Why Do New Technologies Complement Skills

Why Do New Technologies Complement Skills?
Directed Technical Change and Wage Inequality

Author(s): Daron Acemoglu
Source: The Quarterly Journal of Economics, Vol. 113, No. 4 (Nov., 1998), pp. 1055-1089
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2586974 .
Accessed: 18/06/2014 15:17
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to The Quarterly
Journal of Economics.
http://www.jstor.org
This content downloaded from 188.72.126.118 on Wed, 18 Jun 2014 15:17:48 PM

All use subject to JSTOR Terms and Conditions
WHY DO NEW TECHNOLOGIES COMPLEMENT

DIRECTED TECHNICAL CHANGE
AND WAGE INEQUALITY*
SKILLS?
DARONACEMOGLU
A high proportionof skilled workersin the labor forceimplies a large market
size forskill-complementarytechnologies,and encourages fasterupgrading ofthe
productivityof skilled workers. As a result, an increase in the supply of skills
reduces the skill premium in the short run, but then it induces skill-biased
technical change and increases the skill premium,possibly even above its initial
value. This theory suggests that the rapid increase in the proportionof college
graduates in the United States labor forcein the 1970s may have been a causal
factorin both the decline in the college premium during the 1970s and the large
increase in inequality duringthe 1980s.
I. INTRODUCTION
In the 1970s college graduates earned 55 percent more than

high school graduates. This premium fell to 41 percent in 1980,
but then increased to 62 percent in 1995 [Autor, Katz, and
Krueger 1998]. One explanation for the rapid increase in the
college premiumin the 1980s is skill-biased technologicalchange.
According to this explanation, new technologies are by their
nature complementaryto skills, so there has always been some
skill-biased technical change, and the recentpast witnessed rapid
introductionof new technologies, leading to an acceleration in
skill-bias.1Empirical support forthis view includes Autor,Katz,
and Krueger [1998], who calculate that the relative supply of
college equivalent workers (college graduates plus half of those
with some college) increased by 2.73 percent per year between
1940 and 1970, and increased much faster between 1970 and
1995, by 3.66 percent per year. In contrast,the college premium
* I have benefitedfrommany insightfulconversations with Jaume Ventura
during the gestation of this project and frommany of his comments on earlier
versions of this paper. I also thank two anonymous referees, Joshua Angrist,
Olivier Blanchard, Ricardo Caballero, Francesco Caselli, Gilles Durante, Oded
Galor, Lawrence Katz, Michael Kremer,Kevin M. Murphy,Dani Rodrik,Fabrizio
Zilibotti,and various seminar participants foruseful commentsand suggestions.
Financial support fromthe National Science Foundation Grant SBR-9602116 is
gratefullyacknowledged.
1. For example, Bound and Johnson [1992], Berman, Bound, and Griliches
[1994], Caselli [1997], Galor and Tsiddon [1997], and Krusell, Ohanian, Rios-Rull,
and Violante [1997]. See Griliches [1956], Bartel and Lichtenberg [1987], and
Goldin and Katz [1998] formicro estimates of technology-skillcomplementarity,
and Krueger [1993] on the impact ofcomputerson the structureofwages.
? 1998 by the President and Fellows of Harvard College and the Massachusetts Institute of
Technology.
The QuarterlyJournal ofEconomics, November 1998
1055

1056
QUARTERLYJOURNALOF ECONOMICS
fellby 0.63 percentper year duringthe firstperiod and increased

by 0.92 percent per year between 1970 and 1995, indicating a
much more rapid increase in the demand for college graduates
duringthe last 25 years.
The skill-biased change explanation leads to a number of
questions, however. Why did skill-biased change accelerate soon
afteran unprecedentedincrease in the supply ofskills duringthe
1970s? More striking,Katz and Murphy write, "forthe 1963-87
period as a whole and most stronglyforthe 1980s, the groups with
the largest increases in relative supplies tended to have the
largest increases in relative wages" [1992, p. 52]. Whyis this?And
related to these questions, why do new technologies complement
skills? There are in fact many examples in the eighteenth and
early nineteenth centuries of new technologies replacing rather
than complementingskills, such as the spinning jenny, weaving
machines, Jacquard's loom, printing cylinders, and later the
assembly line (see Mokyr [1990]). Current technologies also are
not skill-complementaryby nature. Computers, for example,
simplifysome formerlycomplex tasks such as inventorycontrol,
and can be used byunskilled workers,as theyoftenare in fastfood
restaurants and supermarkets.
Motivated by this reasoning, this paper starts from the
premise that new technologiesare not complementaryto skills by
nature, but by design. I show that a natural model in which the
directionof technical change is endogenous can explain why the
demand for skills and the college premium firstfell and then
increased sharply followingthe large increase in the supply of
skills, and also whyas opposed to the skill-replacingtechnological
advances ofthe eighteenthcentury,today most new technologies
appear to be skill-complementary.
Most technologies,once invented,are largely nonrival goods.
They can be used by many firmsand workersat low marginal cost.
When there are more skilled workers, the market for skillcomplementarytechnologies is larger. The inventor is therefore
able to obtain higherprofits,and more effortwill be devoted to the
invention of skill-complementarytechnologies.2As a result, the
impact ofan increase in the supply ofskills on the skill premiumis
determinedby two competingforces:the firstis the conventional
2. Economic motives do not play a directrole in all inventions."Macroinventions," to use Mokyr's [1990] term, are likely to be exogenous and stem from
advances in basic science. For the thesis in this paper, it is sufficientthat economic
motivesinfluencethe directionin which these macroinventionsare developed.

1057
TECHNICAL CHANGEAND WAGEINEQUALITY
substitution effect which makes the economy move along a

downward sloping relative demand curve. The second is the
directedtechnologyeffect,which shiftsthe relative demand curve
forskills as shown in Figure I, because the increase in the supply
ofskills induces fasterupgradingofskill-complementary
technologies.
A large increase in the supply of college graduates as in the
late 1960s and 1970s firstmoves the economyalong a short-run
(constant technology)relative demand curve,reducingthe college
premium.The relative supply change also increases the size ofthe
market fortechnologies complementaryto skills, and induces a
change in the direction of technical progress and a shift of the
relative demand curve in Figure I. Suppose firstthat the substitution effectdominates the directed technologyeffect.In this case,
the college premiumfirstfalls and then increases, but never above
its initial level. In contrast, if the directed technologyeffectis
sufficientlystrong,the model predicts that in the long run the
college premiumshould increase. This is the case drawn in Figure
I and offersa more complete explanation forthe changes in the
U. S. college premiumover the past 25 years. I will also show how
this mechanism can account for a puzzling aspect of the recent
changes in the structureof wages: the increase in residual wage
Relative
CollegePremium Short-run
Demand
\
Long-run
......
CollegePremium
Initial
CollegePremium
Short-run
Response
\
..........
RelativeDemand
Long-run
For Skills
I
I
+
I
I
I
A ShiftinRelativeDemand
duetoDirectedTechnical
Change
H/L
0
Shiftin RelativeSupply
FIGURE I
DirectedTechnicalChangeand DynamicsofCollegePremium

1058
inequality during the 1970s while the college premium was

falling. The analysis in this paper thereforesuggests that the
unprecedentedincrease in the supply ofcollege graduates during
the 1970s may have been causal both for the technological
developments and the changes in the structure of wages of the
past two decades. Finally, since the proportionof skilled workers
has increased substantially over time,this theoryalso suggests a
natural reason why new technologies should be more skillcomplementarytoday than two centuries ago, and accounts for
the steady increase in the demand for skills in the face of the
rapidlyincreasing supply ofskills over the past century.
There are other episodes in which a large increase in the
supply of skills appears to have affectedthe directionoftechnical
change. High school enrollmentand graduation rates doubled in
the 1910s, mostlydue to changes in the location and curricula of
schools and the decline in transportcosts [Goldin and Katz 1995].
The skill premium(white-collarwage relative to blue-collarwage)
fell sharply in the 1910s. Yet, despite the even faster increase in
the supply of high school skills during the 1920s, the skill
premiumleveled offand started a mild increase. Goldin and Katz
[1995] conclude that the demand forhigh school graduates must
have expanded sharply startingin the 1920s, presumably due to
changes in officetechnologyand higher demand fromnew industries such as electrical machinery,transport,and chemicals.
Thought experiments with exogenous variation in skills
illustrate the main ideas, but in the long run the supply of skills
responds to changes in the skill premium.The model developed in
this paper allows me to simultaneously endogenize the demand
for and the supply of skills. When the directed technologyeffect
dominates, an increase in the supply of college graduates again
firstdepresses and then increases the college premium.But now it
also encourages more workersto enroll in college, and induces an
extended period of adjustment where the supply of skills and the
productivity of skill-complementarytechnologies increase together.
An analysis ofthe implicationsofinternationaltrade on wage
inequality provides another application of this theory.The key
observationis that trade affectsthe directionoftechnical change.
If the United States starts trading with the Less Developed
Countries (LDCs) and sells technologiesto LDC firms,the size of
the market for technologies complementary to unskilled labor
increases and wage inequality declines, or at most increases by

1059
only a small amount. However, if due to lack of international

propertyrights protection,it is not possible to sell new technologies to LDC firms,trade simplyincreases the relative price ofthe
skill-intensivegoods, inducing furthereffortin upgrading skillcomplementarytechnologies.I show that in this case conventional
calculations underestimatethe impact oftrade on wage inequality
because they ignore the change in the direction of technical
change induced by trade.
This paper is related to the older literature on induced
innovations, including theoretical work by Kennedy [1964], Ahmad [1966], and Samuelson [1965], empirical studies by Schmookler [1966] and Hayami and Ruttan [1970], and historical work by
Habakkuk [1962] and David [1975]. These studies discuss the
impact of factor prices on innovations. I treat factor prices as
endogenous and point out the importance of market size. The
market size effect-the fact that an increase in the number of
skilled workers increases the size of the market for skillcomplementary technologies-is crucial in deriving the main
result of the paper, which is that a larger relative supply of a
factorcan lead to fasterupgrading oftechnologiescomplementary
to this factor.Previous papers would predict the opposite result
because they do not featurethe market size effect.My paper also
builds on and extends the work of Romer [1990], Aghion and
Howitt [1992], and Grossman and Helpman [1991] on endogenous
technical change by including two types of workers, each using
differenttechnologies and allowing technological change to be
directed. Finally, a number of recent papers also suggest that
changes in the supply of skills may change the demand forskills.
In Acemoglu [1996], when there is a sufficientfractionofworkers
who are skilled, firmsfindit profitableto create jobs specifically
targeted forthis group, and as a result, unskilled wages fall, and
skilled wages increase. Krugman [1997] has recentlyconstructed
a signaling model with some commonfeatures.Independent work
by Kiley [1997] considers an expanding varieties model and shows
that an increase in the supply of skills can create skill-biased
technical change and increase inequality.Walde [1997] compares
the technologychoice of economies differingwith regard to the
skill level of their high school graduates. He shows that an
economy with less skilled high school graduates may choose a
technologywhich makes little use ofhigh school skills and have a
high skill premium.
The plan of the paper is as follows. Section II analyzes the

1060
basic model and contains the most importantresults ofthe paper.

