Content uploaded by Constantin Ogloblin
Author content
All content in this area was uploaded by Constantin Ogloblin on Apr 25, 2014
Content may be subject to copyright.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
5
GLOBAL OUTSOURCING OF HUMAN CAPITAL
AND THE INCIDENCE OF UNEMPLOYMENT
IN THE UNITED STATES
OGLOBLIN, Constantin
*
Abstract
The study is the first to examine empirically the impact of the new
wave of global job outsourcing on skill-specific patterns of
involuntary unemployment in the U.S. using the latest individual-
level data. The estimates from a probit model show that, so far,
global human-capital outsourcing has not shifted the risk of
unemployment from lower-skilled to higher-skilled American
workers. Overall, the probability of involuntary unemployment is
negatively related with the worker’s level of education. For the
outsourceable occupations, however, high-skilled workers are
currently at a greater risk of unemployment than those with lower
skills.
JEL classification: C1, I2, J2, J6
Key words: outsourcing, globalization, labor, unemployment,
education
1. Introduction
Until recently, an established and generally admitted view in
economic literature had been that global economic integration tends
to shift low-skilled, mainly blue-collar jobs from developed to
developing countries while creating more high-skilled, well-paid jobs
in the developed world. The new wave of globalization seems to
challenge this view.
In the United States, the announcement of the new round of
international trade came with a cover story published in February
2003 by Business Week, which described the new phenomenon as a
*
Constantin Ogloblin is researcher at Georgia Southern University, USA
The author would like to thank all participants at the Georgia Southern
University’s Economics and Public Issues Conference in June 2004 for their
comments. This study was supported by a GSU research grant. E-mail:
coglobi@georgiasouthern.edu
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
6
global shift of “upscale jobs” from the U.S. to developing countries
and characterized it as one of the biggest trends reshaping the global
economy (Engardio et al (2003)). Since then, anecdotal evidence of
companies in developed economies shipping high-skilled jobs
offshore has piled up, raising fears of job losses among high-skilled
workers in the West and sparking hot political debates.
Outsourcing of human capital became “a big sleeper issue” at the
annual meeting of the World Economic Forum in January 2004,
where business, government, and academic leaders, usually
supporters of globalization, worried that its new round might not
produce as many well-paid jobs in Europe and the U.S. as once was
expected, which could erode political support for free trade (Davis
(2004), Friedman (2004)). According to The Economist (2004),
offshore job outsourcing is the main reason why anti-trade sentiment,
especially in the U.S., is currently having one of its strongest revivals
in years.
Although the literature on the labor market effects of international
trade is ample, global outsourcing of human capital has been studied
very little. It is a very recent phenomenon brought about by the
latest advances in information technology and communication
around the globe, along with the significant growth of human capital
in developing countries and trade liberalization in the former
communist economies. So far, much of the research has been
conducted or ordered by politically or financially interested groups
and based on private data sources. The reliability of these studies is
hard to verify and their objectivity is doubtful.
The present study is one of the first independent attempts to
examine the influence of the recent wave of global human-capital
outsourcing on the incidence of unemployment in the U.S. Are the
better educated workers currently more likely to be involuntarily
unemployed than are the lower educated workers? Is the incidence
of unemployment across different levels of education related to
outsourcing? Are the fears that the new wave of globalization is
shifting the risk of unemployment from low-skilled to high-skilled
workers justified?
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
7
I attempt to answer these questions with the help of statistical
inference based on the individual-level data from the U.S. Bureau of
the Census’ Current Population Survey (CPS) combined with the
best available information on occupational categories threatened by
offshore outsourcing. I construct a probit equation that estimates the
expected rate of involuntary unemployment conditional on the
worker’s level of education and on whether her occupation is at risk
of outsourcing. This study is largely empirical. A formal theoretical
examination of the issue is on the author’s agenda for further
research.
2. Data
Most empirical studies of the international job outsourcing are
based on industry-level data. Geishecker and Görg (2003) argue that
this approach has certain problems. First, industry-level data do not
capture individual heterogeneity, which may play an important role
in explaining labor-market outcomes. Second, regressing industry
relative unemployment rate on industry-level measures of
outsourcing activity is likely to cause the endogeneity problem. For
these reasons, I use individual, rather than industry, level data.
The main data sets come from the Current Population Survey
(CPS), a monthly survey of about 50,000 households conducted by
the U.S. Bureau of the Census for the Bureau of Labor Statistics.
