., and Kevin M. Murphy. 1992. “Changes in Relative Wages, 1963-
1987: Supply and Demand Factors.” The Quarterly Journal of Economics 107 (1): 35-
78.

Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. 2000. “IRCA’s impact on the
occupational concentration and mobility of newly-legalized Mexican men.” Journal of
Population Economics 13: 81-98.

Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. 2002. “Coming out of the shadows:
Learning about legal status and wages from the legalized population.” Journal of Labor
Economics 20(3): 598-628.

Legomsky, Stephen H. 1998. “Employer Sanctions: Past and Future.” Pp. 171-190 in
Peter Duignan and Lewis H. Gann, eds., The Debate in the United States over
Immigration. Stanford: Hoover Institution Press.

Papademetriou, Demetrious, Bach, Robert L., Bill, Howard, Cobb-Clark, Deborah A.,
Krane, Roger 6., Lowell, Briant Lindsay, Shea, Martin and Shirley T. Smith. 1991.
Employer Sanctions and U.S. Labor Markets: Second Report. Washington, DC: U.S.
Department. of Labor, Bureau of International Labor Affairs.

Passel, Jeffrey. 1998. “Undocumented Immigration.” Pp. 191-204 in Peter Duignan and
Lewis H. Gann, eds., The Debate in the United States over Immigration. Stanford:
Hoover Institution Press.

Phillips, Julie A. and Douglas S. Massey. 1999. “The New Labor Market: Immigrants
and Wages after IRCA.” Demography 36 (2): 233-246.

40

Reimers, Cordelia W. 1998. “Unskilled Immigration and Changes in the Wage
Distributions of Black, Mexican American, and Non-Hispanic White Male Dropouts.”
Pp. 107-148 in Daniel Hamerrnesh and Frank D. Bean, eds. Help or Hindrance? The
Economic Implications of Immigration for African Americans. New York: Russell Sage
Foundation.

Sorensen, Elaine, and Frank D. Bean. 1994. “The Immigration Reform and Control Act
and the Wages of Mexican Origin Workers: Evidence from the Current Population
Surveys.” Social Science Quarterly 75 (1): 1-17.

Topel, Robert. 1994. “Regional Labor Markets and the Determinants of Wage
Inequality.” American Economic Review 84 (2): 17-22.

U.S. Bureau of the Census. 1984. Statistical Abstract of the United States: 1985.
Washington DC: GPO.

U.S. Immigration and Naturalization Service. 1991. Statistical Yearbook of the
Immigration and Naturalization Service: 1990. Washington DC: GPO.

. 1992. Statistical Yearbook of the Immigration and Naturalization Service: 1991.
Washington DC: GPO.

. 1997. Statistical Yearbook of the Immigration and Naturalization Service: 1996.
Washington DC: GPO.

41

Appendix A: Practical Estimates of IRCA’s Impact.

Table A1.1 gives estimated changes in labor market gaps due to IRCA in 2003
dollars, usual weekly hours, armual weeks worked and employment likelihoods and
probabilities of living in a poor family for three cities. Jackson, MS is representative of
low-PROP cities. Baltimore has a PROP close to the median PROP for a city. Atlanta is

near the 75th percentile city.

42

3-30 05 .3 555:8 88 8 985335 ...

 

 

.8285 m 88.8 228.8 60 s 82885 men: 2885 .é £855 88 5 a 55>: 8 $8325
:83 828.5 828.5 50 e 82885 mo: 28: 8 888885-82 8 $8885
8.83 NV. 3% «.5. 35¢ Nd- 500 2 o>:5_om 503 252 he coo—.83 38>? “5255‘
5:5: Wm- 8858 mm- 8558 Z- 500 9 2553M 503 252 525: >283 ~5st
3.3 8.3 88.3 60 3 2885 men 2588.5 8 8553 885
8.8- 8.3- 8.8 60 252 9 83885 won: 252 8 8583 3:8:
8888: .853 32 5.8 552

38:55- 8.2%- 33$- zoom 25: 283 9 8:585 mean 882 825 8 2:85 82:2
2382 853 33 5.8 82

$85.8 $28.8 . 888.5 88:5 88885 88853
<0 .5822 a: 8.88585 52 .8382 <52

 

6.05836 .55 52$: vogue—om ..8 359:— 635533— m.<0~= —.—< 03.;

43

Chapter 2

Criminal Justice Outcomes and Relative Labor Markets of Blacks

2.1 Introduction

Previous work has documented empirical differences in labor market outcomes
across race; blacks on average perform less favorably than whites.39 Also, arrest rates for
blacks are high relative to those for whites.40 Employers form beliefs about the reliability
of prospective hires based on past experience with employees, especially in regards to the

‘ or particular

stereotypical views they have of workers along race and gender lines.4
concern in this work is the perception of employers about the returns to transactions with
blacks relative to whites.

Lundberg and Startz (2002) model how uncertainty about the returns to
transactions can lead to and perpetuate discrimination against minorities. In particular,
minority status signals greater uncertainty about returns even if the average return is the
same for minorities and non-minorities. However, Lundberg and Startz note that if the

expected return from trade with a minority is lower, the disparity may not be

discriminatory if average wage equals average productivity across groups. Their

 

’9 See, for example, Moss and Tilly (1991), Neumark (1999), Mason (2000), Chandra (2000), and
Heckrnan, Lyons and Todd (2000) for discussions of the black-white wage gap changes over the later part
of the twentieth century.

‘0 This statement is based on examination of national data on incarceration by race in various years of the
Statistical Abstract of the United States over the period 1978 to 1992. The author’s own calculations of
total arrested persons divided by population according to race at the city and metropolitan area level using
data from the Uniform Crime Reporting Program and the National Incident Based Reporting System
support the conclusion that the number of black arrests relative to their population in an area is generally,
though not always, higher than that of whites.

‘1 The Multi-City Study of Urban Inequality (MCSUI) surveys employers as to their perceptions of
employees by race, ethnicity and gender. For a discussion of the findings see Moss and Tilly (2001).

44

statistical discrimination model postulates that black stereotypes may lead to lower
weighting of individual productivity indicators such as education or experience.
Statistical discrimination models generate bias against (favoritism towards) a particular
group(s) because of a feature of the group(s) other than race. This differs from taste
discrimination which implies a dislike for (nepotism towards) the group without apparent
reason.

The Multi-City Study of Urban Inequality (MCSUI) finds negative perceptions
among employers of black males hired for low-skilled jobs.” The concern of this paper
is that factors exogenous to individuals may affect the general perception of black males.
In particular, if employers have observed higher absenteeism among blacks in general,
employers may believe that black candidates are more likely to miss work compared to

whites.43

Corcoran and Duncan (1979) include absenteeism as one of a host of
productivity factors that may affect labor market gaps by race. They note that white
males have lower absenteeism rates than black males.

This study focuses on the effect of one factor exogenous to the individual that
may affect absenteeism by race — relative arrest rates by race. Absenteeism and arrests
are directly related. Hindelag et al. (1981) note that an arrest implies not just immediate

detention (perhaps overnight or more), but may also require future court appearances and

thus is expected to infringe on time on the job. They also report that arrestees may not

 

‘2 See again Moss and Tilly (2001).

‘3 The literature on absenteeism has been used as an explanation for gender wage gaps in labor market
outcomes (Gronau 1988; Ruhm 1998). These works hypothesize that children factor into the career path of
women not only in terms of planned pregnancies, which legally entitles them to paid-leaves of some
duration, but also in terms of more frequent spontaneous absences. Vistnes (1997) finds that the presence
of young children increases the probability of female missing work. She also finds that among females
missing work the presence of young children increased the number of absences, while among men who
missed work it increased the days lost from work. She notes the men were likely to be single or married
with a working wife.

45

truthfully explain arrest-related absences to their employers. Grogger (1995) suggests
that employers may fire workers who report their arrests due to the costs associated with
employee absenteeism.

As blacks have higher arrest rates than whites, ceteris paribus, they will have
higher unexplained absences. Moreover, comparison of survey data on a sample of
arrestees (Hindelag et al., 1981) with actual arrest histories suggests that, when asked,
blacks are less likely than whites to admit past criminal involvement.“ If blacks are
more likely to underreport arrests to survey workers, blacks may be even less likely to
report arrests to employers. Thus, in the absence of information on relative arrest
probabilities and arrests’ subsequent effect on work attendance, employers may be more
likely to associate black absenteeism with a poor work ethic rather than an exogenous
factor such as an arrest. This, in turn, may reduce future black hires.45 In short, all else
equal, employers may statistically discriminate against blacks to avoid employees who do
not show up for work.46

Higher relative arrest rates for blacks may reflect higher levels of participation in
crime by blacks. Relatively greater participation in crime by blacks may reduce
productivity on the job even absenteeism between blacks and whites were the same.
Perhaps, blacks who are working in crime may exert less effort while on the legal job;
motivation and tiredness are potential reasons for less effort. Previous literature has

documented from microdata that past arrests of an individual implies relatively lower

 

“ Supportive of this Grogger (1998) discusses evidence that actual arrest data imply higher relative arrest
rates for blacks when compared to whites but self-reports imply similar ones.

‘5 Moss and Tilly (2001) note that in a survey soliciting impressions of workers by race, “blacks are less
reliable” was a frequent response.

‘6 While it must be recognized that many other factors may affect the perceptions of employers (including
actual work ethic of each race), this work only asks whether the relative arrest rate is a contributing factor
to black-white labor market gaps. The arrest rate is the hypothesized exogenous channel through which
absenteeism can be affected, impacting the relative absenteeism rate and the wider black population.

46

future employment probabilities for the individual (Freeman 1992; Grogger 1992). It
should be noted that Grogger (1995) finds only modest and short-lived effects of arrests
on future outcomes after controlling for individual fixed effects.47 This suggests that
poor labor market outcomes of those with arrest histories are not due to actual arrests;
rather that criminals with arrest histories are low-productivity workers to begin with.

Employers may practice statistical discrimination because they believe that blacks
are more likely to have criminal backgrounds because of higher proportions of the black
populations that are involved in crime. Holzer, Raphael and Stoll (2002) suggest that
such stereotypical views may be responsible for relatively less hiring of blacks by small
firms compared to large firms.‘18 Further, support for employer dislike of criminals is
found in Pager (2003), which uses audits to test whether a criminal background reduces
employment prospects. She finds significant reductions in opportunities for both whites
and blacks due to admission of criminal backgrounds.

The present study hypothesizes that higher arrest rates will lead to more
absenteeism for blacks leading to lower productivity of blacks as a group, potentially
causing statistical discrimination against individual blacks. This work tests whether there
is a relationship between relatively worse labor market outcomes for blacks and their
higher arrest rates. Specifically, it examines whether labor market outcomes for blacks
are relatively worse in areas where blacks are more likely to be arrested. I include city-
specific fixed effects to partially control for the possibility that cities where blacks face

higher discrimination by the criminal justice system are also cities in which blacks face

 

‘7 Findings by Kaestner (1994) support this. He finds that illicit drug use negatively affects labor supply of
young adults when fixed effects are not controlled for. Once individual specific effects are included in the
model, the effect of illicit drugs goes away.

‘8 They find evidence consistent with statistical discrimination against blacks. However, they do not
control for employer tastes and so are unable to rule out taste discrimination.

47

greater taste-based employer discrimination and other effects which may cause relative

labor market outcomes and arrest rates to be spuriously related.
2.2 Theoretical Framework

Let a one-period economy consist of the government, NA type A workers, NB type
B workers, and N identical employers with two jobs each (job 1 and job 2).49 Assume
N< NB < NA so that there will be some unemployment and type B is a minority group.
Let n'; represent the number of type t persons that are hired in job j.

Both types consist of criminals and non-criminals. Assume that type
B has a higher proportion of criminals than type A. Assume that the government arrests
types with a probability that is a monotonically increasing function of the rate at which
types commit crimes. So type B non-criminals have a higher probability of arrest than a
type, A non-criminals due to statistical discrimination by the police, but type B non-
criminals have a lower probability of arrest than a type B criminal.50 Let it denote the
average probability that the government arrests a type t worker. Thus, it: (it‘itC + (1-(1‘)itN
(where at denotes the proportion of type t persons who commit crimes, itC denotes the
probability that a criminal of type t is arrested and im denotes the probability that a non-
‘ criminal of type t is arrested).

Let p‘ denote the average probability that a type t worker will be absent from the
job in a given period. Let a denote the exogenous probability of absenteeism by all
workers regardless of type. Given the above assumptions, pt = a + it. Since a greater

number of type B workers commit crimes, there will be relatively more arrests of type B

 

‘9 Type and race are used interchangeably.

5° Knowles, Persico and Todd (1999) examine whether police discriminated against blacks in motor vehicle
searches for drugs. They test whether this is due to racial prejudice or statistical discrimination and cannot
reject the statistical discrimination model.

48

workers.51 Thus, pB >pA. Let MRP‘J- denote the marginal revenue product generated by a
type t worker in job j, W? denote the wage paid to type t in job j and Cj denote the extra
cost an employer incurs when a worker does not show up for the job j. If workers do not
show up for the job, assuming an optimal initial allocation of inputs by the firm, a more
costly production method is utilized so that output remains unchanged.52 Thus C2 0 is
assumed. There are no other costs of production.53 Employers are unable to observe
whether a given individual is a criminal or not and employees who are arrested do not
inform employers of the arrest. Given the evidence that a significant number of
employers would rather not hire persons with a criminal background (Holzer, Raphael
and Stoll 2002), this is a reasonable assumption. Employers will, however, form
estimates over time of the average rates of absenteeism for each race. Based on the
expectations of absenteeism by types, the optimal allocation of workers within jobs,
therefore, must satisfy the following:

MRPB,- = MRP", (2.1)

“’81" PB C1 WAP‘ PA Ci
Based on the higher average probability of arrest in the black population, and the
necessary ensuing absenteeism, employers overestimate the probability of absenteeism
for non-criminal blacks and of course for blacks who are never arrested. They therefore
incorrectly underestimate the return to hire or job placement of non-criminal and never

arrested blacks. Equation (2.1) implies:

 

5' The model ignores other links between criminality and productivity besides absenteeism.