Section III shows how the model can account forthe increase in
residual wage inequality during the 1970s while the college
premium declined. Section IV endogenizes the relative supply of
skills. Section V discusses the impact ofinternationaltrade on the
direction of technical change and wage inequality. Section VI
concludes.
II. THE BASIC MODEL
I firstoutline the standard part ofthe model and explain the

main results informally.The formal model follows Aghion and
Howitt [1992] and Grossman and Helpman [1991], but allows the
pace of technological improvementsin skill-complementaryand
labor-complementarymachines to differ.
A. Preliminaries and Outline
H skilled workers and L unskilled workers supply labor
inelastically and have identical preferences over the unique
consumptiongoody:
(1)
exp (-r(T - t ))ck(v) dT,
Uk(t)
where Ck(T) is the consumptionof agent k at time v and r is the

discountrate, and due to linear utility,it is also the interestrate. I
will drop the time argumentwhen this causes no confusion.
The consumptiongood is produced fromtwo complementary
intermediategoods,or productionprocesses, one using skilled and
the other unskilled labor. The market forintermediate goods is
competitive.I denote the total output ofthese intermediategoods
by Y1 and Yh, and the aggregate production of the consumption
good is
Y = [YI + 'yY]11P,
(2)
where p ? 1, so the elasticityof substitutionbetween Y1and Yh is

1/(1 - p). I normalize the price ofthe finalgood in each period to 1,
and denote the prices of the two intermediategoods bypi and Ph.
Competitive pricing gives a standard relative demand equation
forintermediategoods:
(3)

TECHNICAL CHANGE AND WAGE INEQUALITY
1061
There are ml and mh firmsin the two intermediate goods

sectors, and fornow, I normalize ml = mh = 1. Since later there
will be constant returns to variable factors,the number of firms
does not matter. The production of Yh, the skill-intensivegood,
requires skilled labor while the productionof the labor-intensive
good, Y1,requires unskilled labor. Namely, firmi in sector s has
productionfunction,
(4)
ys(i)= A,Win,
(i I,
where s = 1,h, and nj(i) is the number of workers employed by

firmi in sectors, P < 1, and A,(i) is the productivityoflabor in this
firm.The labor market is competitiveand clears at everyinstant.
Since firmsin sector1onlyemployunskilled workers,and those in
sector h only hire skilled workers,market clearing implies that
Nh = H. The profitsof these
fnl(i) di N1 = L and fnh(i) d
firms,ifany,are redistributedto consumers.
Firm level productivity,
Aj(i), is determinedby the technologies employedby the firm,and skilled and unskilled workersuse
differenttechnologies. Leaving a detailed discussion to the next
subsection, fornow note that all firmsin a sector face the same
(strictlyconcave) problem,so in equilibriumA,(i) = As, and wages
in termsofthe finalgood are w, = 3pAN (-P) fors = 1,h.
The skill premium-skilled wage Wh,relative to unskilled
wage wl-is the main focus of this paper. Using (3), this skill
premiumwiS3
(5)
CO_ Wh
'y(
h)pt-(-)
The skill premium increases when skilled workers become

more scarce; i.e.,
aHIL
This is the usual substitution effect,and shows that for given
technology,the relative demand curve for skill is downward
3. In this section,forsome parameter values, skilled workersmay have lower
wages than the unskilled, i.e., X c 1. One may want to impose y > (H/L)1-(-P)P to
avoid this, or alternatively,one could assume that skilled workers can use the
machines designed for the unskilled and be more productive at this than the
unskilled. When the supply of skills is endogenized in Section IV, the skill
premiumis always positive,so this parameter restrictionis not necessary.

1062
slopingwith elasticity - 1/(1 - U3p).

Moreover,when p E (0,1],
aAh/Al 0;
that is, improvements in the skill-complementarytechnology

increase the skill-premium.The converse is obtained when p < 0.
The conventional wisdom is that the skill premium increases
when skilled workersbecome more,not less, productive,which is
consistent with p > 0. Most estimates reveal an elasticity of
substitutionbetween skilled and unskilled workersgreaterthan 1
which also implies that p > 0.4 Therefore,in the remainder ofthe
paper I focuson the case p E (0,1), thoughthe formalanalysis does
not depend on this parameter restriction.
The main storyof the paper can now be told informally.As
shown in detail in the rest of this section, endogenous technical
progressimplies that
(6)
AhA= f(p,HIL).
Namely, the relative productivityof skilled workers depends on

the relative price of the skill-intensivegood (p =Ph Ipl) and the
relative supply of skilled workers (HIL). The formereffectis
similar to the impact of factorprices on technical change which
was emphasized by the literatureon induced innovations cited in
the introduction.Intuitively,when a good becomes more expensive, technologiesused in its productioncommand a higherprice,
increasing the incentives to upgrade these technologies.Since an
increase in HIL reduces p, this effectfurtherdepresses the skill
premiumin response to a rise in HIL.
The innovationofthis paper is the second term in f. The size
ofthe marketforskill-complementarytechnologiesrelative to the
market size of technologies complementaryto unskilled labor is
proportional to HIL. An increase in HIL thereforemakes the
invention of a new technologycomplementingskills more profit4. See Freeman [1986]. Practically,all estimates ofthe aggregate elasticityof
substitutionbetween high and low education workers are between U = 1 and 2.
These estimates control for time trends in the demand for skills or use crosssectional data. Thus, in terms ofthe model here, they correspondto the constant
technologyelasticity.So 1/v = 1 - Up < 1, and p > 0. Since a large part of the
substitutionbetween skilled and unskilled workersis within industries, p should
not be interpretedas the elasticityofsubstitutionbetween differentgoods.

1063
able and increases AhIA1,so

dAhiAl
aHIL
The formalanalysis below will establish that as long as p > 0, this
second effectdominates the price effect.Therefore,
dAhlAl _AhAhi
dH/L
ap
p
aP
aAhlAl
aHIL
aHIL
This is the essence ofthe directedtechnologyeffect:an increase in

the relative supply of skilled workersleads to an improvementin
the technologiesused by skilled workers.
Since technologyis given in the shortrun, an increase in HIL
first leaves AhIA, (mostly) unchanged and reduces the skill
premium.Then, the directionoftechnical progresschanges due to
the market size effect.As a result, the skill premium rebounds
fromits short-runlow. Moreover,if the directedtechnologyeffect
is sufficientlypronounced,the skill premium may rise above its
initial level, as drawn in Figure I. The rest ofthis section models
the R&D sector more formally,demonstrates that technical
change takes the formsummarized in equation (6), and analyzes
the dynamicresponse ofthe skill premiumto an increase in HIL.
B. TechnologicalAdvances
Firm level technologyAs(i) is determinedby the quality and
quantity of machines used. There is a continuumis E [0,1] of
machines foreach sector.The fact that each sector uses different
machines is the sense in which skilled and unskilled workersuse
differenttechnologies. The quantity of machine j that firmi in
sector s uses is denoted by x5(ij). Machines, the only form of
capital in this economy,depreciate fullyafteruse, which simplifies
the analysis. I denote the currentlyavailable highest quality of
machines in sectors by q,(j). Incorporatingthe factthat outdated
machines will not be used (which is true in equilibrium as shown
below), the productivityoffirmi takes the form;
(7)
As(i)= 1 -
s((j)x5(i,j)1
dj.
Notice that (4) and (7) implythat productionofYhand Y1is subject

1064
to constant returns to scale. The presence of a continuum of

machines in (7), rather than just one per sector, simplifiesthe
analysis. First, it implies that innovatorsdo not have to take their
impact on factorprices into account. Second, it ensures that the
growthrate ofthe economyis deterministic.
Firms (producingyi and Yh) purchase machines and labor to
maximize static profits.Let us denote the price of machine of
quality qs,(j) by X(i). Since there are no adjustment costs, firmi's
problemat all points in time is
max PA, (i)ns(i) -
ns(i),XS(ij)
Xs(j)x(i, J) dj - wn, (i).
The solution to this problem implies that the aggregate demand

formachines in sectors is Xs(j) = [psqs(j)N /xs(j)] 1/,where Nh=
Hand N1 L.
Technological advances take place as in Aghion and Howitt
[1992] and Grossman and Helpman [1991]. When there is an
innovationformachines, the quality ofthe machine increases by a
factorA > 1. I furtherassume that A > (1 - )-(1-P/, which is a
simplifyingassumption to be discussed in the next subsection.
The R&D firmthat innovates has a monopoly right over that
particular vintage (e.g., it holds a perfectlyenforcedpatent), so it
can charge a profit-maximizing
price and sell as many units ofthe
newly discoveredinput as it wishes. The marginal cost ofproducing input qs(j) is qs(j), so it increases linearly with the quality of
the machine.
Innovations are the result of R&D carried out by research
firmsusing onlyfinal output as factorofproduction.There is free
entryinto the R&D sector.If the total amount ofR&D activityin
technologyj forsectors is z, then the probabilityofinnovation in
this technology is z+(z). The marginal cost of R&D effortfor
inventinga machine ofvintage qs (j) is Bqs (j) (in termsofthe final
good).5 z+(z) is nondecreasing, i.e., P(z) + zP'(z) - 0, and P(.) is
everywhere smoothly decreasing, which implies decreasing returns to R&D effort(see the Appendix for the case of constant
5. Equivalently,the cost ofR&D effort
to improvevintage q,(j) is BXq,(j). The
assumption that R&D inputs are in terms offinal output serves to highlightthat
changes in skill premiumare not drivenby changes in the level ofR&D activity.If
R&D uses more skilled labor, then the skill premium will increase in periods of
high R&D activity,similar to the "skill-biased technologyadoption" effectemphasized in Nelson and Phelps [1966], Galor and Tsiddon [1997], and Greenwood and
Yorukoglu[1997], but this does not change the main results ofthe paper.