The survey provides publicly available individual-level data that are
most current of all government labor market statistics.
†
The CPS
datasets are substantially larger in scope and more reliable than any
other source of the U.S. labor-market data. The April basic datasets
for 2000 and 2004 are used. The April 2004 dataset was the latest
available at the time I started this project and in this sense is
randomly selected to reflect the latest labor-market trends. The 2000
set reflects the time when the new form of human-capital outsourcing
†
The CPS data and methodology are available from the CPS Web site at
http://www.bls.census.gov/cps/cpsmain.htm. The computer programs
(STATA “DO” files ) used to process the data and generate econometric
results are available from the author on request.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
8
was virtually nonexistent and thus provides a baseline for
comparisons with 2004. The sample is restricted to individuals in
civilian labor force, aged 20-64, who are either wage employed or
involuntary unemployed. In all descriptive statistics and estimation
procedures the data are weighted using the CPS final weights.
The new type of international job outsourcing is hard to account
for. Direct and systematic data on the number and composition of
jobs migrating from the U.S. offshore are not available. Measuring
outsourcing as the share of imported intermediate inputs in the
industry’s output, as suggested by Feenstra and Hanson (1996,
1999), does not work well for the new type of outsourcing since it
does not account for the direct cross-border transfers of service jobs
and human-capital services, especially within multinational firms.
Further, within the same industry, there are outsourceable jobs and
those that are hard or impossible to outsource. Also, I believe, in the
face of the rising new wave of outsourcing, the labor-market
outcome for a particular workers depends more on the occupation
than on the industry he is associated with.
In this paper, I construct the measure of international job
outsourcing activity using information from Forrester Research Inc.
(McCarthy et al (2002)), one of the most authoritative and widely
cited proprietary reports. The report predicts that a cumulative
830,000 white-collar jobs will be outsourced from the U.S. offshore
by 2005, and by 2015 the number is expected to rise to 3.4 million
(Hilsenrath (2004)). It also contains a breakdown of U.S. jobs
expected to migrate overseas into nine major groups of the Standard
Occupational Classification (SOC) (Kirkegaard (2003)). Without
questioning Forrester’s assumptions, I use these groups to represent
occupations influenced by offshore outsourcing. The nine SOC
groups threatened by offshore outsourcing are listed in Table 1 along
with the groups that are not at risk of outsourcing.
3. Labor market trends
Tables 2, shows the rate of involuntary unemployment in the U.S.
in 2000 and 2004 by level of education. The involuntary
unemployment rate is calculated as the percentage of laid off or
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
9
otherwise involuntarily unemployed workers in the labor force that is
either employed or involuntarily unemployed.
Table 1. Major SOC Categories by Risk of Offshore Outsourcing
a
Code Category Code
Category
Threatened by outsourcing Not threatened by outsourcing
11 Management 21 Community and social
service
13 Business and financial
operations
25 Education, training, and
library
15 Computer and
mathematical
29 Healthcare practitioner
and technical
17 Architecture and
engineering
31 Healthcare support
19 Life, physical, and
social science
33 Protective service
23 Legal 35 Food preparation and
serving related
27 Arts, design, entertain-
ment, sports, and media
37 Building and grounds
cleaning and maintenance
41 Sales and related 39 Personal care and service
occupations
43 Office and
administrative support
45 Farming, fishing, and
forestry
47 Construction and
extraction
49 Installation, maintenance,
and repair
51 Production
53 Transportation and
material moving
55 Armed forces
a
Standard Occupational Classification (SOC) system is currently used by
Federal statistical agencies; it combines all occupations into 23 major groups.
The nine SOC groups threatened by offshore outsourcing are listed in
Kirkegaard (2003). The complete SOC hierarchical structure and occupational
definitions are available from the U.S. Department of Labor at
http://www.bls.gov/soc/.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
10
The average rate of involuntary unemployment in 2004 is markedly
higher than that in 2000, and in both years involuntary
unemployment shows a clear trend to fall monotonically as the level
of education rises. Thus, for the economy as a whole, the new wave
of international outsourcing does not seem to shift the risk of
unemployment from lower-skilled to higher-skilled workers.
Table 2. Involuntary Unemployment Rate in the U.S. by Level of
Education, 2000 and 2004, %
a
Level of education
b
2000 2004
High school not completed 3.72 5.74
High school diploma 2.13 4.21
Some college 1.58 3.03
Associate degree 1.27 2.39
Undergraduate college degree 0.75 2.15
Graduate or professional degree 0.62 1.44
All levels 1.70 3.27
a.