52 For example, if a fast food restaurant worker does not show up for their particular job at lunch time the
manager may be forced to help out or staff the position in such a way that other workers will work
overtime. Assuming diminishing marginal productivity either a lower quality service or quantity work will
be produced.

53 This is a simplification under the assumption that other costs are not impacted by the race assignments of
employees to these jobs.

49

(i) Since pB >pA , on average, if workers are paid the same wage regardless of type
for job j, then an employer will always prefer to hire a type A worker provided
workers are equally productive when they are on the job;

(ii) assuming equal productivity on the job and equal wages across types for the same
job, if the government requires hire of both types, employers will always prefer to
allocate type B employees to jobs with lower Cg“

(iii) if we assume C2>C1 and that the government requires hire of both types in both
jobs, then it"; > n82 or (nAz /NA)/ (n92 N”) > 1 (where n '1 is the number of type t
workers in job j).55 Thus, a larger share of population of type A relative to type B get
hired for the job where absenteeism is more costly to employers;

(iv) if types are equally productive on the job and employers can pay different wages
within the same job category then ij < wAj.

While there are productivity differences on average by criminal and non-criminal
type, type B non-criminals will pay an unfair share of the costs of absenteeism in their
type. They are paid lower than their productivity warrants and will be disproportionately
denied better (higher Cj) jobs. Non-criminals of type B who ultimately never get arrested
pay the greatest share of the cost on average, even though they are more productive than
type A criminals and type A non-criminals who get arrested.

This project looks at the effect of relative arrest rates on relative labor market
outcomes for blacks compared to whites in cities across the U.S. In the data to be used

here, I cannot test whether a black who never gets arrested is paid less than an observably

 

5‘ See Bulows and Summers (1986) for another explanation of placement of blacks into “bad” jobs in dual

labor markets.
55 See Holzer and Neumark (2000) for a survey of the literature on Affirmative Action, a policy designed to

increase the employment opportunities of minorities and women.

50

similar white who never gets arrested. I test the more limited hypothesis — which the
above model implies should be observable in the data — that groups of workers with
higher arrest rates will be, on average, less productive and will thus have worse labor

market outcomes on average.
2.3 Data

Data on the number of blacks and whites arrested in a city are obtained from the
Uniform Crime Reporting Program (UCR): Arrests by Age, Sex, and Race for Police
Agencies in Metropolitan Statistical Areas for 1979, 1989 and 1999.56 The data contains
arrests by race for each reporting city. Police agencies report their total arrests by age,
sex, race and ethnicity on a voluntary basis. The sample is restricted to police agencies
that report arrests by race. The geographic unit used for this analysis is the city, so only
reporting agencies located in the core city of a metropolitan statistical area are included.

The paper uses the relative arrest rates of blacks to whites in a city to proxy for
relative absenteeism caused by arrests.57 Arrest rates for blacks are calculated by
dividing the number of black arrests by the black population in each city.58 The same
variable is calculated for whites. The relative arrest rate is the ratio of per capita black
arrests to per capita white arrests.

Labor market data on annual wage and salary earnings, usual weekly hours,

employment and weeks worked are obtained from the 1980, 1990 and 2000 Public Use

 

5° If available, for cities not reporting in the stated years, but which reported in the previous year, I calculate
the relative arrest rate for the previous year and substitute it for the current year’s value. As will be
discussed later on in the paper, relative arrest rates are highly serially correlated.

57 Again, though this is not the only factor affecting absenteeism, (Markarn and McKee 1991; Wilson and
Peel 1991) nor is absenteeism the only reason for correlations between arrests and productivity, it is the
only channel explored in this paper.

58 Population estimates by race for cities are obtained from the 1980 Census of the Population:
Characteristics of the Population Series, Decennial Lookup (1990 estimates) tool and American Factfinder
(2000) tool on the U.S. Census Bureau’s web page (www.census.gov). These figures are used to calculate
population grth rates on which I base estimates of populations in the year prior to each Census.

51

Microdata Sample (PUMS) 5% State Samples.59 This is a survey of 5% of the U.S.
Population taken at the time of the decennial censuses. The PUMS provide information
on the earnings and labor force status in the year prior to the Census along with current
demographic information and city of residence. This information is used to construct
average labor market outcomes by race in cities that are identified in the UCR data.

I restrict the sample to males reporting either black or white race, no college
degree and aged 16 to 45 years inclusive.60 The sample also excludes those in the
military, institutional group quarters and those working for government agencies. City
level averages of black and white characteristics —- age, education, marital status
indicators, income, employment likelihoods, eamings— are obtained by using weighted
averages of variables across samples of blacks and whites living in identifiable cities.(’l'62
Only cities with at least 30 persons of each race in the PUMS data in each year are
included.63 Thus, each city contains a sample of at least 30 white males and 30 black
males. These men all report having fewer than four years of college and have jobs
outside the military or government. Note: the UCR data contains actual arrests in a city.
Though figures are likely to be underreported due to omission of certain types of crimes

or less than 12 months of reporting, it is not expected to bias the relative arrests. Arrest

figures are used even if there are very few arrests reported of men of either race as the

 

59 These data sets are freely available at www.ipums.org.

‘0 Based on author calculations in the UCR data for the years and cities used here the 16-44 year old share
of the arrestees range from a minimum of .53 to a maximum of .93 with the average being 0.83. With the
exception of 2 city-year observations the share exceeds 70%. A significant part of the denominator for this
calculation includes under-l6 juvenile arrestees who are considered outside of the labor market not people
over 45 years old.

6' Averages are weighted by the population fiequency weights in the census data.

‘2 Grouped years of education recodes are used so this is not interpretable in the usual terms of years of
education. However, higher values indicate more years of education.

63 See Appendix B Table A2.2 for a list of cities used and their relative arrest rates by year.

52

data are not a random sample of arrestees from which we are calculating proportions. It
is important to recall that the figures are total reported arrestees by race.

In order to investigate whether employers assign jobs to new hires differently
depending on a perception of absenteeism, some proxy for the cost of absenteeism in
various occupations is needed. I use an updated version of the Duncan Socio-Economic
Index for 1980 and 1990 census occupation codes for this purpose.64 The ranking of
costs is determined by matching occupation codes to the index of socio-economic
occupation status for 13 occupation categories.‘55 While occupation prestige rank is not a
perfect measure of the cost of absenteeism on the job, it does provide some measure of
the relative value society places on particular occupations. While jobs with higher annual
incomes are usually those with greater prestige (and here a proxy for higher C), it is
important to note they do not correlate perfectly.66 Further, Stoll, Raphael and Holzer
(2001) and Giuliano (2003) find that race of the employer/hiring manager is an important
determinant of the employment likelihoods of blacks. This adds further motivation for
examining occupational placement outcomes in the context of the impact of
discrimination. In formulating policies to address gaps in labor market outcomes by race,
investigating factors which determine who obtains positions with hiring authority may be
useful. This is the case if indeed having more minority managers/employers increases the

probability of employment of minority workers.

 

6‘ See Hauser and Warren (1996).

65 See Appendix A Table A2.1 for index and occupation categories. While a detailed occupational index is
available it is not used for two reasons. First, the 2000 PUMS codes are more detailed than those of 1990
and 1980 and would require omission of many occupations that could not be identically matched with their
earlier code. Second, up to a point, coarser index information sets may be a better proxy for the
absenteeism costs than the detailed occupation figures.

‘6 Saxenian (2001) states there is little evidence of earnings discrimination against Chinese and Indians in
managerial and professional occupations. It also discusses the initial concentration of these groups into
professional rather than managerial positions based on investor preferences for non-Asian managers.

53

An additional control variable employed is the relative institutionalization rate,
which is calculated from the 1980, 1990 and 2000 PUMS.67 I limit the sample used for
this calculation to 16-45 year old males with less than a college degree to match the
general arrestee population.68 The institutionalization rate for each race is the weighted
population in institutional group quarters divided by the total estimated state population
of that race. While the intent is to control for relative depletion of the low-skilled or low-
productivity employees from a city, it is not possible to ascertain the origin city of

69 Thus, the state level variable is a second best

inmates/ institutionalized persons.
measure of the depletion of the low-skilled population. Omission of this variable may
bias upwards the estimated effect of relative arrests since relative incarceration rates are
positively correlated with relative arrests. For example, suppose the distribution of
abilities is equal across races and cities. If high school dropouts have low levels of
employment and black high school dropouts are incarcerated at higher rates than white
dropouts in a particular city, ceteris paribus, the city will have relatively higher

employment rates for blacks than whites. This might mistakenly be interpreted as a city

with a higher distribution of productive blacks relative to whites.

 

‘7 1990 and 2000 PUMS files do not report group quarters, the variable indicating residence in a
correctional institution, with the detail available in 1980. Thus, persons in mental institutions, old age
homes, homes for the handicapped, hospitals, homes for the mentally handicapped or any other non-
institutional group housing are also included in this variable. Arguably, it still holds value in capturing the
depletion by race of low productivity persons from the labor force. In the 1980 PUMS, for men aged 16-45
years, 65% of those institutionalized were in correctional institutions, 13% were in mental institutions, 5%
were in homes for the elderly and 18% were in other institutions (such as mentally handicapped homes).
These totals may include institutionalized persons as well as resident employees.

68 I assume that those with 4 or more years of college are college graduates.

‘9 Butler and Heckman (1977) discuss the importance in accounting for the withdrawal of low-skilled
workers from the labor force in the context of evaluating Equal Employment Opportunity Act’s (EEO)
effectiveness. They discuss the possible overestimation of the EEO’s effect in narrowing black white labor
market gaps. They argue that concurrent increases in welfare generosity induced selective labor force
withdrawal from low productivity blacks.

54

The institutionalization rate also captures the actual ratio of criminal participation
by race.70 In addition, the period of absence from the job for an arrestee should be short-
lived. If, however, arrests lead to incarceration, the cost to an employer is higher than the
cost of an arrest because recruiting a new employee is necessary in addition to the usual
cost of absenteeism. Then employers would consider both the relative incarceration rates
and the relative arrest rates before hiring.

Data on institutionalized persons aged 16 to 45 years indicates that the population
consists mainly of persons with 12 years of education or less." Unlike the available data
on the prison population, the arrests data do not contain education, so I am unable to
estimate arrest rates of those without a college degree. However, I assume that the
probability of arrest for college graduates is low enough so as not to be an important

2 Another undesirable inclusion in the arrest

consideration to prospective employers.7
data that must be noted is the female arrestees. The proportion of women arrested varies
from city to city. The UCR data from the cities and years used here shows that between
10% and 34% of total reported arrestees in a city are female.73 Unfortunately, I cannot
accurately remove females fi'om the total black arrests or the total white arrests.
However, using data from the National Incident Based Reporting System (NIBRS)

available for a few cities in this data, I compare the difference between the ideal variable,

relative arrest rates for 16-45 year old black males relative to white males, and the one

 

7° Although, if blacks are sentenced at higher rates than whites (and for longer periods) this may
overestimate the criminal participation rates of blacks relative to whites. There should nevertheless be a
significant positive correlation between true localized criminality and incarceration rates.
7' Based on author calculations with the 1980, 1990 and 2000 PUMS, approximately 3% of black and white
males aged 16 to 45 years correctional institutions had 4 or more years of college. Approximately 43% of
those in correctional attended 12"1 grade and/or had some incomplete college education.
72 This is based on the work done for the previous footnote. Arrests lead to incarcerations. Given the low
percentage of college graduates incarcerated, it seems fair to assume the number of arrestees is also minor.

3 Figures calculated from UCR data from 1978, 1979, 1988, 1989, 1998, and 1999. This also shows that
average female share of the arrestee population in the sample of cities and years used is roughly 19%.

55

which I use in this paper relative arrest rates for blacks versus whites. Appendix C
contains a discussion of the findings of this comparison.

A summary of relative arrest rates and labor market outcomes by year are given in
Table 2.1. It is worth repeating that the sample used to calculate labor market outcomes
by city is restricted to those without a 4-year college degree. Thus, in the average city
used .here, the relative education level of blacks without 4 or more years of college
education is higher than that of whites without 4 years or more of college. This is
possibly caused by the inclusion of Hispanics in the white average as they constitute a

disproportionately large portion of the high school dropout population.
2.4 Empirical Methodology

Using a linear regression model and city-year level observations, I estimate the
effect of these relatively higher arrest rates for blacks on their relative labor market
outcomes. As average relative characteristics of the black and white populations are also
expected to affect relative labor market outcomes, these are used as additional controls in
the linear regression model.

