Content uploaded by Ahmed S Rahman
Author content
All content in this area was uploaded by Ahmed S Rahman on Oct 04, 2014
Content may be subject to copyright.
UNITED STATES NAVAL ACADEMY
DEPARTMENT OF ECONOMICS
WORKING PAPER 2011-35
“Trade, Technology and the Great Divergence”
by
Kevin H. O’Rourke
All Souls College, Oxford University
&
Ahmed S. Rahman
U. S. Naval Academy
&
Alan M. Taylor
University of Virginia
Trade, Technology and the Great Divergence∗
(or An Economic History of the World in Just Thirty-Six Equations)
Kevin H. O’Rourke, Ahmed S. Rahman, and Alan M. Taylor
April 2011
Abstract
This paper develops a model that captures the key features of the Industrial Revolution
and the Great Divergence between the industrializing “North” and the lagging “South.”
In particular, a convincing story is needed for why North-South divergence occurred so
dramatically during the late 19th Century, a good hundred years after the beginnings of
the Industrial Revolution. To this end we construct a trade/growth model that includes
both endogenous biased technologies and intercontinental trade. The Industrial Revolution
began as a sequence of unskilled-labor intensive innovations which initially incited fertil-
ity increases and limited human capital formation in both the North and the South. The
subsequent co-evolution of trade and technological growth however fostered an inevitable di-
vergence in living standards - the South increasingly specialized in production that worsened
their terms of trade and spurred even greater fertility increases and educational declines.
Biased technological changes in both regions only reinforced this pattern. The model high-
lights how pronounced divergence ultimately arose from interactions between specialization
from trade and technological forces.
•Keywords: Industrial Revolution, unified growth theory, endogenous growth, demog-
raphy, skill premium, Great Divergence
•JEL Codes: O, F, N
∗Preliminary draft. Please do not quote.
1 Introduction
The last two centuries have witnessed dramatic changes in the global distribution of income
and population. At the dawn of the Industrial Revolution, living standards between the richest
and poorest economies of the world were roughly 2 to 1. With industrialization came both
income and population growth within a few core countries. But massive divergence in living
standards across the globe did not take place until the latter half of the 19th century, the time
when the first great era of globalization started to take shape (see Figure 1). Today the gap
between material living standards in the richest and poorest economies of the world is 30 or 40
to 1, in large part due to the events of the 19th century. It seems an interesting coincidence
then that such unprecedented growth in inter-continental commerce (conceivably creating a great
force for convergence by exploiting comparative advantages and facilitating flows of knowledge)
coincided so precisely with unprecedented divergence in living standards across the world. Why
did incomes diverge just as the world became flat? These phenomena beg for an explanation.
This paper argues that trade and technological growth patterns together sowed the seeds for
divergence, contributing enormously to today’s great wealth disparity.
Some important “stylized facts” from economic history motivate our theory. One concerns
the nature of industrialization itself - technological change was unskilled labor-intensive during
the early Industrial Revolution but became relatively skill-intensive during the latter nineteenth
century. Indeed, England’s early industrialization was in many respects largely a revolution
in the cotton textile industry, with the adoption of the factory system of production and its
associated new machinery. The textile industry was revolutionary in its ability to employ large
numbers of unskilled and uneducated workers with minimal supervision, thus diminishing the
productive role for skilled labor and education (Galor 2005; Clark 2007). By the 1850’s, however,
two major changes in technological growth occurred - it became much more widespread, and it
became far more complementary to skilled workers (Mokyr 2002).
Another historical feature of great importance was the rising role of international trade in the
world economy. Inter-continental commerce between “western” economies and the rest of the
world (what we might mildly mislabel as “North-South” trade) was not particularly robust until
the latter half of the 19th century. By the 1840s steam ships were faster and more reliable than
sailing ships, but their high coal consumption limited how much cargo they could transport;
consequently only very light and valuable freight was shipped (O’Rourke and Williamson 1999).