Calculated from the April CPS. Civilian labor force, aged 20-64,
either wage employed or involuntarily unemployed. All cell counts
are weighted using the CPS final weights.
b
Highest degree obtained.
However, as Table 3 shows, the trends in employment and
unemployment differ depending on whether or not the occupation is
in a category threatened by global outsourcing. The total number of
jobs in 2004 increased by 3.2 million compared to 2000, but since
the number of outsourceable jobs decreased by almost 1.2 million,
the additional employment came entirely from the occupations that
are not likely to get outsourced. The percentage increase in the
number of involuntary unemployed workers whose job falls into the
occupational categories threatened by outsourcing is 35.5 points
higher than that for the rest of the labor force. However, the rate of
involuntary unemployment for the outsourceable occupations is
substantially lower in both 2000 and 2004.
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
11
Table 3 Employment and Involuntary Unemployment in the U.S.,
2000 and 2004 (thousands)
a
2000 2004 Change
% change
Occupations threatened by outsourcing
Employment 54,542 53,365 -1,177 -2.16
Involuntary
unemployment 685 1,531 846 123.53
Rate of involuntary
unemployment, % 1.24
2.79
1.55
124.87
Occupations not threatened by outsourcing
Employment 56,416 60,804 4,388 7.78
Involuntary
unemployment 1,237 2,326 1,089 88.03
Rate of involuntary
unemployment, % 2.15
3.68
1.54
71.72
All occupations
Employment 110,958 114,169 3,212 2.89
Involuntary
unemployment 1,922 3,858 1,935 100.68
Rate of involuntary
unemployment, % 1.70
3.27
1.57
91.93
a.
Calculated from the April CPS. Civilian labor force, aged 20-64, either
wage employed or involuntarily unemployed. For unemployed
individuals, the occupational category is defined by the job lost. All cell
counts are weighted using the CPS final weights.
It is also worthy of note that the outsourceable occupations attract
better educated workers compared to the occupations that are not at
risk of outsourcing (Table 4). In 2004, 37.5% of the labor force
pertained to the outsourceable occupational categories had at least a
bachelor’s degree and only 29.2% did not have education above the
high-school level. For the categories that are not likely to be
outsourced, these numbers were 22.4% and 50.8% respectively.
Thus, the fears of potential loses of high-skilled jobs to international
outsourcing appear to be justified.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
12
4. The empirical strategy
I first estimate a probit model that tests whether the rise of global
human-capital outsourcing is shifting the risk of unemployment from
the low-skilled to the high-skilled labor force and whether the rate of
involuntary unemployment is significantly higher for the
outsourceable compared to non-outsourceable occupations.
Table 4 Distribution of the U.S. Labor Force by Level of Education,
2000 and 2004, %
a
Category by risk of outsourcing
At risk Not at risk
2000 2004 2000 2004
High school not completed 3.4 3.6 17.3 16.0
High school diploma 26.4 25.7 37.8 34.8
Some college 23.6 23.2 17.4 17.3
Associate degree 9.1 9.3 8.4 9.4
Undergraduate college degree
26.6 27.6 12.4 13.8
Graduate/professional degree 10.9 10.6 6.6 8.6
Total 100.0 100.0 100.0 100.0
a.
Calculated from the April CPS. Civilian labor force, aged 20-64. All cell
counts are weighted using the CPS final weights.
The model may be written as follows:
P(U
i
= 1) = Φ(α’E
i
+ βR
i
+ γ’C
i
) (1)
where U
i
= 1 if individual i is involuntarily unemployed and zero
otherwise, F(⋅) is the cumulative probability function for the
standard normal variable;α, β, and γ are probit parameters, E
i
is a set
of dummy variables representing individual i’s level of education
with a high-school diploma as the baseline, R
i
= 1 if individual i’s
occupation is outsourceable and zero otherwise, and C
i
is the vector
of variables that control for individual heterogeneity (age, gender,
race/ethnicity, and marital status) and structural factors (industry,
region, and metropolitan status). All variables included in (1) are
defined and described in Table 5.