The following regression model is estimated:

RelOutcomej, = [30 + [31 RelArrestl J, + [32 RelInstitutionJ-t + B3 RelEdJ-t + [34
RelAgeJ-t + [35 RelAgesqjt + B6 RelMarJ-t + [37 RelSingjt + [33 Y90t+ B9 YOOt+ Cin + an (2.2)

The prefix Rel indicates averages for blacks in city j in year t divided by averages
for whites in city j in year t for the following variables: Outcome - annual wage and
salary income, employment rate, hourly wage or occupational rank; Arrest] —- city arrest
rate for blacks relative to whites in the calendar year prior to the Census; Institution —

state institutionalized rate in census year; Ed — average Education Level; Age — average

56

age; Agesq — average of age-squared; Mar - proportion married; Sing — proportion
single; Y90 - census year is 1990; YOO — census year is 2000; city — city fixed effect; and
e - randomly distributed error term.

Equation (2.2) is estimated by pooled ordinary least squares (OLS) both without
and with city-specific effects. The estimated effects of relative arrest rates have both
positive and negative biases and so the direction of the bias cannot be ascertained. The
city fixed effects control for time invariant factors that are correlated with relative arrest
rates and relative outcomes and should account for much of this bias. For example, the
level of taste-based discrimination in a city might be common to both police and
employers. Thus, a city with relatively higher arrest rates because of taste-based
discrimination within the criminal justice system may also be a place in which employers
are less willing to hire blacks. As employers would learn about relative absenteeism of
workers over time, past arrest rates may also be a factor. However, the correlation
between relative arrest rates in the two years prior to each census year exceeds .90. Thus,
the relative arrest rate in the year prior to the census is expected to capture information
contained in the most recent arrest rates. Employers may be unaware of the cause of
absenteeism differences and attribute it differences in work ethic.

2.4.1 Related Issues

Calculations of relative arrest rates include cities that report less than 12 months
out of the year. Since both black and white arrest reports will be affected, unless there is
seasonality in arrests that differ by race, I do not expect this to bias the estimate of the
relative arrest rate by race. In the 2000 PUMS, it is apparent that Hispanics are reporting

a race other than black or white in greater proportions than in either the 1980 or 1990

57

Censuses. This leads to an undercount of the white population in comparison to the
measure used in prior years and an inflation of their relative arrest rates. This is
problematic as arrest data includes a large number of Hispanics who have been reported
as white.74 Thus Hispanic-whites are overwhelmingly included in the arrest figure, but
not as numerously in the 2000 Census population counts. An examination of the relative
arrest rates by year reveals that with the overestimate of the white arrest rate in 2000,
average relative arrest rates of blacks compared to whites decreased slightly. This trend
holds if the arrest rates are calculated using black and non-black to make the relative
calculations. So the movement is not an artifact of the change in race-reporting by
Hispanics. Another complication is that Hispanics have relatively higher arrest rates
than non-Hispanic whites, but also have relatively lower labor market achievement.
Thus, exclusion of Hispanics would provide a better test of this paper’s hypothesis than
the current data and is an important avenue for future work.75

Irnportantly, though the model predicts a ceteris paribus negative relationship, it is
not possible in a real world setting to obtain data where ability, perceptions and all other
factors are the same for blacks as well as whites. In particular, the unobserved factor of
future arrest is not observable in these data. As a second best, the estimation procedure
controls for a variety of factors including city-specific fixed effects and then asks the
question of whether or not the arrest rates matter to city-level averages of the relative

labor market outcomes.

 

7‘ The National Incident Based Reporting System (NIBRS) contains person level data on arrestees,
including race and Hispanic ethnicity. In the 1999 data, less than 2% of those reported as having Hispanic
ethnicity also had race reported as black.

7’ NIBRS provides person level data on arrestees, but currently many of the largest U.S. cities do not report
using this system. This data set may be useful for future work, to provide a cleaner test, when more cities
are reporting.

58

The city fixed effects control for any features constant in all three census periods
but correlated with arrest rates. However, a bias remains if changes in both relative arrest
rates and relative outcomes are due to changes in underlying prejudices. Additionally, in
cities where blacks face more discrimination when seeking legal employment, they may
be more likely to turn to crime as a means of economic support over time.76 In other
words, relative arrest rates may be endogenous. Ideally, an instrumental variables
procedure would be implemented, but this study is a first step to assess the importance of
impact of arrest rates on general labor market outcomes of blacks relative to whites.

It is important to compare the results without the city-fixed effects to those which
control for the fact that blacks may be worse (better) off in general in particular cities. If
we find that relative arrest rates are related to lower labor market outcomes for blacks
when compared to whites, it would be interesting to see if city-specific effects can
account for all of such effects. It is also possible that, even without controls for city fixed
effects, other factors not accounted for in this analysis are more important than relative
arrest rates for determining relative outcomes. So examination of the basic relationship

without controlling for city fixed effects is an important piece of the analysis.
2.5 Results

Table 2.2 gives the results of the OLS estimates of equation (2.2) without controls
for city specific fixed effects. Relative arrest rates only affect the relative hourly wage
for blacks compared to white. The effect of relative arrest rates on annual income and the

occupation index is negative, but insignificant.

 

76 See Ehrlich (1974) for early work on this avenue of thought.

59

Table 2.2 also shows that relative institutionalization rates are positively related to
city averages of relative labor market outcomes. This supports the hypothesis that low
productivity blacks are being removed from the relevant civilian labor force leaving
behind a relatively more productive group of blacks. Another more causal interpretation
is that institutionalizing relatively more blacks will improve the labor market for the
remaining blacks if labor markets are segmented along racial lines. This would occur as
the labor supply curve for black jobs shifts to the left to a greater degree than that for
whites due to institutionalization.

Table 2.3 gives the results controlling for city-specific fixed effects. The effect of
higher relative arrest rates on the relative labor market outcomes of blacks is no longer
significant. This implies that cities in which blacks are more likely to be arrested are also
those in which blacks have lower hourly wages, independent of the arrest rates.
However, the occupation index is significantly negatively correlated with arrest rates
even after controlling for city-specific effects.77 This finding is significant as the impact
of occupational placement is not limited to individuals who do not move up the
occupation ladder. The findings of Stoll, Raphael and Holzer (2001) and Giuliano (2003)
suggest that assignment of blacks into lower occupation groups may echo in the form of
lowering hiring rates of black males. Secondary effects may operate with a lag. As
mentioned before, due to the high correlation between the relative arrest rates over time,
inclusion of current and lagged arrest rates to capture these effects is beyond the scope of

this paper.

 

77 While the negative effect is robust, the significance is affected by the measure of absenteeism cost used.
In particular, use of a 6-category index rather than the 13-category index renders the effect insignificant.

60

2.6 Conclusion

This paper investigates the possibility that absenteeism as caused by relatively
higher arrest rates impacts relative labor market outcomes of blacks. Effect of relative
arrests on earnings, employment and occupational rank are examined. Hourly wages are
found to be negatively impacted by higher relative arrest rates if one does not control for
city-specific effects. However, the type of occupation assigned to blacks is significantly
negatively related to arrest rates even after city fixed effects are controlled for. While
this paper is unable to explore the repercussions of relatively higher arrest rates as it
operates against the placement of black males in higher status occupations, it does
suggest that further research needs to be done to explore this particular cause of black-

white achievement gaps.

61

Table 2.1 Mean of Relative Independent and Dependent variables.

 

 

Year 1979 1989 1999
Relative Age 1.01 1.00 0.99
Relative Proportion Married 0.81 0.68 0.77
Relative Proportion Single 1.19 1.32 1.22
Relative Education Level 0.99 1.02 1.02
Relative Wage and Salary Income (Annual) 0.75 0.69 0.71
Relative Hourly Wage 1.02 0.95 0.96
Relative Employment Rate 0.87 0.85 0.86
Relative Socio-Economic Index (13 categories) 0.90 0.92 0.93
Relative Arrest Rates (Black over White) 2.31 3.33 2.93
Relative Arrest Rates (Black over Non-Black) 2.49 3.73 3.62
Relative Incarceration Rate (State Level) 4.20 5.55 5.98
N 77 67 74

 

62

Table 2.2 OLS Estimates of the Effect of Relative Arrest Rates on Relative
Labor Market Outcomes of Blacks Compared With Whites

 

Probability
of
Employment
Annual Wage in the
and Salary Previous Hourly Occupation
Income Year Wage Index
Relative Arrest Rate -0.005 0.005 -0.02 -0.003
(0.006) (0.004) (0.012)* (0.002)
Relative Institutionalized
Rate 0.013 0.001 0.02 0.01 1
(0.005)" (0.003) (0.011)* (0.002)***
Relative Education 0.779 0.223 0.343 0.51 1
(0.096)*** (0.059)*** (0.194)* (0.041)***
Relative Age 3.525 3.309 3.669 0.47
(2.190) (1.340)" (4.440) (0.932)
Relative Age Squared -1.334 -1.594 -1.932 -0.175
(1.127) (0.690)" (2.286) (0.480)
Relative Proportion
Married 0.171 0.207 0.177 -0.069
(0.065)*** (0.040)*** (0.133) (0.028)“
Relative Proportion
Single -0.073 0.039 0.069 -0.057
(0.047) (0.029) (0.096) (0.020)***
1990 Census Indicator -0.046 -0.005 -0.073 -0.012
(0.019)" (0.012) (0.038)* (0.008)
2000 Census Indicator -0.053 -0.014 -0.087 O
(0.019)*""" (0.011) (0.038)" (0.008)
Constant -2.317 -1.289 -1.319 0.185
(1.068)M (0.654)“ (2.166) (0.454)
Observations 218 218 21 8 218
R-squared 0.5 0.28 0.09 0.58

 

Standard errors in parentheses
* Significant at 10%; ** at 5%; *** at 1%

63

Table 2.3 Fixed Effect Estimates of the Effect of Relative Arrest Rates on

Relative Labor Market Outcomes of Blacks Compared With Whites

Relative Arrest Rate

Relative Institutionalized

Rate

Relative Education
Relative Age
Relative Age Squared

Relative Proportion
Married

Relative Proportion
Single

1990 Census Indicator
2000 Census Indicator
Constant
ObserVations

R-squared
Number of Cities

 

Probability
of
Employment
Annual Wage in the
and Salary Previous Hourly Occupation
Income Year Wage Index
-0.007 -0.002 0.001 -0.007
(0.010) (0.006) (0.026) (0.004)*
0.002 -0.003 0.014 0.002
(0.009) (0.005) (0.023) (0.004)
1.066 0.313 1.007 0.356
(0.246)**"' (0.143)" (0.633) (0.100)*'"*
1.406 0.985 10.862 -0.257
(2.718) (1.580) (6.994) (1.101)
-0.437 -0.49 -5.81 1 0.307
(1.389) (0.807) (3.573) (0.562)
0.001 0.07 0.48 -0.005
(0.098) (0.057) (0.251)* (0.040)
-0. 161 -0.021 0.426 -0.012
(0.082)* (0.048) (0.21 1)" (0.033)
-0.067 -0.009 -0.1 15 0.01
(0.023)*** (0.013) (0.059)* (0.009)
-0.067 -0.02 -0.09 0.03
(0.024)*"”" (0.014) (0.062) (0.010)***
-1.079 0.059 -5.971 0.523
(1.359) (0.790) (3.497)* (0.551)
218 218 218 218
0.41 0.19 0.13 0.35
95 95 95 95

 

Standard errors are in parentheses
* Significant at 10%; ** 5%; *"'* 1%

64

References

Bulow, Jeremy I. and Lawrence Summers. 1986. “A Theory of Dual Labor Markets
with Application to Industrial Policy, Discrimination and Keynesian Unemployment.”
Journal of Labor Economics 4 (3) Part 1: 376-414.

Butler, Richard and Heckrnan, James J. 1977. “The Government’s Impact on the Labor
Market Status of Black Americans: A Critical Review.” In Leonard J. Hausman, Orley
Ashenfelter, Bayard Rustin, Richard F. Schubert and Donald Slaiman (eds)

Equal Rights and Industrial Relations. Madison, WI: Industrial Relations Research
Association.

Chandra, Amitabh. 2000. “Labor-Market Dropouts and the Racial Wage Gap: 1940-
1990.” American Economic Review 90 (2): 333-338.

Chilton, Roland, and Dee Weber. UNIFORM CRIME REPORTING PROGRAM
[UNITED STATES]: ARRESTS BY AGE, SEX, AND RACE FOR POLICE
AGENCIES IN METROPOLITAN STATISTICAL AREAS, 1960-1997 [Computer file].
2nd ICPSR version. Amherst, MA: University of Massachusetts [producer], 2000. Ann
Arbor, MI: Inter-university Consortium for Political and Social Research [distributor],
2000.

Corcoran, Mary and Greg Duncan. 1979. “Work History, Labor Force Attachment, and
Earnings Differences between the Races and the Sexes.” Journal of Human Resources 14
(1): 3-20.

Ehrlich, Isaac. 1974. “Participation in Illegitimate Activities: An Economic Analysis.”
In Gary S. Becker and William M. Landes (eds.) Essays in the Economics of Crime and

Punishment. New York, London: Columbia University Press.

Freeman, Richard. 1992. “Crime and the Employment of Disadvantaged Youth.” In
Peterson, George and Vroman (eds), Urban Labor Markets and Job Opportunity.
Washington DC: Urban Institute Press.

Giuliano, Laura. 2003. “Race, Gender and Hiring Patterns: Evidence from a Large
Service Sector Employer.” Institute of Industrial Relations, University of Califomia,
Berkeley. Unpublished Paper.

Gronau, Reuben. 1988. “Sex-Related Wage Differentials and Women's Interrupted
Labor Careers-the Chicken or the Egg.” Journal of Labor Economics 6 (3): 277-301.

Hauser, Robert M. and John Robert Warren. 1996. Socioeconomic Indexes for
Occupations: A Review, Update and Critique. CDE Working Paper No.96-01.