1065
returns to scale, 4(z)

1). Also, I impose limzngP(z) = oo and
limz_.(4(z) = 0. These Inada type restrictionson 4(.) ensure an
interiorsolution and smoothdynamics.
C. EquilibriumR&D Effort
The aggregate demands fortechnologycharacterized above
are isoelastic, so the profit-maximizing
price is a constantmarkup
=
over marginal cost: Xs(j)
q,(j)/(l
IP)for vintage qs(j). The
assumption that A > (1 - )-(1-)/
implies that even if the next
best technology,q5(j) = q,(j)/X, were sold at marginal cost, firms
would preferto buy qs (j) sold at the monopolyprice,ensuringthat
the monopolypricing policy is optimal (see Grossman and Helpman [1991]).
Given monopolyprices, everyfirmin the relevant sectorbuys
xS(ij) = Xs(j) = [(1 - 3)ps~N]"n. Therefore, the equilibrium
productivityin sectors, (7), can be writtenas
A
(1
13)(1-23)/f
Qs[psNf](1-f)1/
where I have defined

Q sf
qs (j)dj,
fors = 1,h. Qi and Qh are the average qualities ofmachines used in

the labor and skill-intensivesectors, and are the relevant measure oftechnologicalknow-howand the key state variables ofthis
economy.
The value ofowningthe leading vintage ofmachines ofsector
S is
(8)
rVs(j) = us (j) - zs (j)+(zs(zj))Vs
(j) + Vs (j),
where zs(j) is the current aggregate R&D effortto improve

machine j in sector s and wj(J) = PXs(j)qs(j)/(l - I) is the flow
profit.At the flow rate z5(j)((zj(j)),
the firmloses its monopoly
positionbecause there is an innovation,and the time derivativeof
V in (8) takes care of the fact that zs(j) may be time varying.
Finally,freeentryinto R&D activitiesimplies that
(9)
4(zs(j))Vs(j)
= Bqs(j),
where the left-handside is the marginal return to higher R&D

1066
effortdirected at this machine, and the right-hand side is the

marginal cost.6
An equilibrium in this economyrequires that firmsrent the
profit-maximizingamounts of all inputs, innovators follow the
profit-maximizing
pricingpolicy,product,intermediategood, and
labor markets clear, and there is no opportunityforany research
firmto enter (or exit) and increase its profits.Equations (3), (5),
(8), and (9) ensure these conditions.
D. Characterizing
theEquilibrium
Let us start with the balanced growthpath (BGP) where all
variables either grow at a constant rate or are time invariant. In
particular, since V denotes the value of an existing frontier
technology,we have V = 0. Then, equations (8) and (9) implythat
in BGP
(10)
3(1 -
)l?p p/Ns = B r +
zS(j)(zS
D)
forallj E [0,1] and s = 1,h. This equation states that innovation

effortformachinej is higher when profitsfromtechnologysales,
the left-handside, are higher. The profitswill be higher in turn
when the price ofthe productis higher or when more workersuse
this technology.It also immediatelyfollowsfrom(10) that zs(j) =
zs for all j. In other words, the BGP levels of effortfor all
skill-intensive(labor-intensive)technologies are the same, so we
onlyhave to determinetwo variables, z1and Zh.
Combining(10) fors = 1and s = h, we see thatZh/Zl, relative
research effortat skill-complementarytechnologies,is increasing
in p(H/L)P, as captured by the reduced-formequation f(p, HIL)
above. The price effecthas exactly the same intuition as before:
when Ph is high relative to PI, it is more profitable to invent
skill-complementarytechnologies because their output is more
technolexpensive.Simple algebra using (3) and profit-maximizing
6. There is free entryby small R&D firmswhich ignore their impact on the
inventionprobabilityofotherfirmsworkingto improvethe same machine. If there
had been one large firmdoingR&D on each machine,the left-handside of(9) would
have been [4V'(zs(j))zs (j) + 4(zs(j))]Vs(j). The choice between these two formulations does not change the results. Also, (9) ignores the constraint that total
expenditure on R&D should not exceed current output. Assuming that + is
sufficientlydecreasing would ensure that this constraint never binds. Alternatively,this constraintdoes not apply ifthe economycan borrowfromabroad at the
interestrate r.

1067
ogychoice gives the relative pricep as a functionofH/L:

(11)
p = y3v(Qh/QI ) -1(l-p)v(HI
L)
-f(1-p)v
(1 - (1- ,3)p)-1. Therefore,an increase in HIL dewhere v

presses p, and via the price effect,it induces the inventionofmore
labor-complementarytechnologies. Counteractingthe price effect
is the market size effect,the second argument of f in (6). When
there are more skilled workers, the size of the market for
skill-complementarytechnologies is larger. For p E (0,1], the
market size effectis more powerful.7
To determine the dependence of technologyon relative supplies more formally,note that there is a continuum of skillintensive inputs, so Zh4(Zh) is exactly the rate of improvements.
Hence (X - l)Zh4(Zh)
is the growth rate of Qh - folq(j) dj.
Similarly,(X - 1)z14(zj) is the growthrate ofQi. For BGP we need
Qh/ Qi to be constant; thereforez1 = Zh. Equation (10) then implies
that along the BGP
p = (HIL)-.
(12)
Intuitively,BGP requires that z1 = Zh, so the demand for skillcomplementarytechnologiesrelative to labor-complementarymachines should be independentofHIL, whichimplies (12). Combining (12) with (11), we see that in the BGP, the relative technology
ofskilled workersneeds to satisfy
(13)
Qh
WHp(l -0
Q_ _ 11(L I)(
Equation (13) is an important result. Qh/Qi is the equilibrium

technology level of the skill-intensive sector relative to the
labor-intensivesector,and depends on the relative abundance of
the two types oflabor. Since profitsto innovationare proportional
to market size, they are effectively
proportionalto the number of
workers using the technology.Therefore,when HIL increases,
innovation and R&D in the skill-intensive sector become more
profitable,inducing Qh/Qi to increase (as long as p > 0). This is
7. To see this, note that p is decreasing in HIL with elasticity P(1 - p)/
(1 - (1 - 1)p) which is less than I when p > 0. Although it is plausible for the
market size effectto dominate in the case of skilled and unskilled workers,the
price effectmay dominate in other situations. For example, Hayami and Ruttan
[1970] discuss the different
paths ofagriculturaldevelopmentin the United States
and Japan. The scarcityofland in Japan relative to the United States appears to
have induced a faster rate of innovation and adoption of fertilizers,increasing
outputper acre.

1068
the directed technologyeffect:the greater the fractionof skilled

workersin the economy,the greater their productivityrelative to
unskilled workers. Another implication of (13) is that as the
economy accumulates more skills, technical change responds by
making new technologies more complementaryto skilled labor.
The fact that the demand forskills has increased steadily in the
face of increasing supply of skills over the past century is
thereforeconsistentwiththe approach in this paper.Also, interestingly,duringthe eighteenthand early nineteenthcenturies,there
was a large migration of unskilled workers from villages and
Ireland to English cities (see Williamson [1990]). So this increase
in the supply of unskilled workers might have played a role in
inducing the creation of the well-knownskill-replacingtechnologies ofthat period.
Returning to the formal analysis, the BGP R&D effortlevel
can now be determinedfrom(10), (12), and (13) by imposingz1
Zh = z*, which gives8
-
(14)
B
[r + z*+(z*)1
4(Z*)
P=
(1 -
3)(1)/P
[,yIPP/(l - P) + LP/('1 - P)](l-)/fP
Finally, using the analysis so far and (5), we have (proofin the
Appendix):
PROPOSITION 1. There is a unique balanced growth path (BGP)
where both sectors and total output grow at the rate

(A - 1)z*+(z*) withz* given by (14). Along the BGP, Qh/QI is
given by (13) and the skill premiumis
(15)
where N
X=
3p2/(l - p) -(1-
P)(H/L)9
Up).
In the unique BGP there is a one-to-onerelation between the

relative supply of skilled workers and their relative wage. This
relation can be either increasing or decreasing. The second term
in q, -(1 - Up),is the usual substitutioneffectfromequation (5).
8. As in Aghion and Howitt [1992] and Grossman and Helpman [1991], the
growthrate increases with total population. This is not importantforthe focus of
this paper.All the results ofinterestcontinueto hold ifwe impose Zh + Zj = Z, which
would remove the scale effect.Alternatively,we can impose that forz - z, z4(z) =
+O < co, and the scale effectwould disappear once the population reaches a critical
threshold. Further,an increase in HIL leaving total population unchanged may
increase or decrease the BGP growthrate. The restrictionZh + zj = Z also removes
the effectofHIL on the growthrate.