But by 1870 a number of innovations dramatically reduced the cost of steam ocean transport,
and real ocean freight rates fell by nearly 35% from 1870 to 1910 (Clark and Feenstra 2003). By
1900 the economic centers of the “South” such as Alexandria, Bombay and Shanghai were fully
integrated into the British economy, both in terms of transport costs and capital markets (Clark
2007). Thus, while a closed economy model would be more appropriate to describe the first
stages of the Industrial Revolution (1750-1850), a more open economy framework would better
1
Figure 1: Incomes per Capita Relative to India
sources:Maddison(1989),PradosdelaEscosura(2000),PennWorldTables
2
describe the latter stages of industrialization (1850-1910).
To analyze the intellectual puzzle of the Great Divergence, we develop a model that has a
number of key features which mimic these historical realities. The first feature of our approach is
that we endogenize the direction and extent of technological change in both regions. Technologies
are sector specific, and sectors have different degrees of skill intensities. Following the endogenous
growth literature, we allow potential innovators to observe the employment of factors in different
sectors, and tailor their research efforts towards particular sectors. Thus the scope and direction
of innovation will depend on each region’s employment and demography.
The second key feature is that we endogenize demography itself. More specifically, we allow
households to make education and fertility decisions based on market wages for skilled and
unskilled labor. The method is similar to other endogenous demography models where households
face a quality/quantity tradeoff with respect to their children. Thus, when the premium for skilled
labor rises families choose to have fewer but better educated children.
The final feature is that we allow for burgeoning trade between the North and the South.
During the initial stages of industrialization, trade is not possible due to prohibitively high
transport costs. These costs however exogenously decrease over time; at a certain point trade
becomes feasible, at which time the South exchanges labor-intensive products for the North’s skill-
intensive products. At this stage development paths begin to diverge - the North’s specialization
in skilled production produces a demographic transition, while the South’s specialization in
unskilled production generates unskilled-labor intensive technological growth but consequently
produces more population as well.
With this basic setup, we simulate the model to roughly capture the main features of industri-
alization and divergence between the North and the South from roughly 1700 to 1900. Because of
the great abundance of unskilled labor in the world, innovators first develop unskilled-intensive
technologies. Thus early industrialization is characterized by unskilled intensive technological
growth and population growth both in the North and the South; consequently living standards
in the two regions do not diverge during this time. Once trade becomes possible, however,
the North starts specializing in skill-intensive innovation and production. This induces a de-
mographic transition of falling fertility and rising education rates in the North. The South of
the other hand specializes in unskilled-intensive production, inducing both unskilled-intensive
technological growth and further population growth (see Figure 2). This population divergence
actually fosters deterioration in the South’s terms of trade, forcing the South to produce more
and more primary commodities for skill-intensive products and generating even more fertility
increases (see Figure 3). Thus the South’s static gains from trade become a dynamic impetus to
prosperity, and living standards between the two regions diverge dramatically as a result.
3
Figure 2: Population Growth Rates in the U.K. and India
source:Maddison(2001)
4
Figure 3: Relative Price of Primary Products According to Lewis and Prebisch 1870-1950 (1912
= 100)
source:HadassandWilliamson(2003)
5
Alternative Stories of Divergence
We argue that analyzing the interactions between the North and the South, and between
trade and technological flows, is critical to understanding both the Industrial Revolution and the
Great Divergence. In contrast many explanations of divergence rely on institutional differences
between regions of the world (North and Thomas 1973, Acemoglu et al. 2001, 2005). From
this perspective economic growth is a matter of establishing the right “rules of the game,” and
underdevelopment is simply a function of some form of institutional pathology. In the end
however it is unclear if institutions made any difference to the course of European growth, or if
inefficient institutions had any quantifiable and meaningful affect on growth in peripheral regions
(Clark 2007). For example, Pomeranz (2000) argues that by 1800 China had an economic system
that was as developed, market driven and individually rational as Europe’s. The same has been
said of India, whose institutions of secure property rights, free trade, fixed exchange rates and
open capital markets were nearly ”ideal” for development (Clark and Wolcott 2003).