65

Heckrnan, James J ., Lyons, Thomas M., and Petra E. Todd. 2000. “Understanding
Black-White Wage Differentials, 1960-1990.” American Economic Review 90 (2): 344-
349.

Hindelag, Michael, Hirschi, Travis, and Joseph Weis. 1981. Measuring Delinquency.
Beverly Hills: Sage Publications.

Holzer, Harry, Raphael, Steven and Michael Stoll. 2002. “Perceived Criminality,
Criminal Background Checks and the Racial Hiring Practices of Employers.” JCPR
Working Paper 295.

Kaestner, Robert. 1994. “The Effect of Illicit Drug Use on the Labor Supply of Young
Adults.” The Journal of Human Resources 29 (1): 126-155.

Knowles, John, Persico, Nicole and Petra Todd. 1999. “Racial Bias in Motor Vehicle
Searches: Theory and Evidence.” NBER Working Paper # w7449.

Lundberg, Shelley and Richard Startz. 2002. “Race, Information and Segregation.”
University of Washington. Unpublished paper.

Markham, Steven E., and Gail H. McKee. 1991. “Declining Organizational Size and
Increasing Unemployment Rates: Predicting Employee Absenteeism from within- and

Between-Plant Perspectives” (in Research Notes) The Academy of Management Journal
34 (4): 952-965.

Mason, Patrick L. 2000. “Persistent Discrimination: Racial Disparity in the United
States, 1967-1988.” American Economic Review 90 (2): 312-316.

Moss, Philip and Chris Tilly. 2001. Stories Employers Tell: Race, Skill and Hiring in
America. New York: Russell Sage Foundation.

. 1991. “Why Black Men Are Doing Worse in the Labor Market: A Review of
Supply-Side and Demand-Side Explanations.” New York: Social Science Research
Council.

Neal, Derek A., and William R. Johnson. 1996. “The Role of Premarket Factors in
Black-White Wage Differences.” Journal of Political Economy 104 (5): 869-895.

Neumark, David. 1999. "Wage differentials by race and sex: The roles of taste
discrimination and labor market information." Industrial Relations 38 (3): 414-445.

Pager, Devah. 2003. “The Mark of a Criminal Record.” American Journal of Sociology
108 (5): 937-975.

66

Ruggles, Steven and Matthew Sobek et a1. Integrated Public Use Microdata Series:
Version 3.0 Minneapolis: Historical Census Projects, University of Minnesota, 2003
accessed at www.ipums.org on April 27, 2004.

Ruhm, Christopher J. 1998. “The Economic Consequences of Parental Leave Mandates:
Lessons from Europe.” The Quarterly Journal of Economics 113 (1): 285-317.

Saxenian, AnnaLee. 2001. “Silicon Valley’s New Immigrant Entrepreneurs.” In
Cornelius, Wayne A., Espenshade, Thomas J ., and Idean Salehyan (eds), The
International Migration of the Highly Skilled.

Stoll, Michael, Raphael, Stephen and Harry Holzer. 2001. “Why Are Black Employers
more likely to hire Afiican Americans than White Employers.” JCPR Working Paper
228.

U.S. Dept. of Commerce, Bureau of the Census. I 980 Census of Population. Volume 1,
Characteristics of the population. Chapter C, General social and economic
characteristics. U.S. G.P.O. Washington, DC: Table 1.15. General Characteristics of
Areas and Places. Table 126. Age, Fertility, Relationship, and Educational Characteristics
of White Persons for Areas and Places. Table 132 Age, Fertility, Relationship, and
Educational Characteristics of Black Persons for Areas and Places.

U.S. Dept. of Justice, Federal Bureau of Investigation. UNIFORM CRIME REPORTING
PROGRAM DATA [UNITED STATES]: ARRESTS BY AGE, SEX, AND RACE, 1999
[Computer file]. Compiled by the U.S. Dept. of Justice, Federal Bureau of Investigation.
ICPSR ed. Ann Arbor, MI: Inter-university Consortium for Political and Social Research
[producer and distributor], 2002

Wilson, Nicholas and Michael J. Peel. 1991. “The Impact on Absenteeism and Quits of
Profit-Sharing and Other Forms of Employee Participation.” Industrial and Labor
Relations Review 44 (3): 454-468.

Vistnes, Jessica Primoff. 1997. “Gender Differences in Days Lost from Work Due to
Illness.” Industrial and Labor Relations Review 50 (2): 304-323.

67

Appendix A: Socio-Economic Index for 13 Occupation Categories

 

 

Table A2.1

Socio-Economic Index for 13 Occupation Categories

Professional Specialty 57.84
Executive Administrative and Managerial Occupations 50.64
Technicians and Related Support Occupations 44.35
Sales Occupations 35.96
Protective Service Occupations 35.85
Administrative Support Occupations, including Clerks 34.22
Precision Production, Craft and Repair Occupation 30.91
Transportation and Machine Moving Occupations 25.49
Machine Operators Assemblers and Inspectors 24.03
Service Occupations, Except Protective Household 20.43
Handlers, Equipment Cleaners, Helpers and Laborers 19.84
Farming, Forestry and Fishing Occupations 19.52
Private Household Occupations 17.69

 

Appendix B: Relative Arrest Rates by Year and City

 

Table A2.2

Relative Arrest Rates of Blacks to Whites in Each Year
City 1979 1989 1999
Albany NY 2.66 4.84
Allentown PA 3.26 4.65 3.32
Anaheim CA 2.40 1.95 2.25
Ann Arbor MI 4.05 4.84 4.91
Atlanta GA 1.16

Austin TX 2.04
Bakersfield CA 1 .41 2.94 2.52
Baltimore MD 1 .85 2.32 2.64
Baton Rouge LA 3.10 2.60 3.06
Beaumont TX 1 .36 2.45 2.09
Birmingham AL 0.87

Boise ID 3.55
Boston MA 2.54 4.27 2.86
Bridgeport CT 2.02 1.89 1.20
Brownsville TX 0.48
Buffalo NY 3.26 3.64 3.07
Charlotte NC 2.70 4.02
Chicago IL 3.54 3.11 2.85
Cleveland OH 1 .59 3 .03
Colorado Springs CO 3.19
Columbia SC 2.03 2.98

68

Columbus
Corpus Christi
Dayton
Denver

Des Moines
Detroit

Erie

Eugene
Evansville

F ayetteville
Flint

Fort Wayne
Fort Worth
Fresno
Garden Grove
Gary

Grand Rapids
Greensboro
Hampton
Jackson
Houston
Huntsville
Jackson
Jersey City
Kansas City
Knoxville
Lansing

Las Vegas
Little Rock
Long Beach
Los Angeles
Louisville
McAllen
Minneapolis
Mobile
Modesto
New Haven
New Orleans
New York
Newark
Newport News
Norfolk
Oakland

GA
TX
OH
CO

MI
PA
OR

NC
MI

TX
CA
CA

MI
NC
VA
MS
TX

MS
NJ
MO

99%582

KY
TX

CA
CT
LA

NJ

VA
VA
CA

1.57
1.66
2.34
1.98
3.60
2.71

2.69
2.05
1.26
1.33
3.97
1.48
2.45
2.14
1.62

1.28

2.25
2.28
0.96
2.60
2.18
2.07
2.39
2.23
1.44

5.60
1.51
2.60
2.40
1.18
2.64
1.60

1.40
2.13

69

2.91
4.78
1.62
4.54
3.96

2.96
4.19
2.34
1.90
1.99
1.71
5.58
3.45
2.18
2.31

3.46

1.85
4.99
2.03

2.67
2.02
1.96

6.92
2.70
4.31
2.31
2.27
2.36
2.11
3.73
3.16

2.57
1.36
2.21

4.19
1.19
4.24

4.33
3.08
2.21
4.70

1.74
1.22
0.95
4.24
3.45
2.34

3.02
2.00
2.54

2.45
3.11

3.29
2.16
1.88

0.56
7.89
2.99
2.20
1.84
1.40
1.81
1.66
3.18
3.01

Oklahoma City
Ontario
Oxnard
Paterson
Philadelphia
Phoenix
Pittsburgh
Portland
Portsmouth
Providence
Provo
Raleigh

Reno
Richmond
Riverside
Rochester
Sacramento
Saint Louis
Saint Paul
Salem

Salinas

Salt Lake City
San Antonio
San Bemadino
San Diego
San Francisco
San Jose
Santa Ana
Santa Rosa
Shreveport
South Bend
Spokane
Springfield
Stamford
Stockton
Syracuse
Tacoma
Toledo
Tucson

Tulsa
Virginia Beach
Waterbury
Worcester

OK
CA
CA
NJ

PA

PA
OR

38515;

VA
CA

CA.

MO

OR
CA

TX
CA
CA
CA
CA
CA
CA
LA

WA

,CT

CA

WA
OH

OK
VA
CT

1.91

0.82
1.81
2.56
2.95

3.10

4.67

2.74

1.44
2.73
3.14
1.52
2.62
5.35

5.02
1.36
1.74
2.64
2.70
2.06
1.42

3.30
4.32
2.59
3.91
2.05
3.88
2.67
2.42
2.65
2.02
1.40
3.01
3.99

70

1.57

1.62
2.94
3.56
4.27
4.95
2.04
5.05

2.65
2.17
3.36
4.19
3.95
4.38
8.85
2.29
1.71

1.69

3.51
4.65
2.80
1.93
4.54
3.90
3.32
4.65
2.43
7.78

3.56

1.79
4.03
6.71

2.92
0.96
1.02
1.17
1.72

3.96

2.76
2.29
4.10
4.68
3.54
2.40
2.75
2.13
4.70
8.34

1.06
2.64

1.63

4.75

1.78

4.60
6.29
1.68
4.67
2.12
4.87
3.05
3.46

2.33
2.15
2.51
2.85

Appendix C: Comparison of arrest rates for blacks relative to whites with those for
16-45 year old black males compared to 16-45 year old white males.

NIBRS data fi'om 13 cities used in this paper were used to construct the ideal
measure of total arrests of 16 to 45 year old black males. Using the 2000 PUMS, I
estimated the 16 to 45 year old black male population in each city. The black arrest rate
is thus the number of 16 to 45 year old blacks arrested in a city divided by their estimate
of the 16 to 45 year old black population in the city. Ideally, we would use only arrests
of residents of the city, but resident data has a high rate of “unknown” responses.

The corresponding variable is also calculated for white males aged 16 to 45. This
is used to calculate the preferred relative arrest rate for this paper (which cannot be
calculated using UCR data). The actual measure used in this paper is also calculated
using NIBRS data. The actual measure takes total arrests of all ages and sexes and for
each race and divides by the population of that race. The following gives the mean
relative arrest rates for these 13 cities and the correlation between the actual and ideal
measures. Though relative arrests are higher on average for the restricted group the

correlation between them is quite high at 0.88.

Table A2.3 Mean of Relative Arrest Rates for
Blacks Compared to Whites

 

Both Sexes and All Ages 2.95
16 to 45 year-old Males 3.33
Correlation of the above measures: 0.88

 

71

CHAPTER 3

THE IMPACT OF JOB DISPLACEMENT ON IMMIGRANTS

3.1 Introduction

The U.S. economy grew continuously from the first quarter of 1991 through the
first quarter of 2001.78 In periods such as these when the economy is growing, workers
who lose their jobs may not find it difficult to obtain new employment. However, as in
recent years, when the economy slows, unemployment rises and labor market re-entry
becomes increasingly difficult, increased attention turns to the labor market prospects for
unemployed workers. In times of such slow job growth, the flow of workers into
unemployment is not as likely to be cushioned by job creation elsewhere.79

Current debate centers around the effect of outsourcing jobs from the U.S. and the
continuing ability of the U.S. economy to absorb the people who lose their jobs.80 This
paper joins the growing literature81 on the group of workers who lose their jobs due to
slack economic conditions or plant, shift or company closings, by examining the role of
immigrant status for “displaced workers.” Hereafter, in contrast to persons who lose their
jobs due to incompetence or other personal reasons, I use the term “displaced workers” to

refer solely to persons losing jobs for the aforementioned economic reasons.

 

78 Source: National Bureau of Economic Research Business Cycle Dating Committee accessed at
http://www.nber.org/cycles/cyclesrmin.html on February 15, 2004. Though employment lags the measure
used by the Dating Committee, i.e., output growth, unemployment also fell over much of this period with
the exception of 1991-1992 and 2000-2001 (Bureau of Labor Statistics).
79 McMenamin, Krantz and Krolik (2002) discuss the 2002 job market and note that while some sectors
increased the number employed, there were widespread sectoral job losses that outweighed them.
8° Several recent articles in the New York Times discuss job loss. See, for example, Stephen Lohr (2004).
See also The Econorrric Report of the President 2004 and Kletzer (1998) for discussions of the effects of
eneral outsourcing both inside and outside the U.S.
' For surveys of this literature see Hamermesh (1989) and Fallick (1996). For more recent work see also
Farber, Haltiwanger and Abraham (1997), Kletzer (1998) and Chan and Huff Stevens (2001).

