Globalization, Technology and Inequality
Lecture 4: Trade, O¤shoring and Inequality
Gino Gancia
July 5, 2012
Trends in Wage Inequality
within-country wage inequality has increased in several countries in the past decades
some measures of wage inequality:
1 returns to college education or college premium
2 skill premium = wage of white-collar workers wage of blue collar workers
best-known case:
I US, 1980-2008: college premium +33%
international evidence more scattered due to lack of data:
I sample of 35 countries in the 1980s: skill premium +8% (Epifani &
Gancia, 2008)
I sample of 26 countries 1990-2005: college premium +7% (Parro, 2011)
US Evidence: Skill Premium and Skill Supply
since the late 1970s, in the US both the skill premium and the relative
supply of skilled workers have increased...
Trends in US Real Wages
...and the real wage of unskilled workers has declined/stagnated
Main Explanations for Rising Wage Inequality
wage trends in the US motivated a large literature main explanations:
I Skill Biased Technical Change (SBTC)
I globalization
why globalization?
I timing: the last globalization boom started in the late 1970s
I the correlation between measures of wage inequality and a country openness (import +export
GDP ) is often positive
I Goldberg and Pavcnik (2007):
F trade liberalization in 1980s-90s!rising skill premia in Mexico, Colombia, Argentina, Brazil, Chile, India, Hong Kong
Evidence: Trade and Returns to Education
HND
PER NLD
FIN
GTM S WE
K OR NLD
FIN PHL
ITA ESP
CYP THAVEN
NOR
NOR CHL
KENCAN CHE B RA SLVPOL
NOR AUSAUT
ITA
GRC B OL CRI
I DN CAN DNK
USA AUS GBR BRA P RY
JPN P ANGRC
B RA USA
P RTCOL NLD NLD SWE
AUT ITA PAK
GRC NLD MEX
SWE CHL P RTSWE
CHN
THA KOR
CRI
NLD GHA
PHL
FI N MEX
GTM NI C
-.10.1.2.3Period % Change in the Returns to Education
-.05 0 .05
Period Change in the Openness R atio slope coeff. = 1.23 (t = 2.90)
sample: 40 countries observed between early 60s and late 90s
Evidence: Trade and the Skill Premium
PAN E GY
FIN P ER
KOR TZA
J PN ETH I TA
GRC CAN
GB RGTM
S WE USA B GD
CYP AUS DNK
IND P AK
DEUIRE COL E SP AUT
MEX
VEN
P HL URY
TUR
CHL LUX
FJI MYS
-.4-.20.2.4% Change in the Skill Premia
-.4 -.2 0 .2 .4
Change in the Openness Ratio: 1980-90 slope coeff. = 0.58 (t = 3.96)
sample: 35 countries observed between 1980 and 1990
Evidence: Trade and Income Distribution
TTO MRT
J OR LKA
LSO
CIVLKA HKGTUN
J AM PAN
NLDLUX
JOR IDN VEN
DOM IRN SWE
B GD K OR
FIN FJ I ZMB
ITAGE R DNK P AK P ER PRT KOR ESP
THA CRI TZAGRC
NOR B RA
JPN NOR
PER BEL MUSGBR MEX PAK NZL
J PN BOL EGY VEN
PER
TZA PAK COLNOR DNKI ND JPN
AUS SLV BRA IND ZMB
NOR
ITA GTMGBR
M EX COL
V EN CAN
HNDDOM FRA BRAS LE
NLD MDG J PN I DN THA FIN
BGDP HL
TTO BGD GER
B RA AUS
FRA ETH J PN
ESP IND
SWE NOR
NOR ECU SEN BGD
IND NER
TTO
FIN PAN
PER CHL
LKA MDGIDN IND
FIN COL
CAN GB R ESP
B RA AUS S WE
CANJPN TUR CRIE CU
ITA PAK I RL FRA
GER PRTPRT
DNK GBR CRI
MEX NZL
IDN MLI
PAN GE RTHA DNK
MYSNZL P AK
FRA GA B DZA MEX
CAN CRI CHL
PHL
KOR SWE ZMB
COL GBR CAN
MYS NLD IDN ESP SLV
E SP CRI
K OR GRC B GD EGY TZA
CHL VEN THA TUN K OR
TUR
LKA
FIN
THA NLDNGA
I RLGHA
SWE PHL HND
LKA IDN
J AM DOM
HKG
VEN TUN P HL MEX SYC
HND BELTUN
K ORTHA
HKGMUS J OR
MYS NGA
MYS HKG
SGP
-.1-.050.05.1Period % Change in the Gini Coefficients
-.05 0 .05 .1 .15
Period Change in the Openness R atio slope coeff. = 0.13 (t = 2.40)
sample: 68 countries observed between the early 60s and the late 90s
A Basic Framework
GE model with 2 sectors and 2 factors (high- and low-skill workers) aim:
I study the e¤ects of trade and technology on the skill-premium
CES utility
U = Y = h ( Y
l)
ee1+ ( Y
h)
ee1i
e e 1
, e > 0