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
13
Table 5 Variable Definitions and Descriptions
a
Mean (St. error)
b
Variable Definition
2000 2004
Dependent variable:
unlost 1 if involuntarily
unemployed
1.70
(0.06)
3.27
(0.09)
Education dummy variables (highest level attained):
Baseline High school diploma 32.21
(0.23)
30.50
(0.23)
nohischl High school is not
completed
10.34
(0.15)
10.01
(0.15)
smcolge Attended college but no
degree
20.39
(0.20)
20.03
(0.20)
associate Associate degree 8.79
(0.14)
9.40
(0.14)
bachelor Bachelor’s degree 19.45
(0.20)
20.41
(0.20)
grad/prof Graduate or professional
degree
8.82
(0.14)
9.65
(0.14)
ocouts 1 if the occupation is
outsourceable
48.93
(0.25)
46.51
(0.25)
age Age 39.02
(0.06)
39.76
(0.06)
age_sq age
2
/100 16.47
(0.04)
17.15
(0.05)
gender 1 if female 47.85
(0.25)
47.69
(0.25)
marsta 1 if married 59.30
(0.25)
58.59
(0.24)
Race/ethnicity dummy variables:
Baseline White 72.00
(0.23)
68.78
(0.24)
hisp Hispanic origin 10.75
(0.16)
12.76
(0.18)
black Black 12.46
(0.17)
11.86
(0.17)
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
14
Table 5 Continued
Mean(St. error)
b
Variable Definition
2000 2004
asian Asian 3.93
(0.10)
4.36
(0.11)
other Other/mixed 0.86
(0.04)
2.24
(0.07)
Industry dummy variables:
Baseline Manufacturing, mining 17.51
(0.19)
14.04
(0.17)
agric Agriculture, forestry,
fishing, hunting
1.62
(0.06)
0.89
(0.05)
constr Construction 5.96
(0.12)
6.86
(0.13)
trade Wholesale and retail
trade
18.46
(0.19)
14.14
(0.17)
tranutl Transportation and
utilities
6.12
(0.12)
5.30
(0.11)
infnbus Professional, and
financial services.
19.38
(0.20)
19.03
(0.19)
eduhlth Education and health 18.87
(0.19)
22.64
(0.20)
othsrv Other services 6.94
(0.13)
12.00
(0.16)
pubadm Public administration 5.15
(0.11)
5.10
(0.11)
Metropolitan status dummy variables:
Baseline Metropolitan, MSA is at
least 1 million
56.17
(0.25)
56.09
(0.24)
metr<1m Metropolitan, MSA is
less than 1 million
23.65
(0.21)
24.29
(0.21)
nonmetr Nonmetropolitan 17.14
(0.18)
16.74
(0.18)
nident Not identified 3.03
(0.08)
2.87
(0.07)
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
15
Table 5 Continued
Mean (St. error)
b
Variable Definition
2000 2004
Region dummy variables:
Baseline Midwest 23.86
(0.21)
23.37
(0.20)
Northeast Northeast 19.02
(0.18)
18.85
(0.18)
South South 34.92
(0.24)
35.10
(0.24)
West West 22.20
(0.21)
22.68
(0.21)
a.
Civilian labor force, aged 20-64. Weighted using the CPS final eights.
b
For dummy variables means and standard errors are in percent.
Next, I obtain the estimates of skill-specific effects of outsourcing
on the rate of involuntary unemployment from the following
equation:
P(U
i
= 1) = Φ (α’E
i
+ βR
i
+ δ’R
i
E
i
+ γ’C
i
) (2)
The differential (δ - α) shows the directions of these effects. If for
the level of education j the difference (δ
i
- α
j
) is significantly
positive, then the incidence of involuntary unemployment pertained
to the outsourceable occupations is significantly greater for this level
of education than for the baseline level.
Equations (1) and (2) are estimated separately for 2000 and 2004
by maximum likelihood, with observations weighted using the CPS
final weights and robust estimates of standard errors. For continuous
variables, the marginal effects are calculated for an infinitesimal
change at the means of all explanatory variables in the equation. For
dummy variables the marginal effects are computed for the change of
the relevant variable from 0 to 1 at the means of the rest of the
explanatory variables. The 2004 results are interpreted as a
“snapshot” of the impact of the rising new wave of global
outsourcing on involuntary unemployment in the U.S. The 2000
results serve as a benchmark.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
16
5. Results
The results are presented in Tables 6 and 7. The patterns of
average effects of education on the probability of unemployment
(Table 6) are virtually the same in 2000 and 2004 and consistent with
the trend revealed by the descriptive statistics in Table 2: the risk of
involuntary unemployment falls as the level of education rises.