72

Recent concerns about displaced workers come fi'om the new constituency of
high-skilled workers who are being laid off. In the later part of the 20th century, the
effects of globalization were concentrated on tradable manufactured goods, which led to
displacement in the relatively low-skilled job market in the U.S.82 Displacement
occurring ostensibly due to NAFTA, advances in free trade and improvements in

3 More recently,

technology weakened labor market power of low-skilled workers.8
communications and technological advances are claimed to have led to an exodus of
white-collar service jobs from the U.S. mainland to Asian countries. These advances also
lead to outsourcing of traditional in-house services to outside contractors.84 Farber
(2003) finds that there has been an increase in displacement rates of workers with 16 and
more years of education over the early and mid-90$ and mentions that this is consistent
with reports of “substantial” white-collar job layoffs over the period.85 As a result, there
is heightened interest in the effects of technology, immigration and trade in services on
all workers in the U.S. job market.86

Previous literature regarding displaced workers has not focused separately on the

impact of displacement on immigrants. An exception is Chi and McCall (2004). They

note that rates of displacement are influenced by skill, industry and occupation. Chi and

 

'2 See the Economic Report of the President 2004 pp. 53-82. In particular, see chart 2-9, which shows the
increasing share of inrports that make up the total share of goods purchased in the U.S.

“3 Errrployrrrent Policy Institute briefing paper #147: Robert Scott “The High Price of Free Trade: NAFT A’s
failure has cost jobs across the nation.” November 17, 2003. See also Katz and Murphy (1991) for an
investigation of causes of declines in relative wages of less-skilled workers.

“ The Economic Report of the President 2004 discusses the “prominent” issue of the new irrrpetus to
outsource professional service jobs in addition to concerns about losses of manufacturing jobs and
“offshore outsourcing” (p.229). See also Bartel, Lach and Sicherman (2004), who discuss increasing trends
to outsourcing from of former in-house jobs to outside service firms.

8’ While the rates of displacement are still lower than that of the less-skilled workers, this absolute increase
in rates of displacement may be the result of increased outsourcing of jobs to other countries and outside
firms.

8‘ Kletzer (1998). In addition, Hamermesh (1989) finds that over time there has been a “secular” increase
in rates of displacement beyond the business cycle fluctuations.

73

McCall test whether immigrant status is a factor by which displacement rates differ.
Their paper also tests for differences in immigrant take-up rates of unemployment
insurance benefits. They find that newly arrived and non-citizen immigrants are less
likely to be displaced than natives. In addition, among qualified displaced workers,
newly arrived non-citizen immigrants are less likely to claim benefits than other
immigrants and natives.

The idea that immigrants may have different labor market experiences than those
of natives is not a novel one. There is a large literature concerned with the assimilation
of immigrants into host country labor markets (see, for example, Daneshvary, Herzog,
Hofler and Schlottrnann 1982; Borjas 1985; Schoeni 1987; Schoeni 1988; Butcher and
Piehl 1998; Friedberg 2000; Hu 2000; Butcher and DiNardo 2002). Much of this
previous literature is concerned with the observed achievement gap between immigrants
and natives. Irmni grants, on average, have worse labor market outcomes than similar
natives. These differences might even have an effect on second generation immigrant
performance (Hatton 1997).

Research on displaced workers continues to be relevant for policy formulation
regarding unemployed workers displaced by the recent recession and jobless growth
period,87 immigrant use of government income support programs (Borjas and Hilton
1996; Borjas 1999) and the relatively worse labor market performance of immigrants.
This paper adds to the literature by examining whether displaced immigrants fare worse
than displaced natives or whether they are better able to adjust to job loss. Given the
recent efforts to restrict immigrant use of government welfare programs (Zimmerman and

Tublin 1999; Ku and Freilich 2001) and the expectation and trends towards greater rates

 

'7 McMenamin, Krantz and Krolik (2003).

74

of displacement (Kletzer 1998; Farber 2003; Chi and McCall 2004), added information
on the experiences of these workers is important. Following Farber (2003), I examine the
probability of survey-date employment after displacement in the prior three years and the

change in weekly earnings for workers re-employed full-time after losing full-time jobs.
3.2 Conceptual Framework

Following Farber (2003), I estimate a linear regression model controlling for sex,
non-white race, education, age and tenure on the lost job. To address the question of how
immigrants differ fi'om native, an immigrant dummy is included.88 To address potential
discrimination against people reporting Spanish ancestry, who largely report white as
their race and comprise roughly half the weighted sample of immigrants, an indicator of
Hispanic origin is also included.89

The following regression is estimated:

Y = [30 + [311mm + BzHisp + B3Female + [34Nonwhite + BsEd0_11 + B6Ed13_15 +
B7Ed16_ + BgAge25_34 + BgAge35_44 + BloAge45_54 + BrrAge55_64 + BrzTen2_3 +
013Ten4_10 + [314Ten11_20 + BrsTen20_ + BrsYSDZ + [517YSD3 +1313 Unem + e (3.1)

The following indicators are used as controls: Irnm — non-US country of birth;

Hisp — Hispanic; Female — gender is female; Nonwhite — race is not white; Ed0_11 — 11th

grade or less; Ed13_15 — some college, no degree; Ed 16_ - Bachelor’s or higher degree;

 

88 Reitz and Sklar (1997) found differences in ethnic assimilation pressures by race. Specifically,
European-origin immigrants face cultural assimilation pressures and economic costs due to cultural
differences, but racial minority immigrants experience discrimination on the basis of their skin color rather
than ethnic and cultural differences with the majority. As 71% of the sample of irnnrigrants is white, the
effect on this group is expected to dominate. Thus, racial interactions with immigrant status are not
included.

'9 Additionally, based on penalties to illegal status relative to legal immigrants found in Phillips and
Massey (1999) and Kossoudji and Cobb-Clark (2002), the Hispanic variable is needed to absorb some of
the effect of illegal status that may be captured by the immigrant status variable. Of the estimated 5 million
illegal immigrant residents in the U.S., over 70% are believed to be from Hispanic countries. See Estimates
of the Undocumented Immigrant Population Residing in the US: October 1996 accessed at
http://uscis.gov/graphics/shared/aboutus/statisticslillegalalien/illegalpdf on May 28, 2004.

75

Age25_34 — age is between 25 and 34 years inclusive; Age35_44 - age is between 35 and
44 years inclusive; Age45_54 — age is between 45 and 54 years inclusive; Age55_64 —
age is between 55 and 64 years inclusive; Ten1_3 - tenure on lost job of l to 3 years;
Ten4_10 — tenure on lost job of 4 to 10 years; Ten11_20 — tenure on lost job: 11 to 20
years; Ten20_ - tenure on lost job: 20 or more years; YSD2 — 2 years since job loss;
YSD3 — 3 years since job loss; Unem -— MSA unemployment rate in the year prior to the
survey; a - random error; Y is the outcome variable, which is either a survey date
employment indicator or change in log real earnings between current and old job.90

The motivating question is: do immigrants who are displaced suffer greater
economic losses than natives? In periods of recession, when search queues for jobs are
longer, immigrants have been the target of political rhetoric which claims immigrants
take jobs away from native-bom Americans with the argument that immigrants have
lower reservation wages.91 Have immigrants experienced a backlash in the form of
discrimination in hiring in light of NAFTA, other trade agreements and weakened labor
markets of less-skilled workers? Do immigrants really take jobs away from U.S. bom
Americans due to lower reservation wages? In the latter case, we would expect to see
relatively higher probabilities of employment for displaced immigrants compared to
natives, combined with relatively lower earnings conditional on employment.

New immigrants tend to settle in areas where there are already large populations
of established immigrants (Bartel 1989). To the extent that areas with high
concentrations of immigrants have better (worse) labor markets for displaced workers

relative to those with higher proportions of natives, displaced immigrants may have

 

9° The CPI-U is used to adjust earnings to 2003 dollars.
9' McMenarnin, Krantz and Krolik (2002) note that with weak employment growth late in 2002, duration of
unemployment increased.

76

easier (harder) labor market re-entry experiences than natives. In truth, the distribution of
immigrants across the U.S. may account for any differences found by immigrant status,
when immigrants have the same experience as natives in the same region.

To address this distribution issue, in an alternative specification, I include
metropolitan statistical area (MSA) fixed effects, in addition to the variables used in
equation (3.1), to estimate the effect of immigrant status. Also, accounting for MSA
fixed effects helps to address the possible endogeneity of the initial location of
immigrants within the U.S. on the basis of maximizing expected income.92 In other
words, new immigrants may go where labor market re-entry is easier due to better job

opportunities or lower resistance to immigrant hires.93

3.3 Data

The data come from Current Population Survey (CPS) Displaced Worker
Supplements (DWS) from 1996, 1998, 2000 and 2002.94 These data ask workers 20
years and older whether they have lost a job in the 3 years prior to the survey year. The
sample is restricted to those who have lost jobs due to plant/company closing, insufficient
work or position/shift abolished, rather than all those who lost a job. This is done to
exclude low productivity workers who may induce their job loss and have subsequent

difficulty obtaining employment.

 

92 For example, Borjas (1999) finds that immigrant welfare recipients are more likely than other immigrants
or natives to live in states with higher benefits.

93 One example of this would be immigrants who are hired on the basis of labor shortages in an area for
certain types of jobs. For example, Massey et al. (1994) p. 705 note that farm workers, laborers and crafi
workers were over represented in Puerto Rican rrrigrants to the U.S. partially due to existing job offers in
the U.S.

9‘ While immigrant status is available from the 1994 DWS onwards, MSA identifiers are not available in
1994.

77

Table 3.1 gives weighted sample means and counts for selected variables used in

95 Immigrants are more likely to be nonwhite and are

this paper by immigrant status.
almost ten times more likely than natives to be Hispanic. The proportion of immigrants
re-employed at the survey date is about 4 percent lower than the rate for natives.
However, immigrants who lost full-time jobs and manage to obtain firll-time re-
employment have lower earnings losses than natives. While these immigrants are re-
employed at earnings about 1.4% below their earnings on the lost jobs, natives earn about
10% less on average in the current jobs. For the sample of all workers whose weekly
earnings on both jobs are reported, displaced immigrants surprisingly see weekly earning

growth on average.96 Unsurprisingly, based on the immigrant assimilation literature,

immigrants earned less than natives both before and after displacement.
3.4 Results

To address the issue regarding the effect of immigrant status at different points in
the business cycle, I will briefly discuss some results not shown here. I estimate equation
(3.1) without MSA fixed effects in each of the survey years. As MSA specific effects are
not included in these annual regressions, I include the 1994 CPS DWS in this exercise.
As a check on the methodology, I closely replicate Farber’s results with the estimation of
equation (3.1) without the immigrant status and Hispanic indicators. I find that the effect
of immigrant status does not significantly affect re-employment likelihoods in most of the
survey years (mainly boom periods). However, data from the 1994 survey suggests that

immigrants were significantly less likely than natives to be re-employed in early 1994.

 

9’ Sampling weights from the CPS Outgoing Rotation Files are used.

9° Estimates of weekly earnings growth among immigrants and natives from the CPS March files show that
immigrants had faster weekly earnings growth overall from 1994-2002. See Appendix A for details on
these calculations.

78

These workers were displaced over the 1991-1993 period when the economy was just
starting to recover from the previous recession. Generally, immigrant status positively
affects the change in weekly earnings, so that immigrants have lower earnings losses than
natives. Workers displaced in 1991 through 1993 have a negative (though insignificant)
estimated effect of immigrant status. This period led up to the 1992 Presidential
elections, when the economy was emerging from a recession and included heated debate
about the labor market effects of immigration and trade. Hence, it is possible that
immigrants face greater difficulty when jobs are hard to come by, but are no worse of in
other years when the economy was generally experiencing growth

Table 3.2 column (1) gives coefficients from a regression pooling the 1996-2002
DWS surveys to estimate equation (3.1). It examines the likelihood of survey—date
employment of workers displaced from 1993-2001. Results show that immigrants were
no different from natives in their propensity to re-enter the employed part of the labor
force, while Hispanics are significantly less likely to obtain re-employment by the survey
date. Roughly half of immigrant workers are Hispanic; with the omission of the Hispanic
indicator the immigrant coefficient is significantly negative.

Table 3.2 column (3) examines the effect of immigrant status on the change in log
weekly earnings. Immigrants employed full-time, who had lost full-time jobs, had
eamings losses 8.5% lower than those for similar natives. Hispanics did not have
different earnings losses conditional on full-time to full-time employment transitions.

Table 3.2 columns (2) and (4) re-estimate the effect of immigrant status including
MSA specific effects to control for pre-existing conditions such as growing labor markets

which may have historically attracted immigrants. Column (2) results suggest that

79

immigrants were actually concentrated in cities with lower probabilities of re-
employrnent. From Table 3.1 recall, immigrants were less likely to be employed at the
survey date. From the results in column (2), however, if an immigrant and native reside
in the same MSA, the immigrant is about four percentage points more likely to be
employed at the survey date; this difference is significant. It should be noted these results
include annual MSA unemployment rates. Column (4) shows that the result that
immigrant status implies lower earnings losses is robust to the addition of MSA effects,
though the point estimate is slightly lower. Hispanics are less likely to be re-employed,
but are no different from non-Hispanics in their loss of earnings from the old job to the

current job.