TECHNICAL CHANGE AND WAGEINEQUALITY
1069
If technologywere exogenous in this economy,i.e., if AhiAI (or

Q /Qi) were constant, the skill premium would be a decreasing
functionofHIL. When technologyis endogenous, the increase in
HIL changes the direction of technical progress, and leads to
more R&D activityin the skill-complementarytechnologies.As a
result, Qh
/ Q1
increases, and the "short-runrelative demand
curve" shifts to the right as in Figure I. The long-run relative

demand curve is thereforemore elastic than the short-rundemand curve.9 Moreover, if Pp2/(l- p), the directed technology
effect,is large enough, q is positive, and the long-run relative
demand curve for skills is upward sloping.10 In this case, an
increase in the supply of skills leads to a higher relative price of
skill in the long run. This "perverse"case is more likelyto happen
when p is close to 1 so that the skill-intensiveand labor-intensive
productionprocesses (intermediate goods) are close substitutes,
and when a is close to 1 so that there are only limited decreasing
returnsto labor withineach sector.
A differentintuitionforthe possibly upward sloping relative
demand curve is that there is an importantnonconvexityin this
economy.There is a fixedup-frontcost ofdiscoveringnew technologies, and once discovered, they can be sold to many firms at
constant marginal cost. This nonconvexity(nonrivalryoftechnologyuse) implies that it is more profitableto improvetechnologies
designed fora larger clientele. This is the essence ofthe directed
technologyeffect.Interestingly,the size of this nonconvexity,B,
does not matterforthis result. Only when B = 0, the nonconvexity
and hence our results disappear, but in this case the growthrate of
the economy becomes infinity.To see the importance of the
nonconvexityand market size forthe results more clearly,one can
consider a modifiedmodel where technological monopolists can
only sell X units oftheir machines, whereX < min IL,HI.The rest
ofthe marketwill then be filledby imitatorswho produce a unit of
a machine of quality q at the marginal cost q/(l - a) (so that all
9. The long-run relative demand curve is more elastic than the short-run
demand curve even when p < 0, because from(13), an increase HIL reduces Qh/Q1,
but in this case, as (5) shows, the skill premium is decreasing in Qh/Q1. So even
when p < 0, an increase in HIL firstreduces and then increases the skill premium.
However,when p < 0, the long-runrelative demand curve can never slope up.
10. HIL has two effectson technology:first,it determines the quantities of
machinespurchasedat a givenQh /Qi, and second,it changesQh /QI. The second
effectis the importantone. To see this, suppose that there is no R&D, but As is still
determined by purchases of machines froma monopolist. In this case, the skill
premium is again a decreasing function of HIL with the smaller elasticity,
-(1 - p)/(l - (1 - U)p),which is also the short-runelasticityin Proposition3.

1070
price
machines sell at the same price, and the profit-maximizing
for the monopolist is the same as above). This modification
removes the market size effectbecause an increase in H does not
increase the market size of skill-complementarytechnologies for
the innovator.Following the same steps as above, we see that in
this case (13) becomes Qh/QI = y"/('-P)(H/L)-'. So an increase in
H/L only creates the price effecton technologicalimprovements,
technologies.
and causes slowerupgrading ofskill-complementary
The relative demand curve is always downward sloping in this
case.
The next propositionsummarizes the equilibrium dynamics
out ofthe BGP. Denoting the BGP level of Qh/Qi given in (13) by
Q*, we have (proofin the Appendix):
PROPOSITION2. (a) Locally, there exists a unique transition path
convergingto BGP, so that ifQh/Ql + Q*, then Zh and z, jump,
and QhIQl monotonically
adjuststo Q*. If Qh/Ql < Q*, then
z1along the transitionpath and vice versa.

(b) Suppose that the elasticity of the function +, E,4(z), is
Q *, there is a
nonincreasing in z. Then, for all QhAI Q
globally unique saddle path to BGP along which Qh /Qi
convergesto Q*. If Qh/Qi
Q *, thenZh > Z1
monotonically
along the transitionpath and vice versa.
Zh >
This propositionshows that the BGP is locally (saddle path)

stable, and under a fairlyweak assumption on the + function,also
globally stable. When Qh/Qi is below its BGP level, the economy
invests more in skill-complementarytechnologies (Zh > z1), raising Qh /Ql toward the BGP.
E. TheDynamicResponsetoa RelativeSupplyShock
The followingresult summarizes the dynamicresponse ofthe
economyto an unanticipated relative supply shock (proofin the
Appendix)."
PROPOSITION 3. Consider an unanticipated increase fromHIL to
3H/L starting from a BGP with skill premium w = w0.

Immediately after the shift,the skill premium falls to wSR,
11. The increase in the supply of skills during the 1970s may have been
anticipated. This does not change the qualitative conclusions reached in Proposition 3. The reason is that it is not profitable to invent skill-complementary
technologiesmuch beforeworkers who will use them are in the market, because
otherfirmsare likelyto improveon these technologiesbeforethe original inventor
has had access to the larger market.

1071
wherelog (0SR - log wo= -0 log 8 and 0 (1 - p)/- (1 3)p)> 0, and Zh /Z1 jumps up. The new BGP skill premium(0LR
is such that log wLR-
log w0= q log 8.
Following the relative supply shock, the skill premium falls

by Olog8. The short-run response is for given technological
know-howbecause Qh/Qi is a stock variable and changes slowly.
In terms of Figure I in the Introduction,this is a move along the
short-run(constant technology)relative demand curve,causing a
decline in the skill premium. As the economy adjusts to its new
BGP, Qh/Qi increases, and the skill premium starts increasing
from its short-term low. In terms of Figure I the constant
technologyrelative demand curve shiftsto the right. Therefore,
an increase in the relative supply of skills creates a period of
rising skill premium in response to the induced shift in the
relative demand forskills. This result is not very surprising: an
intuition based on the LeChatelier Principle suggests that the
elasticity of substitution between skilled and unskilled workers
should increase as other factorsare adjusted [Samuelson 1947].
In this paper the otherfactorsare measures oftechnology,Al and
Ah, and in the case q < 0, the results confirmthis intuition. In
contrast,when q > 0, the model's predictionsare more surprising
and original. Because of the nonconvexity introduced by the
market size effect,the increase in H/L has a large impact on
AhIA1,and eventually raises the skill premium above its initial
value.
Proposition 3, especially in the case where q > 0, offersan
alternative explanation for the behavior of the U. S. economy
duringthe past 25 years. There was a large increase in the supply
of skills in the 1970s. Between 1940 and 1970 the relative supply
ofcollege equivalent workersgrew at the rate of2.73 percentper
year. In contrast, between 1970 and 1980 this relative supply
grew almost twice as fast,at 5.19 percentper year (in otherwords
by 52 percent in the course of ten years, see Autor, Katz, and
Krueger [1998]). This is a verylarge change in the relative supply
of skills precedingthe rise in the college premium. Furthermore,
these supply changes were at least partly"exogenous"ratherthan
a simple response to anticipated higher returns to education in
the future. Enrollment rates had been increasing since the
mid-1950s, and this trend continued in the 1960s (see Freeman
[1976], Figure 6, p. 35). The main reason forthe increase in the
proportionofcollege graduates in the labor forceduringthe 1970s

1072
was the interactionof these higherenrollmentrates with the

large cohortsizes arrivingin the marketduringthe 1970s. In
particular,the cohortsretiringin the 1970s had veryfewcollege
graduates.Forexample,only7.5 percentofthose55 yearsorolder
in 1970 were collegegraduates[Bureau ofthe Census 1971]. In
trendsdatingback to the 1950s,
contrast,due to the enrollment
21.3 percentofthoseaged 30 to 34 in 1981,a cohortthatentered
themarketbetween1970 and 1980,had a collegedegree[Bureau
due tothebabyboom,theentering
oftheCensus 1983].Moreover,
cohortswerelargerelativeto the laborforce:in 1981 therewere
38 millionpeople betweenthe ages of 25-34 as comparedwith
48.5 millionpeoplebetweenthe ages of35 and 54 [Bureau ofthe
Census 1983].
by increasingthe enrollTwo otherfactorsalso contributed
mentrates furtherduringthe late 1960s and 1970s: (1) until
almostthe end ofthe war,the Vietnamera draftlaws exempted
service.Thisinducedmany
malesenrolledin collegefrommilitary
moreyoungmalesto stayin collegeduringthelate 1960sin order
to avoid the draft[Baskir and Strauss 1978]; (2) government
financialaid forcollegeincreasedby a large amountduringthis
era. For example,the total federalaid to collegestudentsthat
$2 billionin 1963 increasedto $14 billion
stoodat approximately
in 1970-1971and thento$24 billionin 1975-1976(all numbersin
1989-1990 dollars,see McPhersonand Schapiro [1991]).12 My
theorypredictsthatin responsetothislargeincreaseinHIL, the
collegepremiumshould fall first,and then increasedue to the
induced skill-biasedtechnical change. This pattern matches
thebroadbehavioroftheU. S. collegepremiumfrom1970 to the
present,and suggestsan explanationforwhytherelativedemand
forcollegegraduatesincreasedmuchfasterin the past 25 years
thanbetween1940 and 1970 [Autor,Katz, and Krueger1998].13
The model also predictsthat aftera large relativesupply
change,the growthrate ofthe economyshoulddecline.During
adjustmenttothenewBGP,Zh increases,and z1falls,and because
12. It can be argued, however, that federal aid increased because the
governmentforecastthe increased need forcollege graduates. Even ifthis were the
case, which is unlikely,the large part of the increase in the supply of college
graduates due to the cohortsize effectis still predeterminedand "exogenous."
13. The model does not predict a fall in unskilled wages, which has been a
feature ofthe changes in wage structureduringthe 1980s. If there is depreciation
ofexistingtechnologies,forexample, because some ofthe old technologiesare not
compatible with new ones, the model also predicts that during adjustment to an
increase in HIL unskilled wages may fall. See Acemoglu [1996] foran alternative
model forthe decline in unskilled wages.