Table 6. Probit Equation Estimates: Average Effects of Education,
2000 and 2004
a
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
nohischl 0.176***
(0.049)
0.670 0.121***
(0.042)
0.794
smcolge -0.089**
(0.045)
-0.272 -0.102***
(0.037)
-0.576
associate -0.127**
(0.063)
-0.366 -0.167***
(0.050)
-0.879
bachelor -0.271***
(0.056)
-0.735 -0.169***
(0.040)
-0.920
grad/prof -0.234***
(0.085)
-0.614 -0.235***
(0.061)
-1.175
age 0.008
(0.010)
0.027 0.010
(0.008)
0.062
age_sq -0.008
(0.013)
-0.026 -0.009
(0.010)
-0.056
ocouts -0.079**
(0.038)
-0.255 -0.043
(0.032)
-0.257
gender -0.018
(0.036)
-0.059 -0.072**
(0.029)
-0.430
marsta -0.258***
(0.035)
-0.890 -0.282***
(0.028)
-1.785
hisp 0.009
(0.054)
0.029 -0.080*
(0.045)
-0.451
black 0.278***
(0.046)
1.141 0.289***
(0.039)
2.151
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
17
Table 6. Continued
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
asian -0.151
(0.103)
-0.420 -0.034
(0.073)
-0.197
other 0.290***
(0.112)
1.293 0.087
(0.075)
0.565
agric 0.331***
(0.093)
1.536 0.081
(0.117)
0.524
constr 0.240***
(0.057)
0.986 0.256***
(0.047)
1.900
trade -0.090*
(0.051)
-0.272 -0.122***
(0.046)
-0.671
tranutl -0.217***
(0.078)
-0.570 -0.312***
(0.066)
-1.429
infnbus -0.007
(0.052)
-0.024 -0.092**
(0.044)
-0.523
eduhlth -0.473***
(0.070)
-1.137 -0.508***
(0.052)
-2.382
othsrv -0.137*
(0.071)
-0.390 -0.166***
(0.048)
-0.882
pubadm -0.414***
(0.103)
-0.899 -0.592***
(0.086)
-2.158
metr<1m 0.046
(0.041)
0.151 0.058*
(0.032)
0.356
nonmetr 0.138***
(0.045)
0.494 -0.017
(0.036)
-0.103
nident 0.150*
(0.090)
0.568 0.095
(0.075)
0.624
Northeast 0.036
(0.047)
0.118 0.023
(0.037)
0.137
South -0.109**
(0.045)
-0.341 -0.167***
(0.035)
-0.958
West 0.100**
(0.046)
0.346 0.048
(0.037)
0.296
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
18
Table 6. Continued
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
constant -2.084***
(0.200)
1.250
d
-1.693***
(0.160)
2.576
d
N 49,905 55,839
Wald X
2
(28) 393.28*** 603.10***
Pseudo R
2
0.066 0.062
a
The dependent variable is unlost; observations are weighted using the
CPS final weights.
b
Standard errors are in parentheses.
c
In percent;
calculated for an infinitesimal change in a continuous variable and for the
change from 0 to 1 in a dummy variable at the means of (the rest of) the
explanatory variables.
d
The probability estimate at the means of the
explanatory variables. * Statistically significant at the 0.1 level; ** at the
0.05 level; *** at the 0.01 level.
For instance, in 2004, a worker with a bachelor’s degree is 0.9%
less likely to be involuntarily unemployed than one with a high-
school diploma, and the latter is 0.8% less likely to be unemployed
than an individual who has not finished high school. Both
differentials are statistically significant. That is, so far, the new wave
of global outsourcing has not shifted the risk of unemployment from
low-skilled to high-skilled workers in general.
Further, the results in Table 6 show that outsourcing has no
significant influence on overall involuntary unemployment. The
coefficient of ocouts is negative in 2000 and not statistically different
from zero in 2004. This suggests that the higher unemployment rate
in 2004 compared to 2000 is not related to the new wave of global
outsourcing but is due largely to the business cycle and, possibly,
structural changes in the economy. The latter, however, are hardly
evident, judging by the coefficients of the industry variables reported
in the table.