3.5 Why Might Immigrants Be Hurt Less by Displacement?
Exploring the Results

As displaced immigrants appear to suffer somewhat less than natives, an
investigation of potential reasons for this result is in order. I explore three possibilities
for the more positive results for immigrant workers. First, immigrants are concentrated
in jobs with low earnings and as such cannot experience as much earnings loss as higher
paid natives when moving from one full time job to another. For example, a person who
loses a full time minimum wage job and then is reemployed full time is unlikely to
experience any wage loss at all. In addition, if immigrants accept lower wages than
similar natives, immigrants may be more likely to be re-employed. The hypothesis that
immigrants exhibit greater flexibility in the labor market after displacement is explored
by examining whether immigrants are more likely than natives to turn to self-
employment or switch occupations. Finally in this section, I explore the possibility that

immigrants search more intensely for jobs following displacement because they may

80

have different access to government income support programs that ease the burden of
being unemployed than natives (Chi and McCall 2003; Borjas 2003) or they have less
non-labor income.
3.5.1 Earnings on the Old Job

Recall from Table 3.1 that immigrants have lower weekly earnings than natives
(among those formerly full-time workers who are re-employed full-time) in both the
current and pre-displacement jobs. I test the hypothesis that lower earnings losses of
immigrants are due to their concentration at the lower end of the weekly earnings
distribution limiting their potential loss conditional on full-time re-employment. I
calculate quintiles for real weekly earnings on the lost job in each of the survey years.
Indicators of the highest four earnings quintiles on the lost job are included as additional
variables in the specification that adds MSA fixed effects to equation (3.1). Results for
the survey-date and earnings loss regressions are shown in Table 3.3 columns (1) and (2).

Clearly, the higher the earnings quintile on the old job the greater is the
probability of re-employed at the survey date, holding constant other observable wage
determining factors like tenure and education. However, adding the pre-displacement
earnings quintiles has almost no effect on the coefficient of the immigrant dummy, which
remains positive and significant in the re-employment equation. Adding the earnings
quintile variables affects the earnings loss regression, making the coefficient of the
immigrant dummy statistically insignificant. It appears that the apparently smaller
percentage job losses for immigrants are due to the fact that, even after controlling for
observable wage determining characteristics, they tend to be concentrated in lower

earning jobs, and thus have less to lose.

81

3.5.2 Labor Market Flexibility of Immigrants

In this section, I investigate whether displaced immigrants are willing to do more
to be gainfully employed than similar natives. Are immigrants more likely to go into
self-employment rather than remain unemployed? Are immigrants more willing to
switch occupations? To the extent that workers are displaced from declining
occupations, they will be re-employed faster if they are willing to learn a new trade.
Controls from the preceding regression (Table 3.3 columns 1 and 2) are used in
regressions of indicators of self-employed or switched occupations; results are reported in
Table 3.3 columns (3) and (4) respectively.

In this sample, immigrants are not different from natives in their propensity to
enter self-employment after displacement. Immigrants are, however, significantly more
likely to report that they are currently in a different occupation from the occupation on
the job they lost. This is consistent with the idea that immigrants are more willing to lose
occupation-specific human capital and try something new in order to become re-
employed.

3.5.3 Government Income Support, Non-Labor Income and Search Intensity

Borjas (2003) finds evidence that after welfare reforms enacted in 1996 curtailed
the eligibility of non-citizen immigrants to participate in various income support
programs —- Aid to Families with Dependent Children (AFDC)/ Temporary Assistance to
Needy Families (TANF), Medicaid, Food Assistance or Supplemental Security Income
(SSI) — immigrant labor supply increased in response. In this section, I test whether
greater success of immigrants in finding re-employment results from greater search

intensity, as their non-work alternatives (government income support) will appear less

82

attractive. Lower welfare eligibility (or even perceptions of lower eligibility) for non-
citizen immigrants (almost 57 % of the sample of immigrants) may have led to lower
participation rates in government income support programs among non-citizen
immigrants.

Borjas (1999) shows although there were substantial declines in the participation
rates after the 1996 reforms in a several welfare programs, immigrant participation rates
remained above native rates. He also notes that in response to restrictions on welfare use
by non-citizens many qualified non-citizens would apply to become naturalized citizens,
nullifying the restrictions on welfare use.97 However, there are several reasons to believe
that immigrants may be less likely than natives to use welfare programs in the period
used here. First, as mentioned earlier there were welfare reforms in 1996 which restricted
non-citizen immigrant use of public welfare (Zimmerman and Tublin 1999; Ku and
Freilich 2001). Though welfare reforms mainly restricted use by immigrants arriving
after 1996, if immigrants mistakenly believe they are ineligible participation rates will
decline. 98’” Second, Borjas (1999) notes immigrants stay longer on welfare. Time limits
for welfare use, which were part of the 1996 welfare reforms, may reduce the eligibility
of immigrants to a greater degree than that of natives. Finally, Haider, Schoeni, Bao and
Danielson (forthcoming) note that much of the decline in welfare participation found in

Borjas (1999) and similar studies can be explained by macroeconomic improvements

 

97 In this analysis I do not distinguish between citizen and non-citizen immigrants.

98 Eligibility may not be an issue in this sample as over 90% of the immigrants arrived prior to 1996.

99 Borjas (1999) discusses the possibility that immigrants will reduce their participation in welfare even if
their eligibility does not decline. He also discusses the fact that even though the language of the legislation
restricted use of non-citizen immigrants who were in the U.S. prior to the legislation, this was not fully
enforced.

83

over the period. If immigrants had greater opportunities over this period than earlier ones,
their relative welfare participation may fall.

Borjas’ findings may not be representative of government assistance use by
immigrant status in the current work for a few reasons. First, the age group of the
displaced worker sample in the current work is restricted to 20 through 64 years, whilst
Borjas’ study of immigrant welfare use is at the household level and there are no
restrictions on the age of the household head. Second, Bod as’ work shows participation
rates only up to 1998, which includes only 2 of the 4 years used in this paper.100 In
addition, I define immigrants solely as foreign-bom persons. In Borjas’ work, second
generation immigrants’ welfare use is classified as immigrant use if the household head is
an immigrant. Another issue is that displaced workers are atypical. Harnermesh (1989)
notes displaced workers are considered to have strong labor force attachment. This
cannot be said of the typical welfare recipient.

In order to obtain representative government assistance participation rates for the
relevant groups used in this section of the analysis, I first calculate take-up rates of
unemployment insurance benefits for the current sample. Appendix B Table A3.2 shows
the take-up rate of unemployment compensation by sex, employment status and
immigrant status in the current sample. Calculations use CPS sampling weights. Not
surprisingly, currently unemployed displaced workers have a higher take-up rate of
unemployment benefits than do currently employed displaced workers. Surprisingly,
considering findings of higher welfare participation rate of immigrants (Borjas 1999;

Borjas and Hilton 1996), displaced male immigrants are 8 percentage points less likely to

 

'00 Recall, after the passage of time, time limit use on welfare should further reduce immigrant participation
even relative to native declines for this reason.

84

receive unemployment benefits than displaced male natives from the group employed at
the survey date. Unemployed displaced male immigrants are 4 percentage points less
likely than unemployed displaced natives to receive unemployment benefits. Even
female immigrants are equally or only slightly more likely to use unemployment

'01 it seems unlikely

compensation. While I am unable to restrict the sample by eligibility
that difference between the participation rates of male natives and immigrants is due
entirely to eligibility or eligibility perceptions, since the take-up rates for female
immigrants is so similar to that of natives. From the results for currently unemployed
workers in Table A3.2, male natives have the greatest use of unemployment benefits,
while female immigrants, male immigrants and female natives follow (in that order). The
figures do not, however, differ markedly from one another.

Since women are the main beneficiaries of public welfare, the social safety net for
displaced women may be greater than that for men. Borjas and Hilton look at several
welfare programs whose main beneficiaries are women (Currie and Grogger 2000;
Kaestner and Kashaul 1997) and find that on average immigrants have much greater
participation rates in these programs than natives. Thus, male immigrants may have the
lowest income support from these programs, while female immigrants the greatest.

The difficulty in ascertaining which group has the greatest motivation to search
for a new job is apparent. I have chosen to calculate the relative degree to which each
group will be motivated to search using the CPS Annual Demographic (March) files for

1996, 1998, 2000 and 2002. These data contain information on income received from

various government income support programs. In addition, they contain income of other

 

'0' Recent immigrants may not have the required duration of employment to qualify. However, fewer than
10% of immigrants in the sample arrived within the three years prior to the survey. Thus, controlling for
tenure on the job, this form of income support should be open to most of men and women in the sample.

85

household members. First, I examine participation in various government income
support programs (public assistance, supplemental security and unemployment
compensation) in the previous year using a linear regression model with a variety of
demographic, geographic and time controls. I control for weeks unemployed in the
previous year and non-labor income available fi‘om the household. Annual non-labor
income available to an individual is calculated as household income minus own wage and
salary income. Second, I examine differences in access to non-labor income between
immigrant-gender groups. Further details of the procedure are in Appendix B.

To test the hypothesis that differences in govemment income support
participation influences search intensity, I replace the immigrant dummy by three
indicators (male immigrant, female immigrant and female native) in the survey-date
regression including MSA fixed effects and quintile dummies. Under the first hypothesis
that government income support differences drive search intensity I expect that female
immigrants should be re-employed in the highest proportion as controlling for other
factors they are the least likely to receive part of their income from the government (see
Appendix B Table A3.3 column 4). Male immigrants and female natives should be re-
employed in the next highest proportions in that order. As overall government income
support use by female and male immigrants are not significantly different, all immigrants
may have higher search intensity than natives. The lowest re-employment rate is
expected for the male natives. Under the hypothesis that non-labor income represents the
cushion that individuals can fall back on in the event of displacement, the following order
of re-employment is expected: male immigrants should search the hardest followed by

male natives, female immigrants then female natives (Appendix B Table A3.3 column 5).

86

Table 3.3 column 5 shows that female immigrants are 4.8 percentage points less
likely to be re-employed than male immigrants, who are the most likely to be re-
employed. Female immigrants are about 2.5 percentage points more likely than female
natives to be re-employed, but the difference is insignificant. Female natives are 1.2
percentage points more likely than male natives to be re-employed at the survey date. 80
the hypothesis of search intensity (greatest to least: female immigrants, male immigrants,
female natives, male natives) is contradicted by the finding of re-employment order here
(highest to lowest: male immigrants, female immigrants, female natives, male natives).

Similarly, the findings presented in Table 3.3 colurrms 5 through 7 are contrary to
what the non-labor income motivation hypothesis predicts (greatest search motivation to
least: male immigrants, male natives, female immigrants, and female natives).102 Results
in Table 3.3 column 6 control for marital status. The regression results in column 7
control for marital status interacted with female indicators. None of the results suggests
that either the government income support or non-labor income fully determine the
degree to which each group returns to work. The ranking of this last hypothesis holds
with the exception of male native re-employment rates. Quite likely, access to assets can
explain the lower than expected re-employment rates of male natives. I cannot account

for asset holdings with the CPS March files.
3.6 Heterogeneity among Immigrants

Immigrants are a rather heterogeneous group. This section examines differences
in re-employment outcomes between immigrants. As discussed earlier, non-citizen

immigrant access to welfare was limited by welfare reforms in 1996. In addition, citizens

 

"’2 See Appendix A for details on the calculation of the ranking of search intensity.

87

qualify for a number of federal jobs that are not available to non-citizen immigrants. I
estimate the regressions by citizenship status of immigrants to see if there are any
differences in their re-employment outcomes. Borjas (1985) suggests that source country
of immigrants matters for immigrant quality and assimilation in the U.S. To address
concerns that immigrants’ source country may capture differences in human capital (or
value of such in the U.S. labor market), I break immigrants into groups by their region of
origin. Hispanics, however, are grouped together due to language and cultural
similarities. Finally, I ask whether the blow to immigrants created by displacement
might be cushioned if they live in areas with greater populations from their own country.
3.6.1 Citizenship

Table 3.4 columns (1) and (6) are estimates of regression (1) with MSA fixed and
quintiles. They include only immigrants who report that they are citizens. Columns (2)
and (7) are similar but include only non-citizen immigrants. Besides the fact the non-
citizen immigrants have higher point estimates than naturalized immigrants, the earlier
results that immigrants are more likely to be re-employed and have similar earnings
losses, when compared to natives, holds.
3.6.2 Region of Origin and Immigrant Type

As in section 3.5.3, I split the original immigrant indicator into several indicators.
Firstly, all immigrants reporting Hispanic ancestry are assigned to one group regardless
of region of origin. Indicators are also created separately for Asian, European and West
Indian immigrants. I group Middle Eastern and African immigrants into one indicator

due to small sample sizes from these areas. Immigrants from other part of the world are

88

assigned to one of the aforementioned groups.103 Hispanics are the only immigrant group
more likely than natives to be re-employed. Most other immigrant groups (Asians are the
exception) are less likely to be re-employed than natives (Table 3.4 column 3). Table 3.4
column (8) shows that while European and Asian immigrants have lower re-employment
likelihoods, they have lower earnings losses. This suggests that these groups may have
higher reservation wages than other groups.
3.6.3 Ethnic Networks

The literature on the impact of ethnic networks provides mixed evidence on the
extent to which networks improve labor market outcomes of co-etlmics. Here, the term
co-etlmic refers to immigrants from the same country. Reitz and Sklar (1997) note
traditional theories view ethnic network use as limiting opportunities. Similarly, Nee,
Sanders and Semau (1994), who interview Asian immigrants, finds that many prefer jobs
outside the ethnic enclaves because of higher wages available to them in the outside labor
market. Tienda and Lii (1987) test whether enclaves improve outcomes for minorities by
protecting against discrimination and raising self-esteem, but find no support for this
argument.104 In contrast, Wilson and Portes (1980) find that Cuban immigrants working
within an enclave earn similar returns to Cuban immigrants in the general labor market.
Potentially, therefore, larger populations from one’s own country may improve the
outcome of immigrants but it remains an empirical question.

This section tests whether larger ethnic networks mitigated effects of

discrimination and/or possibly provided access to information and opportunities that may

 

'03 Immigrants from Oceania are grouped with Asians. Immigrants from Canada, New Zealand, Australia
and the former Soviet Union are grouped with Europeans.