I Yh =skill-intensive good
I Yl =low-skill intensive good
I e= elasticity of substitution between Yl and Yh
Demand and Supply
to …nd relative demand, solve:
max
Yh,Yl
h
( Y
l)
ee1+ ( Y
h)
ee1i
e e 1
P
hY
hP
lY
lI Ph, Pl =prices of Yh, Yl
I FOCs:
Y1e (Yh) e1 =Ph ! Yh
Yl = Ph Pl
e
each sector employs one type of labor:
Y
h= A
hH and Y
l= A
lL
I H=supply of skilled workers, productivity Ah, wage wh
I L=supply of low-skill workers, productivity Al, wage wl
The Skill Premium
perfect competition
I price = marginal cost
Ph = wh
Ah and Pl = wl Al
skill premium:
w
hw
l= P
hP
lA
hA
lI use PPh
l = YYh
l
1/e, Yh=AhH and Yl =AlL
wh
wl = Ah Al
e 1
e L
H
1 e
The Skill Premium: Determinants
determinants of the skill premium w
hw
l= A
hA
le 1
e
L
H
1 e
1 technology:
F if goods are gross-substitutes(e>1), skill-biased technical change (higher AAhl) increases wwhl
F if e<1 it is the opposite, but this case is empirically less plausible
2 endowments:
F an increase in the relative supply of one factor reduces its relative reward, stronger e¤ect when e is low
Trade and the Skill Premium
e¤ect of trade:
I similar to a change in endowments
integration between two identical countries:
I equal to doubling H and L
I no change in the skill-premium
integration between a skill-abundant North and a skill-scarce South : H
L > H + H L + L > H
L w
hw
l< w
hw
ltrade
< w
hw
lI trade raises the reward of the relatively abundant factor
N-S Trade and the Skill Premium
North-South trade can explain the increase in the skill premium in the North
yet, this mechanism fails to convince in practice why?
1 most of trade is between similar countries:
F the volume of N-S trade is considered too low (particularly in the 80s and 90s)
2 wage inequality has increased also in many less-developed countries
F see Goldberg and Pavcnik (2007)
Epifani and Gancia (2008)
models of new trade theory with increasing returns (IRS) and monopolistic competition can explain:
I why trade between identical countries may increase wage inequality
I why trade may lead to a pervasive increase in skill premia
same framework as before, but:
1 skilled workers produce di¤erentiated goods with IRS
2 unskilled workers produce homogenous goods
e¤ect of trade: create bigger markets
I di¤erentiated goods are subject to IRS!bene…t more from bigger markets
I skill is more valuable in large global markets
Skill-Intensive Sector
Y is a CES baskets of di¤erentiated varieties:
Y = y
lα+ ∑
ni=1( y
h,i)
α 1/α, α 2 ( 0, 1 )
I n=number of skill-intensive varieties (endogenous)
I e=1/(1 α) >1=elasticity of substitution
demand for any variety y
i:
max
yi[ Y ∑
ni=1p
h,iy
h,ip
ly
l]
I from the FOC, pi =Y1 αyiα 1 :
yi =p
1 α 1
i Y
I negative function of its own price, with elasticity σ=1/(1 α)
Skill-Intensive Firms: Monopolistic Competition
one …rm = one variety, total cost function:
TC = ( F + cy
h,i) w
hI F = …xed cost, c = variable cost, all costs are in units of labor
prices set to solve:
max
ph,iπ
h,i= p
h,iy
h,i[ F + cy
h,i] w
hs.t. : y
h,i= p
1 α 1
h,i
Y
I …rm takes Y as given, FOC:
ph,i =ph = cwh α
I markup over the marginal cost
I simplify: set c=α!ph =wh
Firms’Scale and Number
Free Entry (FE) ! π
h,i= 0:
( p
hcw
h) y
h= w
hF ! y
h= F 1 c
I FE pins down …rm scale
I simplify: set F =1 c!yh =1
labor market clearing (demand = supply):
( F + cy
h) n = H ! n = H
I with a …xed scale, an increase in H is accommodated by an increase in the number of …rms
L-intensive sector:
I single good, perfect competition:
yl =L
Trade and the Skill Premium
skill premium (use p
i= Y
1 αy
iα 1):
w
hw
l= p
hp
l= y
ly
h1 α
= L
1 αI skill premium is increasing in L and independent of H!