The probit estimates from equation (2) presented in Table 7
indicate, however, that in 2004 the negative relation between the
level of education and the probability of involuntary unemployment
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
19
holds only for the occupational categories that are not threatened by
global outsourcing. For these occupations the relation (shown by the
“main effects” of the education variables in Table 7) appears even
more pronounced than on the average for all jobs. But, as the (d - a)
differentials reported in Table 7 suggest, for the outsourceable
occupations, more highly educated workers are at a greater risk of
unemployment than those with lower education.
Table 7.Probit Equation Estimates: Occupation-Specific
Effects of Education, 2000 and 2004
a
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
nohischl 0.176***
(0.054)
0.668 0.100**
(0.047)
0.642
smcolge -0.093
(0.061)
-0.284 -0.045
(0.048)
-0.261
associate -0.120
(0.086)
-0.348 -0.134**
(0.064)
-0.714
bachelor -0.214**
(0.088)
-0.602 -0.308***
(0.069)
-1.543
grad/prof -0.115
(0.139)
-0.336 -0.426***
(0.104)
-1.826
age 0.008
(0.010)
0.027 0.010
(0.008)
0.060
age_sq -0.008
(0.013)
-0.026 -0.009
(0.010)
-0.053
ocouts -0.062
(0.055)
-0.201 -0.065
(0.049)
-0.386
ocouts ×
nohischl
0.025
(0.128)
0.082 0.106
(0.101)
0.692
ocouts ×
smcolge
0.003
(0.090)
0.011 -0.111
(0.074)
-0.605
ocouts ×
associate
-0.016
(0.125)
-0.051 -0.074
(0.100)
-0.411
ocouts ×
bachelor
-0.088
(0.115)
-0.265 0.201**
(0.088)
1.377
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
20
Table 7. Continued
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
ocouts ×
grad/prof
-0.184
(0.174)
-0.496 0.286**
(0.129)
2.182
gender -0.022
(0.036)
-0.072 -0.063**
(0.030)
-0.374
marsta -0.258***
(0.035)
-0.890 -0.283***
(0.028)
-1.771
hisp 0.009
(0.054)
0.030 -0.075*
(0.045)
-0.422
black 0.278***
(0.046)
1.145 0.286***
(0.039)
2.102
asian -0.149
(0.103)
-0.415 -0.035
(0.073)
-0.202
other 0.289***
(0.112)
1.287 0.090
(0.075)
0.581
agric 0.330***
(0.093)
1.529 0.087
(0.117)
0.560
constr 0.239***
(0.057)
0.980 0.259***
(0.047)
1.906
trade -0.095*
(0.051)
-0.288 -0.112**
(0.047)
-0.616
tranutl -0.221***
(0.078)
-0.578 -0.307***
(0.066)
-1.396
infnbus -0.009
(0.052)
-0.030 -0.091**
(0.044)
-0.510
eduhlth -0.493***
(0.072)
-1.170 -0.479***
(0.054)
-2.249
othsrv -0.138*
(0.071)
-0.391 -0.165***
(0.048)
-0.868
pubadm -0.416***
(0.103)
-0.901 -0.587***
(0.086)
-2.120
metr<1m 0.046
(0.041)
0.153 0.058*
(0.032)
0.352
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
21
Table 7. Continued
2000 2004
Variable
Coeffi-
cient
b
Marginal
effect
c
Coeffi-
cient
b
Marginal
effect
c
nonmetr 0.138***
(0.045)
0.494 -0.018
(0.036)
-0.104
nident 0.149*
(0.090)
0.564 0.098
(0.075)
0.637
Northeast 0.035
(0.047)
0.117 0.021
(0.037)
0.129
South -0.110**
(0.045)
-0.343 -0.166***
(0.035)
-0.942
West 0.100**
(0.046)
0.344 0.050
(0.037)
0.303
constant -2.086***
(0.201)
1.249
d
-1.694***
(0.161)
2.543
d
Parameter differentials (δ- α):
E
i
=nohisch -0.151
(0.158)
-0.586 0.006
(0.129)
0.050
E
i
=smcolge 0.097
(0.139)
0.295 -0.066
(0.111)
-0.344
E
i
=associate 0.104
(0.194)
0.297 0.060
(0.150)
0.302
E
i
=bachelor 0.126
(0.192)
0.336 0.509***
(0.148)
2.919
E
i
=grad/prof -0.069
(0.297)
-0.161 0.712***
(0.221)
4.008
N 49,905 55,839
Wald X
2
(33) 400.55*** 635.51***
Pseudo R
2
0.067 0.063
a
The dependent variable is unlost; observations are weighted using the CPS final
weights.
b
Standard errors are in parentheses.
c
In percent; calculated for an infinitesimal change in a continuous variable and for
the change from 0 to 1 in a dummy variable at the means of (the rest of) the
explanatory variables.
d
The probability estimate at the means of the explanatory variables.