104 They also discuss previous literature suggesting that initial limitations give way to greater political
power as the minority population increases, leading to economic access on a previously unattainable scale.

89

not exist without networks to assist job search. To measure ethnic networks, population
counts by country of birth of immigrants were obtained at the MSA level from the 2000
Public-Use Microdata Sample (PUMS) fi'om the U.S. Census. The sample is restricted to
residents in MSAs identified in both the CPS and PUMS files. Each immigrant’s
network is measured as the number of peOple from the immigrant’s country residing in
his or her MSA in 2000, divided by the MSA population.105 The network measure for
non-immigrants is set to zero.106 Figure 1 provides the weighted sample distribution of
values of the ethnic population share variable. The largest population shares are for
Mexicans and Cubans. Most other immigrant shares were below 15% of the MSA
population.

Table 3.4 column (4) shows that after controlling for network strength, immigrant
status still increases the probability of re-employment. However, larger networks do
positively affect immigrants’ survey-date employment likelihoods. Since as expected,
having a non-zero ethnic network exerts a positive effect on the probability of survey-
date employment, immigrant groups will have different experiences based on the size of
the population from their own country in an area. For example, displaced Mexicans in
New York, where approximately 1.2% of the population is Mexican, have a higher
probability of being re-employed than natives of 5.9 percentage points, while displaced
Mexican immigrants in Fort Lauderdale, where the Mexican population share is only

0.6%, experience only 4.8 percentage points increase in their relative probability of

 

"’5 It must be noted that this calculated measure will generally understate the population it targets as several
continent level answers are allowed (e.g. Other Africa). Thus, country of birth cannot be assigned for some
innnigrants.

'06 I assume that natives have the same network. Under this assumption, the effect of the general
preferences for natives over and above immigrants (since they are a majority) is captured by the immigrant
dummy.

90

being re-employed. Meanwhile, displaced Jamaicans in New York, where they constitute
1.4% of the population, have an increased probability of finding re-employment of 6.1
percentage points when compared to natives. In Fort Lauderdale, where Jamaicans make
up 3.9% of the population, Jamaicans should have an even greater chance of re-
employment (8.8 percentage points more) when compared to natives in Fort Lauderdale.
It should be noted, however, that the immigrant dummy is still positive and significant -
immigrants in cities in which no co—ethnics reside still have higher re-employment
probabilities than comparable natives.

On the other hand, larger ethnic networks imply significantly larger earnings
losses for immigrants displaced from full time jobs and re-employed in full time jobs. As
mentioned earlier, the earnings loss result is not completely surprising given the findings
of Nee, Saunders and Semau (1994) for Asian immigrants. Even the results of Wilson
and Portes (1980) — Cuban immigrants in enclaves earn returns similar to Cubans

'07 without culturally assimilating (e. g., acquiring English

earnings in the primary market
proficiency) — can be consistent with these findings. Lazear (1999) argues that larger
ethnic communities lower pressure to assimilate. Thus, workers displaced from an
enclave job may be disproportionately poor English speakers. It is well documented that
English language proficiency affects wages in the general labor market (see McManus
1990 and McManus, Gould and Welch 1983). While there may be no penalties for poor
English skills on the old job, poorer English-speaking immigrants (who are

disproportionately in cities with large enclaves) will experience lower wages, if rc-

employment is not found within enclaves.

 

'07 The term primary market comes from theories of a dual labor market. The primary market contains
good jobs while the secondary market contains less desirable or bad jobs.

91

I allow for differences by region of source country and include a set of
interactions between immigrant type and the network variable (Table 3.4 columns 5 and
10), I find that Hispanics drive the positive results for immigrants’ re-employment. No
other group is significantly more likely to be re-employed than natives. Even after
accounting for earnings quintile on the old job, Hispanics immigrants have larger
earnings losses than natives, while Asian and European immigrants have lower earnings
losses. This is consistent with the hypothesis that Hispanic immigrants, at least, have
lower reservations wages and this is assisting them to obtain re-employment.

To summarize, the fact that even after accounting for the ethnic network size,
immigrants are more likely to be re-employed than natives implies that networks do not
fully explain the success of immigrants. Overall immigrants have better re-employment
outcomes than non-immigrants even with no ethnic network. Nevertheless, the ethnic
network analysis provides some interesting insight into the factors affect the success of
displaced immigrants.

3.6 Conclusion

This work examines whether displaced immigrants face more difficult re-entry
into the labor markets than natives. I find that immigrant status increases the likelihood
of re-employment after exogenous displacement. In addition, immigrant status mitigates
the earnings losses for those displaced from full-time jobs, and re-employed full-time.
Taken together, the employment and earnings results are consistent with the hypothesis
that immigrants who are displaced face a higher demand for their services than do
observably similar natives. However, the lower earnings losses of immigrants can be

explained by their concentration in the low end of the earnings distribution on the lost

92

job. Immigrants exhibit greater flexibility in their re-entry to the labor market as they are
more likely to switch occupations to gain re-employment. Immigrants are not different in
their propensity to become self-employed. Results of analyses suggesting that welfare
reforms affecting immigrant participation in government income support programs or
differences in non-labor income caused differences in search intensity for re-employment
yielded inconsistent results.

Larger ethnic networks positively affect re-employment probabilities for
immigrants, but this effect alone cannot explain the higher re—employment probabilities
of immigrants relative to natives. On the negative side, larger ethnic networks increase
eamings losses among currently full-time immigrant workers who were previously
displaced from full-time jobs. Further exploration disaggregating immigrants by source
region revealed results consistent with the idea that Hispanic immigrants have lower
reservation wages than natives and are thus more likely to obtain re-employment, while
Asian and European immigrants have higher reservation and are less likely to obtain re-
employment by the survey date.

The immigrant population share is increasing. As such, knowledge of their
assimilation and labor market experiences is important as the debate continues on
whether immigrants are net contributors in their use of government income supports
programs. This work documents a surprising and robust finding that immigrants
displaced are more likely to be re-employed than natives. Further investigation of this
feature may reveal policy prescriptions to help cushion the blow to increasing populations

of displaced workers as technology allows for greater trade in goods and services and

93

perhaps international wage convergence. The findings suggest that if natives lower their

reservation wages their re-employment prospects will increase after displacement.

94

Weighted Fraction of Sample

0

Figure 3.1 Ethnic Population Share Distribution

      

 

 

 

— H
H
E
a H
E;
H
H
‘
3-'E=a:nm_.aa_.a_ 5. -rr- -
003013 .0'5 .i .115 .2' .23 .291671
Ethnic Population Share

95

Table 3.1. Mean of Selected Variables by Immigrant Status.*

 

N

Survey Date Employment
Weekly Earnings Old Job
Weekly Earnings Current Job
Change in Earnings (%)
Hispanic

Female

Nonwhite

Education (median years)
Tenure on the lost job (years)
Age (years)

N**
Weekly Earnings Old Job***

Weekly Earnings Current Job
Change in Log Earnings (%)

 

 

Natives Immigrants
12490 1641
0.72 0.68
646.35 541.36
600.18 557.73
-7. 14% 3.02%
0.06 0.50
0.45 0.42
0.14 0.27
13-15 12
4.7 3.6
38 39
7587 935
$671 $573
$620 $565
-10.2% -1.4%

 

96

Ox: ...-... 2% ...... ”:2 a 8:88:83 88885 ...

 

 

a 8.8 88.8 ...-.888 ...-...: 8.8
888 888- 888 888 88889 88:8
68.8 88.8 ...-.8888 ...:888
888- 288- 888 888 8285 oz 88:8 2:8
88.8 88.8 ...-.888 ...:688
. 888. 888- 388- 388- 2:an .8888 88E 5.8 33
£88 28.8 ...-...: 8.8 1.1.88.8
888- 88- 828- 8888- 233-82
.888 88.8 ...-2888 ...-A888
888 ~88 888- 888- 225.8
88.8 888 ...-1.88.8 ...-:88
8.88 888 888- 888- 2582:
...-$8.8 2.88.8 ...-58.8 88.8
88 388 888 888 88886—
888 88.8 ...-88.8 ...-288.8
888 388 8:8 888 8388
meEwm NEoQB we; E 09:25 “58829:”; 380 525
A8 5 S E

 

88-88 528 83285 $883 88 seem 852.8858 888m 88 288-8

97

ox; ...: axon ...... 2.8. a 850888: 888:. ..

 

 

 

mo> Oz mo> Oz
88 88 :8 .8
3% $8 8:; 8:;
888.8 ...-c.8888 3,988.8 ...-.8888
_ 88- 88- _88- :88-
..._.....G_8.8 ...:8 8.8 ...-3.8888 ...-.888
888 :88 8.8 888
...-.688 ...-.888 ...-$888.8 3288.8
888 888 £- _ 8 8 _8
:2888 ...-88.8 88.8 28.8
88- 288- 888- 888-
1....888 ...-.588 ...-228.8 ...-2.6888
8:8- 838- 888 888
188.8 ...-.888 1...: 8.8 ...:358
£88- .88- 888 888
8 8.8 888 ...-L888 ...-...: 8.8
«.88 888- 888 .88
188.8 ...-88.8 ...-58.8 ...-.888
888- 888- 8 .8- 8:8-
1.888 ...-3.88.8 .888 68.8
8.88- 388- 888- 888-
...-.888 £888 $8.8 38.8
888- 888- 88- 888-
68.8 88.8 .888 G 8.8
888- 88- 888 :88

 

meF—wm bxoog mo; 5 owe—EU

A8

5

EoEAOEEm Ban tag-Sm

5

A:

mace-«mo cox: <92
nouns—um-”—
3033.530

83— “conic—9:023 _m::=<

804 no.- ooEm mane» m
8o... no.- ooEm Emu» N
memo» + om ounce-r
83% ON 9 ~ _ ounce-r
88% A: 9 v 0.55%
Emu» m 3 _ 23:“:-
m-Eo» we 3 mm ow<
memo» Vm 8 mv ow<
mum?»- vv 3 mm ow<

Emu» VM 3 mm ow<

88-88 :58 88.85 38:83 .8 2:3 88888 8 88m 8858 N8 833

98

.o\°_ ...: 88m ...... 88a— 8 858558 3:823. .. 8805:08a :_ 0.8 825 van—8:85

 

 

 

 

3.8 888 3.8 888 88 88 888 8388?:
83.2 832 88.2 88 8:8 8% 83: 22.82880
......«GSdfl
wmod Com-Cm: ao>oz ... 2&th
.....iQ-odv
hwod cofifiqom 603023 609530 ... 288$
...-88.8 888.8
888- 888- 8882 .262
...-88.8 388.8
888- _ 88 832838 .8523 883::
3.888 88.8 288.8
_88- _ 88 ~88 2:82 225:
38.8 258.8 12.88
888 888 :88 22888: 2258
...-88.8 ...-38.8 ...-G88
8888 888 288 2.8855 222
.888 c 8.8 88.8 :18 8.8
888 888- 88- 888 2.2985
...-88.88 2:88.88 ...-A288 ...-388 88.8 ...:888 ...-88.8
888 $88 288 2:8 8 n 88- 888 W 2:580
...-.388 ...-38.8 ...-38.8 ...-:88 ...-888.8 ...-.688 ...-A288
388 88 38 888 $88- 8888- 9-88 8 28:30
...-888 1.588 ...-88.8 58.8 ...-888.8 ...-288.8 ...-88.8
8 _ 88 88 88 888 288- 8.38- .88 8 085:0
...-:88 ...-:88 ...-.288 38.8 ...-888.8 ...-68.8 ...-:88
888 888 888 M88- 388- 888- 888 N 28.80
...-:88 ...-38.8 ...-.288 ...-88.8 38.8 ...-.888 ...-28.8
388 888 3.88 888 888 888 888 .8528
EofihoEEm cyan-hotam :otaasooo Baa—QEm mfiflpfim «SE-336m
vowgzu ..zom mod :_ owSEU o:aD->o>._:m
E 8 68 E 88 E E

 

 

8858.5 «:Ewafim 80:3 wifiwzmgfi 328538 80¢ wEoEEooU .m.m 038-.-

99

88. ...: 88m ...... 882 8 3:85:38 3:329: ... 88055.8: 5 28 act-“— ESESm

 

 

 

8.8 2 8 a 8 8.8 8 888 88 88 888 88 8233-8
88 88 88 88:. 8:. N _ a. N. E N. m: on: _ 82 _ 22823.8
82.: @888
8:8 88.? 89502 585 8.03
2388 $8. _ 8 8382
2:: 82.8- §E§E§m 2822
8888 33.8 88382 582.
:88- mm:- ~o§__e.s<\=§osm
38.8 ...-288.:
838 3..”- v.5382 5?.
33.8 ...-:88
58- .888 88382 25%:
...-A388 ...-82.8
.8 _- m _ 88 82202 8.58
8. .8 828 $8.8 ...-$8.8 eeseé 585 ~83
888- 288 8888- .N _ 8-
5 _ .8 5 _ .8 $888 ..A _ 8.8 sages. 88::
E8 .28 88. .28- Eagm 2822
£2 _.8 ...-am. .8 L _ 8.8 ...-$8.88 88888 828
$8 888 888- 8.8- 3.8838288
:8. ..8 L _ 2.8 A _ 8.8 $8.8
N88 88 N88 888- 82825 :22
£288 a: ..8 £88 ...-£88
888- :3 888 :88 8.2888 2588:
£88 @888 $8.8 ...-$8.8 ...-88.8 2.388
888 888- 88 :88 888 888 23855
358% :_ owcanu «coaonEm Ban-828m
a: 5 8 E 8 6 E 5 € €

 

 

58:0 820: EB :55 mczaéam mcofimomoz 8 858580 .3“ 033-

100

References

Bartel, Ann, Lach, Saul, and Nachum Sicherrnan. 2004. “Outsourcing and
Technological Change.” Unpublished Paper: Columbia University.