I why? an increase in H creates its own demand
e¤ects of trade:
I trade integration is like an increase in H and L
I new e¤ect: a bigger market can sustain a larger number of skill-intensive varieties!similar to SBTC?
I trade integration always increases the skill-premium!
O¤shoring and Wages
we now introduce o¤shoring in a way inspired to:
I Grossman & Rossi-Hansberg (2006, 2008)
I Acemoglu, Gancia & Zilibotti (2012)
focus on o¤shoring of L-jobs (more relevant case)
I production of Yl requires intermediates that can be separated geographically
I bene…t of o¤shoring: move production to low-wage countries (South)
I but only a fraction κ< ¯κ= L+LL of intermediates can be o¤shored
new result:
I o¤shoring of unskilled jobs can, in some cases, bene…t domestic unskilled workers!
Introducing L-O¤shoring
preferences:
Y = h ( Y
l)
ee1+ ( Y
h)
ee1i
e e 1
I Yl is a CES function of a unit set of intermediates
I intermediates[0, κ]are o¤shored to the South
Yl = Z 1
0 yiαdi
1/α
=h(1 κ)1 αLα+κ1 α(L )αi1/α
F we used yi =L /κ and yi=L/(1 κ)
I no o¤shoring in the H sector: Yh =H
wages:
I equal to MPL
wh = ∂Y
∂H =Y1/eYh 1/e wl = ∂Y
∂L =Y1/eYl 1/eYl1 α(1 κ)1 αLα 1
L-O¤shoring and the Skill Premium
w
hw
l= L
1 α
Y
h1/e( 1 κ )
α 1Y
l1 α 1/ee¤ects of κ:
1 direct e¤ect: less demand for L in North!higher skill premium
2 e¢ ciency e¤ect: higher Yl ! (?)
if tasks are su¢ ciently complementary (α < 1 1/e)
I wh
wl is a U function of κ
I why? recall ∂Yl(κ'0)
∂κ !∞, and ∂Yl(κ∂κ'¯κ) !0
intuition:
I with enough complementarity, cost saving on[0, κ]increases the demand for workers on[κ, 1]too!lower skill premium
I but this e¤ect disappears as wl !wl
L-O¤shoring and the Skill Premium
0.0 0.1 0.2 0.3 0.4 0.5
1.0 1.2 1.4 1.6
kappa wh/wl
e = 1.6, solid: σ = 5, dashed: σ = 1.25, σ = 1/ ( 1 α )
Feenstra & Hanson (1997, 1999)
the removal of barriers to Foreign Direct Investment has triggered o¤shoring from US to Mexico
this relocation of economic activity can increase wage inequality in both countries
why? because o¤shored activity are low-skill intensive relative to US production, but skill-intensive relative to Mexican production
thus, the skill-intensity of production (and thus the demand for skill)
increases both in US and Mexico
Alternative Approaches: Trade, Sorting and Inequality
so far, only two skill levels (H and L):
I in reality there are many skill levels and inequality has increased also within skill groups
I with more skill levels sorting (who works with who) becomes important
Trade, Sorting and Inequality:
I trade bene…ts disproportionately more productive …rms, more productive …rms hire more able workers!trade bene…ts relatively more high ability workers
F Yeaple (2005), Ohnshorge and Tre‡er (2007), Costinot and Vogel (2010), Helpman, Itskhoki, and Redding (2010, 2011), Monte (2011)
O¤shoring, Sorting and Inequality:
I globalization makes international teams of workers possible!"good"
workers in the South form teams with workers in the North!"bad"
workers in the South are segregated!their wage falls
F Antras, Garicano & Rossi-Hansberg (2006), Kremer & Maskin (2006)
What Did We Learn?
trade/o¤shoring can a¤ect factor prices (the skill premium) in rich ways
trade (even between similar countries) may lead to a generalized increase in wage inequality
trade/o¤shoring can have e¤ects similar to (biased) changes in productivity
I trade may increase relatively more the productivity of skilled workers
I o¤shoring of labor-intensive tasks may increase the e¢ ciency of unskilled sectors
F but only if tasks are very complementary
I in more advanced models, trade can even trigger SBTC