1073
W.)is decreasing-i.e., returnto R&D is concave-the faster

technologies
in skill-complementary
technological
improvements
growthof
do notcompensatefortheslowdownin theproductivity
unskilledworkers.Therefore,a large increasein HIL not only
changesthe structureof wages, but also causes a productivity
slowdown.Unfortunately,(0, which determinesthe extentof
is noteasilyobservedinthedata,so it
thisproductivity
slowdown,
is not possible to know whetherthis is an importanteffect.
Greenwoodand Yorukoglu[1997] also obtain slower growth
duringthe processofadjustmentto new technologiesbecause of
costsofadoptionand learning.In contrastto thatpaper,technologicalchangeis endogenoushere,and theslowergrowthis due to
skill-complementhe factthatthe economyinvestsin improving
used byunskilled
at theexpenseoftechnologies
tarytechnologies
labor.
Finally,it is instructiveto do some back-of-the-envelope
calculationsto see whetherthe model can generateplausible
effects.
I take the relativesupplyshockto be the increasein the
ratioofcollegegraduatesto highschoolgraduatesfrom1971 to
1979,since1971 was the startingyearforthelargechangein the
skillcomposition
ofthelaborforce.This givestherelativesupply
shock,A log (HIL) = log 8, as 0.4 (TableVIII in Katz and Murphy
[1992]).I take the long-runresponseto thisincreasein supplyto
be the proportional
changein the collegepremiumfrom1971 to
1987, whichis reportedas 0.024 by Katz and Murphy[1992]. I
comparethis numberwiththe long-runresponseimpliedbythe
response
model,q log 8 (= log ()LR - log w0).I takethe"short-run"
to be thechangein theskillpremiumfrom1971to 1979,whichis
-0.10 [Katz and Murphy1992]. This numbercannotbe directly
ofa one-timeshockgivenin
comparedwiththe immediateeffect
3 becausein realitythesupplyshocktookplace overa
Proposition
musthave adjustedduringthis
numberofyears,so technology
period. Therefore,I compare it with (log wSR + log wLR)/2- log wo,
afterthe
whichis the simpleaverageofthe changeimmediately
shockand thelong-run
response.
For a rangeofparameterchoices,themodelimpliesnumbers
verycloseto the data. For example,when 3 = 0.35 and p = 0.75,
the modelyields a short-runelasticityof substitutionbetween
collegeand noncollegeworkersin the rightrange and = 0.06.
fall in the college premiumof 11
This implies a short-term
percent,comparabletotheonein thedata,and a subsequentlarge
increase,takingittoabout2.5 percentaboveitsinitialvalue.This

1074
is quite close to the actual behaviorof the college premium.

Similarresultsare obtainedwhen R= 0.4 and p = 0.73,orwhen
P = 0.45 and p = 0.7. These parameterchoicescan be givensome
justificationin terms of more microestimates(see Acemoglu
arbitrary.
So these resultsshouldbe
[1997]),but are inherently
interpreted
as purelyillustrative.In particular,small changesin
p changethe quantitativeimplicationsby a large amount.Also,
furtherincreasesin HIL in the 1980s make these calculations
difficult
to interpret.
III. DIRECTED TECHNICALCHANGE AND RESIDUAL
WAGEINEQUALITY
Equating skilled workers to those with a college degree, I

suggested that a model incorporating the directed technical
change can match the evolution of the U. S. college premium.
Anotherimportantaspect ofthe changes in the structureofwages
is increased residual wage inequality. Using March Current
Population Surveys, Juhn, Murphy,and Pierce [1993] and Katz
and Murphy[1992] findthat residual (withingroup)wage inequalitybegan increasing duringthe 1970s while the college premium
fell. Bernard and Jensen [1997] find the same pattern in the
census data. (But DiNardo, Fortin,and Lemieux [1996] do not find
it using May CurrentPopulation Surveys.) To date, therehas been
no unified explanation forthe simultaneous increase in residual
inequalityand the declinein the college premiumduringthe

1970s.14In thissectionI suggesta simpleextensionofthe model
which accounts forthis pattern.
educationand skills
Supposethatskillsare two-dimensional:
For lack ofa betterterm,I call
unobservedto theeconometrician.
the latter ability.A fractionAh of college graduates have high
is ,ul< Jh
ability,and theremainderhave lowability.Thisfraction
amonglow educationworkers.For example,ability(unobserved
so
skills)couldbe relatedto collegeeducation,but notperfectly,
that some of the college graduates do not acquire the necessary
skills,whilesomeotherworkersdo in spiteofnothavingattended
14. Galor and Tsiddon [1997] argue that abilityis morevaluable in periods of
rapid technological change, which offersanother explanation forthe increase in
residual inequality duringthe 1970s. Caselli [1997] suggests,but does not develop,
a related storywhere due to on-the-job-training,
high abilityworkersbenefitfrom
rapid technologicalchange first.Both stories can explain why residual inequality
mighthave grownfasterthan the college premium duringthe 1970s, but without
further modifications,not why college premium fell and residual inequality
increased.

1075
skills are not innate,but

college.Therefore,ability/unobserved
acquiredpartlythrougheducation.ThereareH collegegraduates
and L low educationworkers.Suppose also that the aggregate
function
oftheeconomyis
production
Y=
[(Ahh(uhH)
)P ? (Alh((l
YOh)H) )
+ (Ahl(ulL))
P + (All((1
- 41) )PL)]
/P
whichbasicallycombinesthe equivalentequationsto (2) and (4),

whichis
and also imposesthatall firmsuse the same technology,
truein equilibrium.The importantassumptionwe make is that
machines,qh(j), whileAlh
Ahh and Ahi use skill-complementary
andAlluse theq1(j) machines.Forexample,
Ah
= 1/(1 - 1)
f0 qh(j)Xh
(j)
dj,
machinesused
whereXhl(j) is thequantityofskill-complementary
in hi. This formulationimplies that both high abilitycollege
graduates and high abilityhigh school graduates use similar
(see Barteland Sicherman[1997] forsomeevidence
technologies
to ability/unobthat advancedtechnologiesare complementary
served skills). An analysis similarto the one used previously
impliesthatin BGP
Qh
Q(6((1
Pv +
(t1L) Pv
h)H)pv + ((1 - YL)PvJ
(JhH)
1
l/(l-p)v
where v -(1
- (1 - r)p)-1 (see the Appendix for details). The
Wh = Whh /Wlh
and xl = Whl W11,which are the ratios ofthe wages of
of high abilityworkersincreasesthe relativeabunproportion

to ability.This has exactly
dance oftechnologiescomplementary
thesame intuitionas theresultsin SectionII.
Nowconsideran exogenousincreaseinHIL overa numberof
periods.Because Jh > ?2 > yl, the proportionof high ability
workersin thelaborforceincreases,inducinga risein Qh /Qi. The
college premiumbehaves as in the previous section:first,it
declinesfora while in responseto the increasedsupplyof H
workers.Then, as Qh /Qi increases,the college premiumalso
ofhighability
startsincreasing,because thereis a largerfraction
workersamongthe collegegraduates.In contrastto the college
premium,which falls first,however,residual wage inequality
startsincreasingimmediatelyafterthe shock.To see this,considerthetwomeasuresofresidualwage inequalityin thismodel,
highto low abilityworkerswithintheireducationgroups.Argu-

1076
mentssimilartothosedevelopedbeforeimplythat
=
wh
1
\l y
IQh~~
YpiIh
-o
and
(1O=
h IpLl
-0
/Q~\~
) k11 Ii)
where0 and v are as definedabove(see theAppendix).Noticethat

the
theincreasein the supplyofcollegegraduatesequallyaffects
ofthemeasuresofresidualwage
numeratorand thedenominator
inequality,wh and wx,so there is no contractionin residual
inequalityin responsetotheincreaseinHIL. But thesemeasures
the
to Qh /Qi, so the increasein Qh/ Qi following
are proportional
change in relativesupplies leads to an immediateincrease in
residualwageinequality.
to unobserved
iftechnologiesare complementary
Therefore,
skills, the approach developedin this paper predictsthat in
responseto increasesin therelativesuppliesofeducatedworkers
as in the 1970s, the college premiumdeclines firstand then
increases,whereas residual wage inequalitystarts increasing
immediately.
IV. ENDOGENOUSSUPPLYOF SKILLS
Section II treated the relative supply of skills as exogenous.
The increasein the supplyof collegegraduatesduringthe late

1960sand 1970swas arguedtobe largelyexogenousratherthana
simple response to anticipated higher returns in the future.

Nevertheless, education choices are to some degree forwardlooking and respond to returns. It is therefore important to
endogenize the supply of skills and verifythat the main results
are robust. This analysis also provides new results regardingthe
joint behavior ofskills and technology.
Suppose now that a continuumv ofunskilled agents are born
everyperiod,and each faces a flowrate ofdeath equal to v, so that
population is constant at 1 (as in Blanchard [1985]). Each agent
chooses upon birth whether to acquire education to become a
skilled worker.For agent x it takes KXperiods to become skilled,
and duringthis time,he earns no labor income. The distributionof
KXis given by the function F(K) which is the only source of
heterogeneityin this economy,due to creditmarket imperfections
or differencesin innate ability.The rest ofthe setup is unchanged.
To simplifythe exposition,I assume that F(K) has no mass points.
I now definea BGP as a situation in which HIL and the skill
premium remains constant. In BGP there is a single-crossing

1077
property:if an individual with cost of educationKX chooses

schooling,anotherwithKxt< KXmustalso acquireskills.Thereleveloftalent,K, suchthatall Kx> K do
fore,thereexistsa cutoff
not get education.AlthoughHIL is in general a complicated
function
ofpast educationdecisions,whenv is small,alongBGP it
takesthesimpleform,
H
L
(17)
F(K)
1 - F(K)
between
The agent with talent K needs to be indifferent
acquiringskillsand not.Whenhe does notacquireanyskills,his
returnat timet is
Rne= Jtexp[-(r + v)(T - t)]w(T) dT

-w1
J0exp
[-(r + v - g)T] dT = r + v
wherer + v is the effective

discountrate and I have used thefact
that along the BGP wages grow at the rate g = (A - 1)z *(z*).
If
in contrastK decidesto acquireeducation,he receivesnothingfor

a segmentof time of lengthK, and receivesWh fromthen on.
thereturnto agentK fromacquiringeducation,Re(K),
Therefore,
can be writtenas
Re(K)
S- exp [-(r +
v)(T
t)]Wh(T)
dT
-exp [-(r + v - g)K]
In BGP, forK to be indifferent,

we need Re(K) = Rneat all times,
this equationand
sO W WhiW1 = exp [(r + v - g)K]. Inverting
into(17), we obtainthe relativesupplyofskillsas a

substituting
function
oftheskillpremium(whenv is small):
(18)-
F(ln w/(r+ v - g))
1 - F(ln w/(r+ v - g))