Bartel, Ann P. 1989. “Where Do the New U.S. Immigrants Live?” Journal of Labor
Economics 7 (4): 371-391.

Blau, Francine. 1984. “The Use of Transfer Payments by Immigrants.” Industrial and
Labor Relations Review 37 (2): 222-239.

Borjas, George. 1985. “Assimilation, Changes in Cohort Quality and Earnings of
Immigrants.” Journal of Labor Economics 3 (4): 463-489.

Borjas, George. 1987. “Self-Selection and the Earnings of Immigrants.” American
Economic Review 77 (4): 531-553.

Borjas, George. 1994. “Long-Run Convergence of Ethnic Skill Differentials: The
Children and Grand Children of the Great Migration.” Industrial and Labor Relations
Review 47 (4): 553-573.

Borjas , George J. 1999. “Immigration and Welfare Magnets.” Journal of Labor
Economics 17 (4) Part 1: 607-637.

Borjas, George. 2003. “Welfare Reform, Labor Supply, and Health Insurance in the
Immigrant Population.” Journal of Human Resources 22: 933-958.

Borjas, George, and Lynette Hilton. 1996. “Immigration and the Welfare State:
Immigrant Participation in Means-Tested Programs.” Quarterly Journal of Economics
111 (2): 575-604.

Butcher, Kristin F., and Anne Morrison Piehl. 1998. “Recent Immigrants: Unexpected
Implications for Crime and Incarceration.” Industrial and Labor Relations Review (51)
4: 654-679.

Butcher, Kristin F. and John DiNardo. 2002. “The Immigrant and Native-Bom Wage
Distributions: Evidence from United States Censuses.” Industrial and Labor Relations
Review 56 (1): 97-.

Bureau of Labor Statistics. Table 1. Employment Status of Civilian Non-Institutional
Population, 1940 to date. Frequently requested information, Department of Labor
accessed at fip://fip.bls. gov/pub/special.requests/lf/aat1 .txt on February 15, 2004.

Carrington, William J., Detragiache, Enrica, and Tara Vishwanath. 1996. “Migration
with Endogenous Moving Costs.” American Economic Review 86 (4): 903-930.

101

Chan, Sewin and Ann Huff Stevens. 2001. “Job Loss and Employment Patterns of Older
Workers.” Journal of Labor Economics 19 (2): 484-521.

Chi, Wei and Brian P. McCall. 2004. “Job Loss and the Receipt of Unemployment
Insurance Benefits of Immigrants.” Paper presented at the 2004 Meetings of the Society
of Labor Economists.

Currie, Janet and Jeffrey Grogger. 2000. “Medicaid Expansions and Welfare
Contractions: Offsetting Effects of Prenatal Care and Infant Health.” NBER Working
Paper 7667.

Fallick, Bruce C. 1996. “A Review of the Recent Literature on Displaced Workers.”
Industrial and Labor Relations Review 50 (1):5-16.

Farber, Henry. 2003. “Job Loss in the United States, 1981-2001.” Princeton University
Industrial Relations Section Working Paper #471.

Farber, Henry S., Haltiwanger, John and Katherine G. Abraham. 1997. “The Changing
Face of Job Loss in the United States, 1981-1995.” Brookings Papers on Economic
Activity, Microeconomics 1997: 55-142.

Friedberg, Rachel M. 2000. “You Can’t Take It With You? Immigrant Assimilation and
the Portability of Human Capital.” Journal of Labor Economics 18 (2): 221-251.

Funkhouser, Edward and Stephen Trejo. 1995. “The Labor Market Skills of Recent
Male Immigrants: Evidence from the Current Population Survey.” Industrial and Labor
Relations Review 48 (4): 792-811.

Green, Gary Paul, Tigges, Leann M. and Daniel Diaz. 1999. “Racial and Ethnic
Differences in Job-Search.” Social Science Quarterly 80(2): 263-277.

Haider, Steven J ., Schoeni, Robert F ., Bao, Yuhua, and Caroline Danielson.
“Immigrants, Welfare Reform and the Economy.” Journal of Policy Analysis and
Management (forthcoming).

Harnmermesh, Daniel S. 1989. “What Do We Know about Worker Displacement in the
United States?” Industrial Relations 28 (1): 51-59.

Hatton, Timothy J. 1997. “The Immigrant Assimilation Puzzle in Late Nineteenth
Century America.” Journal of Economic History 57 (1): 34-62.

Hu, Wei-Yin. 2000. “Immigrant Earnings Assimilation: Estimates From Longitudinal

Data.” American Economic Review 90 (2), Papers and Proceedings of the One Hundred
and Twelfth Annual Meeting of the American Economic Association: 368-372.

102

Immigration and Naturalization Service. “Estimates of the Undocumented Immigrant
Population Residing in the US: October 1996-updated.” Accessed at
http://uscis.gov/graphics/shared/aboutus/statistics/illegalalien/illegal.pdf

Kaestner, Robert and Neeraj Kashaul. 2003. “Welfare Reform and Health Insurance
Coverage of Low-Income Families.” NBER Working Paper #10033.

Katz, Lawrence F. and Murphy, Kevin M. 1992. “Changes in Relative Wages, 1963-
1987: Supply and Demand Factors.” The Quarterly Journal of Economics 107 (1): 35-
78.

Kletzer, Lori G. 1998. “Job Displacement.” Journal of Economic Perspectives 12 (1):
1 15-136.

Kossoudji, Sherrie A., and Deborah A. Cobb-Clark. 2002. “Coming out of the shadows:
Learning about legal status and wages from the legalized population.” Journal of Labor
Economics 20(3): 598-628.

Ku, Leighton and Alyse Freilich. 2001. “Caring for Immigrants: Health Care Safety
Nets in Los Angeles, New York, Miami and Houston.” Urban Institute: Washington
DC.

Lazear, Edward P. 1999. “Culture and Language.” Journal of Political Economy 107
(6) part 2: S95-S126.

Lohr, Steven. “Jobs Lost Abroad: Host of New Causes for an Old Problem.” New York
Times. February 15, 2004.

Mankiw, N. Gregory, Forbes, Kristin J ., Rosen, and Harvey S. Rosen. 2004. “Economic
Report of the President.” Washington DC: US GPO.

Massey, Douglas, Arango, Joaquin, Hugo, Graeme, Kouaouci, Ali, Pellegrino, Adela, and
J. Edward Taylor. 1994. “An Evaluation of International Migration Theory: The North
American Case.” Population and Development Review.20 (4): 699-751.

McManus, Walter S. 1990. “Labor Market Effects of Language Enclaves: Hispanic Men
in the United States.” Journal of Human Resources 25 (2): 228-252.

McManus, Walter, Gould, William and Finis Welch. 1983. “The Role of English
Language Proficiency.” Journal of Labor Economics 1 (2): 101-130.

McMenamin, Krantz and Krolik. 2003. “U.S. Labor Market in 2002: Continued
Weakness.” Monthly Labor Review, February.

NBER's Business Cycle Dating Committee. “US Business Cycle Expansions and
Contractions.” http://www.nber.org/cycles/cyclesmain.html accessed February 15, 2004.

103

Nee, Victor, Sanders, J imy M. and Scott Semau. 1994. “Job Transitions in an Immigrant
Metropolis: Ethnic Boundaries and the Mixed Economy.” American Sociological
Review 59 (6): 849-872.

Reitz, Jeffrey G. and Sklar, Sherrilyn. “Culture, Race and Economic Assimilation of
Immigration.” Sociological Forum. Vol. 12(2), 1997.

Schoeni, Robert. 1998. “Labor Market Assimilation of Immigrant Women.” Industrial
and Labor Relations Review 51 (3): 483-504.

. 1997. “New Evidence on the Economic Progress of Foreign Born Men in the
19708 and 808.” Journal of Human Resources 32 (4): 683-740.

Tienda, Marta and Ding-Tzann Lii. 1987. “Minority Concentration and Earnings
Inequality: Blacks, Hispanics, and Asians Compared.” American Journal of Sociology.

Wilson, Kenneth L. and Portes. 1980. “Immigrant Enclaves: An Analysis of Labor
Market Experiences of Cubans in Miami.” American Journal of Sociology 86 (2): 295-
319.

Wilson, Franklin D. “Ethnic Concentrations and Labor-Market Opportunities.” In Frank
D. Bean and Stephanie Bell-Rose (Eds.) Immigration and Opportunity: Race Ethnicity
and Employment in the United States. Russell Sage Foundation: New York.

Zimmerman, Wendy and Karen Tumlin. 1999. “Patchwork Policies: State Assistance
for Immigrants Under Welfare Reform.” Urban Institute Occasional Paper 24.

104

Appendix A: Weekly Earnings Change by Immigrant Status: 1994-2002.

Table A3.1 Weekly Earnings Growth by Immigrant Status: 1994-2002.

 

Year

1994

1995

1996

1997

1998

1999

2000

2001

2002 -

Average Wage Growth
Cumulative Growth: 1994-2002

 

 

Natives Immigrants
$495.40 - $388.61 -
$609.71 18.75% $532.42 27.01%
$668.11 8.74% $589.56 9.69%
$672.93 0.72% $558.65 ' -5.53%
$682.00 1.33% $584.55 4.43%
$701.55 2.79% $611.96 4.48%
$696.16 -O.77% $599.65 -2.05%
$732.53 4.97% $638.17 6.03%
$758.63 3.44% $646.10 1.23%

4.99% 5.66%
53.14% 66.26%

 

Using a sample limited to men and women ages 20 to 64, who are not in school or

retired and report a desire to work, I calculate average weekly earnings from the CPS

March files.

calculations. Average annual wage growth is slightly higher for immigrants using this
sample, but is not significantly different from natives. Average weekly earnings grew
66% from 1994 to 2000 for immigrants. While natives had a lower rate of growth, 53%,
I must note that this is mainly due to changes in the unemployment rate. If I restrict the
sample to exclude people who worked at least part of the year, immigrants have slightly
lower weekly earnings growth. Nevertheless, the figure which this calculation is

intended to be compared with includes displaced workers who are not re-employed at the

Persons who are unemployed have weekly earnings of $0 in these

survey date. Thus, excluding unemployed persons would be inappropriate.

105

Appendix B: Participation Rates in Government Income Support Programs and
Access to Non-Labor Income

Table A3.2 Proportion of Weighted Sample Receiving
Unemployment Insurance Benefits By Immigrant Status

 

 

 

 

 

 

 

and Sex
Survey Date Employment Status
Male

Employed Not Employed
Natives 0.38 0.53
N 4,958 1,762
Immigrants 0.30 0.49
N 641 254

Female

Employed Not Employed
Natives 0.38 0.48
N 3,895 1,723
Immigrants 0.40 0.50
N 450 273

 

*Sample excludes non-responses

Table A3.2 uses the displaced worker sample in this paper and calculates usage
rates of Unemployment Insurance benefit by immigrant status, gender and whether or not
the worker was employed at the survey date. In addition to the above calculations, I
utilize data in the CPS March files (1996, 1998, 2000, 2002) on income from public
welfare, supplemental security income and unemployment insurance benefits to calculate
participation rates of 20 to 64 year olds by immigrant status and sex. In a linear
regression, controlling for age, sex, education, race, year of survey, weeks worked in the
previous year, non-labor income in household and MSA fixed effects, I calculate relative
usage rates of male immigrants, female immigrants and female natives compared to male
natives in the top 1/2 of the Table. Table A3.3 columns (1) to (3) give coefficients

indicating usage rates of Supplement Security Income, public welfare assistance and

106

unemployment compensation relative to male natives. Column (4) shows relative
participation in at least one of the three programs. Column (5) uses non-labor income (an
independent variable in the first four regressions) as the dependent variable. The second
1/2 of Table A3.3 gives the expected search intensity based on the value of relative income
support available for each column. Program participation (or non-labor income) is
expected to be negatively related to search intensity. So that in Panel A column (1)
where female natives use public assistance in the greatest proportion, they are expectedto
be re-employed at the lowest rates. The sample is restricted to those who want to work

and are not in school.

Table A3.3. Coefficients from Estimates of Government Income Support Use and Access
to Non-Labor Income and Implied Search Intensity Rank.

 

 

 

 

 

 

 

Panel A
(1) (2) (3) (4) (5)
Public Assistance SSI UI Benefits All Non-Labor Income
Male Immigrant -0.007 -0.002 ~0.016 -0.026 4,169.55
Female Native 0.014 0.001 -0.023 —0.009 11,694.16
Female Immigrant 0.008 -0.003 -0.032 -0.027 7,210.33
N 281,168 281,168 281,168 281,168 281,168
0.07 0.01 0.04 0.08 0.14
Panel B
Implied Search Intensity Rank Based on above Use Rates

Male Native 2 3 4 4 2
Male Immigrant 1 2 3 2 1
Female Native 4 4 2 3 4
Female Immigrant 3 l 1 1 3

 

Notes: All estimates are significantly different from male natives and in most cases each other.
The exception is that immigrants do not differ by sex in their SSI usage rates. This drives the
overall usage of Government Income Supports so that there is no difl‘erence between
immigrants by gender.

107

1"."