* Statistically significant at the 0.1 level; ** at the 0.05 level; *** at the 0.01 level.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
22
Compared to a worker who has only a high-school diploma, the
probability of involuntary unemployment is 2.9% higher for a holder
of a bachelor’s degree and 4.0% higher for one with a graduate or
professional degree! In 2000, as was expected, none of the
interaction term coefficients and (d - a) differentials is statistically
significant, i.e. the outsourceable occupation factor does not alter the
overall negative relation between education and unemployment.
Finally, the coefficients of the control variables in equations (1)
and (2) are largely as expected and may provide some useful insights
into the current patterns of involuntary unemployment in the U.S.
Since these results are not the focus of the present study, I do not
discuss them here. Their interpretation is fairly straightforward, and
I simply refer an interested reader to Tables 6 and 7.
6. Conclusions
The recent wave of outsourcing of human capital from the U.S. to
the developing and transition economies has become a big economic
and political issue. The main concern is that workers in the U.S. are
going to lose not just low-skilled jobs, but also high-skilled, well
paid jobs to the rest of the world.
This paper has presented a comparative static empirical study of
the influence of the recent wave of global human-capital outsourcing
on the incidence of unemployment in the U.S. based on the latest
individual-level data from the Current Population Survey augmented
with information on outsourceable jobs from a private source.
The analysis has shown that so far, global outsourcing has not
shifted the risk of unemployment from lower-skilled to higher-skilled
workers in general. In both 2000, when the new form of human-
capital outsourcing was virtually nonexistent and 2004, when it is on
the rise, the probability of involuntary unemployment is negatively
associated with the worker’s level of education. Currently,
belonging in an outsourceable occupational category does not
increase the risk of unemployment for an average U.S. worker,
Oglobin, C. Global Outsourcing of Human Capital and Unemployment in the U.S.
23
which suggests that the higher unemployment rate in 2004, compared
to 2000, is due largely to the business cycle.
At the same time, the negative relation between the level of
education and the probability of involuntary unemployment does not
hold specifically for the outsourceable occupations. On the contrary,
more highly educated workers in these occupations are currently
significantly more likely to be involuntarily unemployed than
workers with lower education. It remains to be seen, however,
whether or not this relation continues to hold and becomes prevalent
for the entire economy as the new wave of global outsourcing gets
more strength.
References
Davis, Bob. “Migration of Skilled Jobs Abroad Unsettles Global-
Economy Fans.” The Wall Street Journal Online, January 26, 2004.
Economist, The. “The new jobs migration.” February 19, 2004.
Engardio, Pete et al. “The New Global Job Shift.” Business Week,
February 3, 2003.
Feenstra, Robert C. and Gordon H. Hanson. “Globalization,
outsourcing, and wage inequality.” The American Economic Review,
May 1996; 86, 2, p. 240-45.
Feenstra, Robert C. and Gordon H. Hanson. “The Impact of
Outsourcing and High-Technology Capital on Wages: Estimates for
the United States, 1979-1990.” The Quarterly Journal of Economics,
114(3), August 1999, p. 907-40
Friedman, Alan. “Embrace Outsourcing, Don't Fear It.” The Wall
Street Journal Online, January 26, 2004.
Geishecker, Ingo and Holger Görg. “Winners and Losers:
Fragmentation, Trade and Wages Revisited.” GEP Research Paper
03/41.
Applied Econometrics and International Development. AEID. Vol. 4-3 (2004)
24
Hilsenrath, Jon E. “Forrester Revises Loss Estimates to Overseas
Jobs.” The Wall Street Journal, May 17, 2004, p. A8.
Kirkegaard, Jacob F. “Outsourcing—Stains on the White Collar?”
Institute for International Economics. February 2004.
McCarthy, John C. et al. “3.3 Million US Services Jobs to Go
Offshore.” Forrester Research, Inc., November 2002.
McCarthy, John C. et al. “Near-Term Growth of Offshoring
Accelerating Resizing US Services Jobs Going Offshore.” Forrester
Research, Inc., May 2004.
________________________
Journal published by the Euro-American Association of Economic
Development. http://www.usc.es/economet/eaa.htm