_
A BGP equilibriumwithendogenousskillformation
is given
by the intersectionof the relative supply (18) with relative
demandforskillsgivenby(15). IgnoringtheimpactofHIL ong,
which is likelyto be small, equation (18) definesan upward
slopingrelativesupplyrelation.When q > 0, (15) also definesan

1078
upwardslopingrelativedemandcurve,and multipleBGP equilibria, as drawnin FigureII, are possible.Intuitively,

when-q> 0, a
higherHIL increases w, encouragingworkerswith high K to
obtaineducationand increasingH/L further.
Government
policy(e.g.,the grantprogramsor theVietnam
era draftlaws) can be thoughtofas reducingthecostofeducation
therelativesupplycurvein FigureII to theright(or
and shifting
the function
shifting
F(K) to the left).When q > 0, the returnto
educationX wouldalso rise,thusraisingHIL further.
Therefore,
the predictionof the model in this case is that subsidies to
educationlead to an increasedtendencyto acquireeducation,and
also to a largereducationpremiumdue tothedirectedtechnology
effect.
when-q> 0, educationsubsidiesmayharm
Interestingly,
thoseagentswhodo nottake advantageoftheincreasedincentive
to attend college, which is different
fromthe predictionsof
standardtheorywhereeven thosewho do nottake advantageof
educationsubsidiesbenefit.Also noticethat a small exogenous
increasein the demandforskills-an increasein -yin equation
(2)-increases the skillpremiumimmediately,
encouragingmore
effectand
skill acquisition.This inducesthe directedtechnology
If the supplyofskillsis
increasesthe demandforskillsfurther.
sufficiently
responsiveto the skill premium,the exogenousin-
RelativeSupply---Equation
(18)
SkillPremium(o
Wo3
I'dR
elativeDemand--Equation(15)
(021. ... ... .. ... .. ... ... .. ... .. ... ..
(H/L)1
(H/L)2
(H/L)3
FIGURE II
HIL
BalancedGrowthPathEquilibriawithEndogenousSkills

1079
crease in the demand for skills causes a decline in the skill

premiumbeforeincreasing it above its initial value.
A formalanalysis oftransitorydynamicsis more complicated.
To simplifythe discussion, suppose that there is a unique BGP
and the supply ofskills is not too responsive to the skill premium
(i.e., F'(K) is sufficientlysmall or the relative supply curve in
Figure II is sufficientlysteep). Then, it can be shown that in the
neighborhoodofthe BGP, an increase in HIL, due to a decline in
the cost of education or other reasons, first reduces the skill
premium. Then, HIL and Qh/Qi increase together,creating both
skill-biased technical change and a larger supply of skills. In the
case where -q> 0, the economyends up withhigherskill premium,
more skill-complementarytechnologies,and more skilled workers
(see the Appendix). The process of adjustment is likely to take a
long time because technologyadjusts slowly(especially in the case
where 4(.) is steeply decreasing). Therefore,the skill premium
will rise over time, and earlier generations will be less willing to
invest in skills as the returnsare farin the future.This pattern of
slowly increasing HIL and a gradual shift toward more skillcomplementaritytechnologies is similar to the experience of the
United States economyover the past century.
V. THE IMPACTOF TRADEON TECHNOLOGY
Increased trade withthe LDCs where skilled labor is scarce is
oftensuggested as a potential cause ofincreased wage inequality
and contrasted with the explanations based on technology.Since
technologyhas been treated as exogenous in the wage inequality
literature,there has been littleeffortto uncoverthe links between
these two explanations. This section shows that the direction of
technical change is influencedby trade, modifyingor qualifying
many ofthe conclusions reached in the previous literatureregarding the impact oftrade on inequality.
Suppose that the North,where the ratio ofskilled to unskilled
workersis equal to HNILN, begins trading with an economy,the
South, which has a skill ratio Hs ILS < HNILN. There is no
endogenous skill accumulation in either economy.What happens
to wage inequality in the North?I will answer this question under
three differentscenarios: (a) no directed technical change; (b)
directed technical change and new technologies sold to firmsin
the South on the same terms as firmsin the North; (c) directed
technical change and no propertyrightsenforcementin the South.

1080
The firstscenariois fora benchmark,and the truthpresumably

lies somewherebetween(b) and (c), so thatthereis somesale of
of
technologyto the firmsin the South, but the enforcement
intellectualproperty
rightsis less thanperfect.Notethat,in this
model,factorprice equalizationis guaranteedwithoutfurther
restrictions
becauseeach sectoremploysonlyoneofthenontraded
factors(see Ventura[1997]fora similarstructure).
A. No Technical Change
Denote the
Suppose that Qi and Qh are givenexogenously.
steady state (BGP) skill premiumin the North beforetrade
openingbywN,and theBGP skillpremiumaftertradeopeningby
(w. Also defineA log w log ww- log wN,HW = HN + HS, LW =
LN + Ls and HW/LW = 8HN/LN, where 8 < 1 by the fact that the
than the South (i.e., log6 < 0).

North is more skill-intensive
Equation(5) fromSectionII impliesthat
(19)
A log1 NTC= -log
8>
0,
NTC denotes"notechnicalchange"and 0 was

wherethesubscript
tradewith
definedabove. Since the South is less skill-intensive,
theSouthincreasestherelativepriceofskillsin theNorth,which
is thestandardeffect
oftrade.
B. Endogenous Technical Change and Full PropertyRights
SupposethatAl and Ahare endogenousas in SectionII, and

assume that (i) beforetrade opening,there were no sales of
technologyto the firmsin the South (and no foreigndirect
in theSouthbyfirmsin theNorth);and (ii) aftertrade
investment
and
opening,firmsin the South and the Northare symmetric
property
rightsofR&D producersin theNorthare fullyenforced
in the South.Tradeopeningis thenequivalentto a declinein the
relativesupplyofskillsfromHNILN to HW/Lw. We can nowuse
equation(15) fromSectionII to obtainthe changein BGP skill
premiumaftertradeopening:
(20)
A log (opR= mlog 8,
where the subscriptPR indicates that in this case there is

endogenoustechnicalchangeand propertyrightsofR&D firms
are enforcedin the South. If mq> 0, contraryto conventional
wisdom,tradeopeningmayactuallyreducethe skillpremiumfor
exactlythe same reasonsthata highersupplyofskilledworkers

1081
increased it in Section II (though as in Proposition 3, trade

opening would firstincrease inequality and then reduce it). More
generally,the directedtechnologyeffectimplies that when intellectual propertyrightsare fullyenforced,it is unlikelythat international trade increases the skill-premiumby a large amount.
C. No IntellectualProperty
Rightsin theSouth
A simple way of modeling the lack of intellectual property
rights is to suppose that imitators produce and sell the latest
machines invented by R&D firmsin the North to firmsin the
South, so that Northern R&D firmsdo not receive any revenue
fromfirmsin the South. To simplifythe analysis, suppose that
imitators can produce a machine of quality q at marginal cost
q/(l - A), so that firmsin the South use the same technologyas
the firmsin the North. This specificationimplies that the market
sizes for differentmachines are unchanged after trade opening.
However, there is still an impact on the direction of technical
change because the relative price ofskill-intensivegoods changes.
Trade opening first depresses the price of labor-intensive
goods (i.e., reduces p). However, in the BGP, (10) has to hold in
order to equate the return to R&D in skill-complementaryand
labor-complementarymachines, because R&D firmscan only sell
technologiesto firmsin the North.This implies that
p =
(HNILN)-P
so relative prices have to adjust back to their original level to

restore equilibrium, and this requires an adjustment in the
relative technologyofskilled workersaway fromits original level.
l
-1.
In particular,in the new BGP, QhQl Q = Y/(l-P)
(HN/LN)P/(l-P)
The relative wage is now fw = 'yl/(l-p)
8-1, and the
(HNILN)
change in the skill premiumaftertrade opening is
(21)
A log WNPR
log8 > 0,
where the subscript NPR indicates that there is endogenous

technical change but no enforcementof intellectual property
rights in the South. The assumption of no intellectual property
rightsin the South has removedthe market size effectcompletely
(the relative market size is stillHN/LN), but the price effecton the
incentives to innovate is present.As noted in Section II, the price
effectmagnifiesthe negative effectof the increase in HIL on the
skill premium,which explains the large impact of trade on wage

1082
trade openingin this case may

Also, interestingly,
inequality.15
also increasethe skillpremiumin LDCs, because firmsin LDCs
and trade openinghas inducedskilluse the same technology,
biased technicalchange in the North.This contrastswith the
ofa modelwithexogenoustechnology.
prediction
In summary,
we have A log (NPR > Alog WNTC > Alog WPR and
rightsare fully
in factif q > 0, A log (WPR< 0. That is, ifproperty
enforcedin the South,the declinein the relativesupplyofskills
should not lead to a large increase in the skill premium.In
contrast,if intellectualpropertyrightsare not enforcedin the
South,simplecalculationsthatignorethe inducedchangein the
directionoftechnicalprogressmaybe seriouslyunderestimating
tradeon inequality.
theimpactofinternational
VI. CONCLUDING COMMENTS
The wages ofcollegegraduatesand ofotherskilledworkers

relativeto unskilledlabor increaseddramaticallyin the United
States over the past fifteenyears. To many economistsand
thisis a directconsequenceofthe complementarcommentators,
itybetweenskill and new technologies.It is not clear,however,
skills.Historyis fullof
shouldcomplement
whynewtechnologies
examplesofnew technologiesdesignedto save on skilledlabor.
More generally,inventionsand technologyadoption are the
we couldhave chosento
outcomeofa processofchoice;as a society,
It is
technologies.
developor attemptedto developmanydifferent
necessaryto analyzethe directionoftechnicalchange
therefore,
as well as its magnitude.In its simplestform,thismeansto pose
skills?"This
complement
the question:"whydo new technologies
paper suggestedthat the directionoftechnicalchangeis deterinventions.When
minedby the size ofthe marketfordifferent
thereare moreskilledworkers,the marketfortechnologiesthat
skillsis larger,hencemoreofthemwillbe invented,
complement
to skills.
willbe complementary
and newtechnologies
I
I formalized
thisobservationand discusseditsimplications.
showedthatan exogenousincreasein theratioofskilledworkers
or a reductionin the costofacquiringskillscouldincreasewage
15. This argument is related to Wood's [1994] suggestion that trade with the
South may have induced "defensive"unskilled-labor-savinginnovations.However,
it is not clear what mechanism Wood has in mind since a decline in unskilled wages
should normally lead to the introductionof skill-replacing or unskilled-laborcomplementarytechnologies.

1083
inequality.The likelypath is firsta decline,and then a large

increasein the skill premium.These observationsfitthe U. S.
factswherethe large increasein the ratio of collegegraduates
during the late 1960s and 1970s firstdepressed the college
premiumand thenincreasedit tohigherlevelsthanbefore.
The mostimportant
area forfutureworkis todevelopa testof
directedtechnicalchange, and its impact on the structureof
wages. The testable implicationof the model is that afteran
increase in the supply of college graduates,R&D directedat
to collegegraduatesshouldincrease.
technologies
complementary
in generalto determinewhichtechnologies
Althoughit is difficult
are complementary
to skilledworkers,mosteconomistsbelieve
to skilledand educated
thatcomputersare morecomplementary
workersthan to the unskilled.For example,Autor,Katz, and
Krueger[1998]reportthatin 1993only34.6 percentofhighschool
graduatesuse computersin contrastto 70.2 percentof college
for
graduates.Moreover,Krueger[1993] shows that controlling
education,workersusing a computerobtain a wage premium
which suggests that they are more skilled. From the R&D
expendituredata reportedby the NSF, we see that in 1960
R&D forofficecomputingwas 3 percentofthe
company-funded
R&D expenditure.
This ratiohas increased
totalcompany-funded
thatduringthisperiodofrapid
to 13 percentby1987,16suggesting
more
increasein the supplyofskills,therehas been significantly
to
R&D directedto one ofthe technologiesmostcomplementary
skills. If othertechnologiesand R&D expenditurecan also be
to collegegraduates,the hypothesis
classifiedas complementary
ofthispapercan be tested.
A secondarea forfutureresearchis the applicationofthese
Since the 1970s,the
ideas to the male-femalewage differentials.
labor forceparticipationof womenhas increasedsubstantially,
and their wages relative to those of male workershave also
increased.Partofthischangeis likelytohavebeendue toreduced
discrimination.
However,to the extentthat male workersuse
different
technologiesthan femaleworkers,the approachin this
paper suggestsa new explanationbased on directedtechnical
technologies
change.The degreeto whichwomenuse different
thanmenwithina plantor sectoris probablylimited.Nevertheand
sectorsand occupations,
less,womentendtoworkin different
16. These data come fromvarious issues of Research and Development in
IndustryDetailedStatisticalTablespublishedbytheNSF. I am gratefulto Sam
Kortumforprovidingme withthese data.

1084
these jobs use differenttechnologies than traditionallymale jobs

(e.g., desk jobs versus construction).It is thereforeconceivable
that the greater participation of women may have affectedthe
direction of technical change, and via this channel, reduced
male-female wage differentials.This hypothesis can be investigated more carefullyby studyingthe relative growthofindustries
that employ more women, and the relative rate of technical
change in these industries.
APPENDIX
ProofofProposition 1
Equation (13) uniquely definesQh/ Ql forZh= z1. Then, using
(10), (12), and (13) and imposingz* = Zh= z1 gives equation (14),
which uniquely defines z* because the left-hand side of (14) is
strictlyincreasing in z * and the right-handside is constant. This
establishes that the BGP exists and is uniquely defined. Now
substitutingforAhIAI and forp, (5) implies
(,O =
'-l/[(l-(l-P)p1
[(Qh/Q0~PP1/(l-(l-P)p)-
Finally,substitutingforQh/ Ql using (13) gives (15) in the BGP.

TransitionDynamics and ProofofProposition2
Equation (10) immediatelyimplies that z,(j) = z, out ofBGP
as well as along it. So we only have to determinethe time path of
Zh, Zl, Qi, and Qh. Equation (9) holds at all times, so differentiating
it with respectto time and using (8), we obtain
_s =-ss)V/
44~V~4(Zs)V~
s ()
s-VZS1E'(Zs)Vs)
P(1 - )(1P-/ in this Appendix to

fors = 1, h. I normalize B
simplifythe notation. Combining this with (8) and using (9), we
obtain
r + zs((zs) - (Zs)ps11PNs
~ ~~~~~~~~En
(ZS )IZS
(22)
Qh
Finally,notingthatQS/Qs = (X - 1)4(zs) zs and defining

Q
/Ql, we also have
(23)
= (X - 1)(Zh4(zh)
Z- 14z1))Q.
Equations (22) and (23) completelydescribe the dynamics of the

system. To analyze local dynamics and stabilityin the neighbor-

1085
hood of the BGP, I linearize these equations. Then, around the

BGP,zl =Zh =Z, Q =Q*, and ignoringconstants,we have
Zh
07(Z*)(Zh
j =
(yZ*)(Z
+ + 1(Z*,Q*)(Q
- Z-)/(E+(Zp)/Z*)
-Q),
-
k(Z*,Q)(Q
Q*)
- z 1), where u(z*)

4'(z*)z* + 4+(z) > 0. 4p
and Q ((Z*)Q*(Zh
and 4'2are analogously defined,and are both positive. The reason
is that
why deviations of Q fromQ* affectz1 and Zh differently
whenQ > Q*, Ph is aboveits BGP value, andPI is belowits BGP
value. This linearization enables us to reduce the three variable
systemto two variables: Q and =Zh - z. Specifically,
=
z (Z))Q
U+(Z Z
? 1VR 'Q
+(Z
)(Q - Q*),
where vI(z*,Q*) = hi(Z*,Q*) + ti2(Z*,Q*) > 0. This linear system

has one negative and one positive eigenvalue, and thus a unique
saddle path converging to the BGP equilibrium. The rate of
convergenceto the BGP is
/1 u(z*)
V2E,(Z *)/Z*
u(Z*)
+ ((Z
*)4(Z *,Q)
2E-
(Z *)Iz
Thus, when u(z*) = 4' (z*)z* + 4(z*) is lower,or when 4(.) is more
steeply decreasing, convergenceis slower. In fact,in the extreme
case of4(z) =1 (where u(z*) is at its highest),all our BGP results
would be unchanged, but ignoringnonnegativityconstraints on
consumption, there would be no transitory dynamics. That is,
when Q < Q*, we would have z1 = 0 and Zh ' o? for an
infinitesimallyshortwhile. There would be once again transitory
dynamics,however,if nonnegativityconstraints on consumption
are imposed.
Next, I establish that if E,(z) is nonincreasing,the system is
also globally saddle path stable. Since paths cannot cross and
there are no other stationarypoints ofthe system,all paths that
do not cycle must go to infinity.Therefore, we only have to
establish that there are no cycles. Suppose that Q < Q*. Note that
in this case phH' > p1LP. Then consider case (A), where z1 > Zh.
Then using (22) and the fact that E,~(z) is nonincreasing, 1/;zl >
Zh /Zh. Therefore,z1 will remain larger than Zh, and Q < 0. Thus,
there cannot be any cycles, and all paths go to infinitywhen z1 >
Zh. Now consider case (B), whereZh > z1 and 1/zl < Zh lZh. Now Q >

1086
0, and also as Q increasesPh falls,and PI increases.Therefore,

it
willalwaysbe the case that 1/z < h /Zh. Hence,in thiscase too,
cyclesare notpossible.Now considercase (C), whereZh > z1 and
00, Zh > z1, and i1/z, > ZhlZh, then we
i1/Z1 > ZhlZh. If as t
convergetoZh = zi = z* and Q = Q*,and we knowthereis a unique
we mustbe
saddle pathlocallyand pathscannotcross.Therefore,
on thatpath.InsteadifZh = z1 at somepointwhereQ 0 Q*,then
onceagain cyclescan be ruledout.We musthave eitherthatZh >
z1 and Q > Q* whichputs us in case (A), and cyclesare not
Or,it couldbe the case that
possible,and all paths go to infinity.
>
>
z1
by
the
argument
analogoustocase (A),
Zh
and Q Q*,which,
Thus,
there
must
be
a unique saddle path
again rulesout cycles.
<
The
proof
for
the
case of Q > Q* is
fromall pointsQ
Q*.
analogous.
-
ProofofProposition3
Take Qh /Qi as given.Then,givenoptimalmonopolypricing

we have
and profit
maximization
byfirms,
AhiAl = (Qh/QI)P(
-P?1 )H/L)(1-?)
Nowsubstituting
forPh andpI and rearranging,
_h
h_)J]
[y(1-y)/ (Qh)
into(5), we obtain
Substituting
where 0
(1 - p)/(l - (1 - 3)p)and v
(1 - (1 -
3)p)-1as de-
at giventechnology
(i.e.,givenQh /QI),
finedinthetext.Therefore,
=
adjuststo its new BGP level,
A log w -0 log 6. Oncetechnology
we have theresultofProposition1. Thus,Alog w = q log 6.
Details of theModel ofSection III
The demandsforsectorh machinesnow come fromfirms

employinghigh abilitycollegegraduatesand high abilityhigh
school graduates,and vice versa forsector1 machines.These
demandcurveshave the same elasticityas in thetext.Thus,the
optimal pricingpolicyis the same. Therefore,the free-entry
conditionforsectorh machines,analogousto (10), can be written
as
(24)
KZh )[phhJhH + p )j/j1L] =
r+
Zh4(Zh),

1087
and similarlyforzj, wherePhhis thepriceoftheintermediate

good
producedby high abilitycollegegraduatesin termsofthe final
good,Phl is forhighabilityhighschoolgraduates,etc.(recallthat
B
(1 - 3)(1?/P in this Appendix). Then, using competitive
pricingand thedefinition
ofAhh,
=
A=
(1
)(1) 2p)v/Qh(lP)v(hH)-(1-p)vy(1-p)v
where v
(1 - (1 - P)p)-1 and Phl, Plh, and Plu are similarly
defined.Substitutingthese into (24) and simplifying,
we obtain
(16) in thetext.
Finally,considerresidual wage inequalityamong college
graduates:
Whh _
Wh=
Ahh\p
A h /(1
JhH
-(13p)
-Jh)H)
Substituting
forAhh and Alh givesthe expressionin thetext.wxis
derivedsimilarly.
TransitoryDynamics withEndogenous Skills
I focuson the case ofinterest,whichhas mq> 0. Linearizing

arounda BGP,Q stillonlydependson 4.Thus,ignoring
constants,
Q = a21A,wherea21> 0. In contrast,4 = ajjy + a12Q- a(HIL),
whereall coefficients
are positive.Thisis becausez, dependson V,
whichis an increasingfunction
ofQ, and decreasinginNs. Finally,
aroundthe BGP, we have thatH
-vH + F(log w/[r+ v - gD,
and similarlyforL. Therefore,
the rate of change ofHIL is a
decreasingfunctionof HIL and an increasingfunctionof the
relativewage w,whichis itselfdecreasingin HIL and increasing
in Qh/Ql. Thus, in the neighborhood
ofthe BGP the linearized
differential
equation is (HIL) = a32Q - a33(H/L), where once
are positive.Notethata32is theresponseof
again thecoefficients
the supplyof skills to the skill premium,so it is smallerwhen
F'(K) is lower; that is, when educationchoices are not very
responsiveto wages. Therefore,
aroundtheBGP,thissystemhas
two state variables and one controlvariable. Thus, for wellbehavedtransitiondynamics,thelinearizeddifferential
equation
systemneeds to have two negative (xbx2) and one positive
eigenvalues(x3). Standard argumentsestablish that x1x2x3=
small-so that
a2l(al2a33a13a32). So when a32 is sufficiently
educationchoices are not too responsive-eitherall roots are
positive,ortwoofthemare negative.Also,x1x2+ x1x3 + x2a33+ a12
a21)< 0, whichimpliesthatall rootscannotbe positive.So longas
educationis nottooresponsivetotheskillpremium,
thesystemis

1088
saddle path stable around the BGP. Therefore,when HIL and

Qi are below their BGP value, we have Zh > zj, and the
economy converges to BGP by accumulating skills and more
skill-complementarytechnologies.
Qhl
MASSACHUSETTS INSTITUTE OF nTCHNOLOGY
REFERENCES
Acemoglu, Daron, "Changes in Unemploymentand Wage Inequality: An Alternative Theoryand Some Evidence,"CEPR Discussion Paper No. 1459, July1996.
,"Why Do New Technologies Complement Skills? Directed Technical Change
and Wage Inequality,"CEPR Discussion Paper No. 1707, September 1997.
Ahmad, Syed, "On The Theory of Induced Invention,"Economic Journal, LXXVI,
(1966), 344-357.
Aghion, Philippe, and Peter Howitt, "A Model of Growth through Creative
Destruction,"Econometrica,LX (1992), 323-351.
Autor, David, Alan Krueger, and Lawrence Katz, "Computing Inequality: Have
Computers Changed the Labor Market?" Quarterly Journal of Economics,
CXIII (1998), 1169-1213.
Barro, Robert, and Xavier Sala-i-Martin, Economic Growth(New York: McGraw
Hill, 1995).
Bartel, Ann, and Frank Lichtenberg,"The Comparative Advantage of Educated
Workers in Implementing New Technologies," Review of Economics and
Statistics, LXIX (1987), 1-11.
Bartel, Ann, and Nachum Sicherman, "Technological Change and Wages: An
Inter-IndustryAnalysis,"NBER WorkingPaper No 5941, 1997.
Baskir, Lawrence M., and William Strauss, Chance and Circumstances:theDraft,
the Warand the VietnamGeneration(New York:AlfredKnopf,1978).
Berman, Eli, JohnBound, and Zvi Griliches,"Changes in the Demand forSkilled
Labor within U. S. Manufacturing Industries: Evidence from the Annual
Survey of Manufacturing," Quarterly Journal of Economics, CIX (1994),
367-398.
Bernard, Andrew,and BradfordJensen,"UnderstandingRising and Falling Wage
Inequality: Evidence fromU. S. States," Yale University,mimeo, December
1997.
Blanchard, Olivier, "Debt, Deficits and Finite Horizons," Journal of Political
Economy,XCIII (1985), 223-247.
Bound, John, and George Johnson, "Changes in the Structure of Wages in the
1980s: An Evaluation of Alternative Explanations," American Economic
Review,LXXXII (1992),371-391.
Bureau ofthe Census, Statistical Abstractofthe United States, 1971 (Washington,
DC: 1971).
, Statistical Abstractofthe United States, 1982-83 (Washington,DC: 1983).
Caselli, Francesco, "Technological Revolutions," University of Chicago, mimeo,
June 1997.
David, Paul, Technical Choice, Innovation and Economic Growth: Essays on
American and British Experience in the NineteenthCentury(London: CambridgeUniversityPress, 1975).
DiNardo, John,Nicole Fortin, and Thomas Lemieux, "Labor Market Institutions
and the Distribution of Wages: 1973-1992," Econometrica, LXIV (1996),
1001-1044.
Freeman, Richard, The OvereducatedAmerican (New York:Academic Press, 1976).
___
"Demand For Education," Chapter 6 in Orley Ashenfelter and Richard
Layard, editors, Handbook of Labor Economics, Vol I (Amsterdam: NorthHolland, 1986), pp. 357-386.
Galor, Oded, and Daniel Tsiddon, "TechnologicalProgress,Mobilityand Economic
Growth,"AmericanEconomic Review,LXXXVII (1997), 363-382.

1089
Goldin, Claudia, and Lawrence Katz, "The Decline of Noncompeting Groups:

Changes in the Premium to Education, 1890 to 1940," NBER WorkingPaper
No.5202,1995.
Goldin, Claudia, and Lawrence F. Katz, "The Origins ofTechnology-SkillComplementarity,"QuarterlyJournal ofEconomics, CXIII (1998), 693-732.
Greenwood, Jeremy,and Mehmet Yorukoglu, "1974" Carnegie-RochesterConference Series on Public Policy,XLVI (1997), 49-95.
Review ofEconomics and StatisGriliches, Zvi, "Capital-Skill Complementarity,"
tics,LI (1956), 465-468.
Grossman, Gene, and Elhanan Helpman, "Quality Ladders in the Theory of
Growth,"Review ofEconomic Studies, LVIII (1991),43-61.
Habakkuk, H. J., American and British Technologyin the Nineteenth Century:
Search for Labor Saving Inventions, (Cambridge: Cambridge University
Press, 1962).
Hayami, Yujiro, and Vernon Ruttan, "Factor Prices and Technical Change in
Agricultural Development: The U. S. and Japan, 1880-1960," Journal of
Political Economy,LXXII (1970), 1115-1141.
John,Chinhoi, Kevin M. Murphy,Brook Pierce, "Wage Inequality and the Rise in
Return to Skills,"Journal ofPolitical Economy,CI (1993), 410-442.
Katz Lawrence, and Kevin Murphy, "Changes in Relative Wages: Supply and
Demand Factors,"QuarterlyJournal ofEconomics, CVII (1992), 35-78.
Kennedy,Charles, "Induced Bias in Innovation and the Theory of Distribution,"
Economic Journal, LXXIV (1964), 541-547.
Kiley, Michael, "The Supply of Skilled Labor and Skill-Biased Technological
Progress,"Board of Governorsofthe Federal Reserve System, mimeo,March
1997.
Krueger, Alan, "How Computers Have Changed the Wage Structure: Evidence
fromMicrodata, 1984-1989," QuarterlyJournal of Economics, CVIII (1993),
33-60.
Krugman, Paul, "And Now for Something Completely Different:An Alternative
Theory of Trade, Education and Wages," Massachusetts Institute of Technologymimeo,March 1997.
Krusell, Per, Lee Ohanian, VictorRios-Rull, and Giovanni Violante, "Capital Skill
Complementaryand Inequality,"UniversityofRochester,mimeo, 1997.
McPherson, Michael, and Morton Schapiro, Keeping College Affordable:Governmentand Educational Opportunity,(Washington,DC: Brookings Institution,
1991).
Mokyr,Joel,The Levers ofRiches: TechnologicalCreativityand Economic Progress
(New York: OxfordUniversityPress, 1990).
Nelson, Richard, and Edmund Phelps, "Investment in Humans, Technological
Diffusion,and Economic Growth,"American Economic Review Papers and
Proceedings,LVI (1966),69-75.
Romer,Paul, "Endogenous Technological Change," Journal of Political Economy,
XCVIII (1990), S71-S102.
Samuelson, Paul, The Foundations of Economic Analysis (Cambridge, MA:
Harvard UniversityPress, 1947).
"A Theory of Induced Innovations along Kennedy-WeisackerLines," Review
ofEconomics and Statistics,XLVII (1965), 444-464.
Schmookler,Jacob, Invention and Economic Growth (Cambridge, MA: Harvard
UniversityPress, 1966).
Ventura,Jaume, "Growthand Interdependence,"QuarterlyJournal ofEconomics,
CXII (1997), 57-84.
Walde, Klaus, "Relative Quality ofEducation, Induced Technological Change and
Wages,"Universitat DortmundDiscussion Paper, February 1997.
Williamson, Jeffrey,Coping with City Growth during the British Industrial
Revolution(Cambridge: Cambridge UniversityPress, 1990).
Wood, Adrian, North-South Trade, Employment and Inequality: Changing Fortunes in a Skill Driven World(Oxford:Clarendon Press, 1994).


Why Do New Technologies Complement Skills

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Why Do New Technologies Complement Skills

Hochgeladen von

Copyright:

Verfügbare Formate

Why Do New Technologies Complement Skills?

Directed Technical Change and Wage Inequality

This content downloaded from 188.72.126.118 on Wed, 18 Jun 2014 15:17:48 PM