Another potential explanation for the divergence is that peripheral countries were specializing
in inherently less-productive industries (Galor and Mountford 2006, 2008). But this also is not
very convincing - so called low-technology sectors such as agriculture enjoyed large productivity
advances during the early stages of the Industrial Revolution (Bekar and Lipsey 1997, Clark
2007). And in the twentieth century, developing countries specialized in textile production which
of course experienced massive technological improvements more than a century before.
Another related puzzle is the scale of the developing world. If a full one third of the world
had become either Indian or Chinese by the twentieth century (Galor and Mountford 2002), why
were Indians and Chinese not more wealthy? After all, most semi-endogenous and endogenous
growth theories have some form of scale effect, whereby large populations can create a boon
to innovative activities (Acemoglu, forthcoming).1Any divergence story that focuses on the
explosive population expansion in peripheral economies faces this awkward implication from the
canonical growth literature.
Relation to Galor and Mountford’s “Trade and the Great Divergence”
The paper presented here relates most closely and obviously to Oded Galor and Andrew
Mountford’s theoretical works on the Great Divergence (Galor and Mountford 2006, 2008)
(henceforward ‘GM’). These papers similarly suggest that the South’s specialization in unskilled-
intensive production stimulated fertility increases which lowered per capita living standards.
However, we offer a unique narrative in describing both the North’s launch into modernity and
1More specifically, in such seminal endogenous growth models as Romer (1986, 1990), Segerstrom, Anant and
Dinopolous (1990), Aghion and Howitt (1992), and Grossman and Helpman (1991), a larger labor force implies
faster growth of technology. In ”semi-endogenous” growth models such as Jones (1995), Young (1998), and Howitt
(1999), a larger labor force implies a higher level of technology.
6
the South’s vicious cycle of underdevelopment that is quite distinct in a number of ways.
The first involves the nature of trade - since we are concerned with an era with limited
exchanges of differentiated products and little intra-industry trade, trade is best modeled as
Ricardian (based on productivity differences) or Hecksher-Ohlin oriented (based on factor dif-
ferences).2GM uses the former approach, while we use both. Such a hybrid trade model seems
most appropriate to us given the rather large differences in factor technologies and skewed factor
endowments between the North and the South for the period we study.
Second, we endogenize both the scope and the direction of technological progress in both re-
gions. GM on the other hand make assumptions concerning the timing and speed of technological
growth which they claim are “consistent with historical evidence.” Specifically, they assume that
1) modernization either in agriculture or in manufacturing is not initially feasible, 2) moderniza-
tion occurs first in the agricultural sector, and 3) growth in industrial-sector productivity is faster
than growth in agricultural-sector productivity. Compelling as these assumptions are to us, they
are by no means universally held.3To give stronger credence to the so-called “Crafts-Harley
view” that early industrialization was confined to just a few industries (Crafts and Harley 1992),
we form a technological growth model that can endogenously mimic these historical trends.
Finally, rather than suddenly open up the North and South to trade, we allow for gradual
increases in North-South commerce. The British economy (and other Western economies) pre-
sumably did not undergo a discontinuous switch from a closed to an open state, and thus we
impose continuously declining transport costs to achieve such a transition.
These key differences allow us to conclude something that is somewhat under-explored in GM
- growth in trade and technologies together created the dramatic divergence in living standards
between the North and the South. More specifically, while the South’s specialization in agricul-
ture and low-end manufacturing allowed for plenty of technological advances in these areas, they
did not help the South grow for two reasons. One is that it fostered fertility increases similar to
the process outlined in GM. The other is that the South’s terms of trade deteriorated over time.
As the South grew in population it increasingly made up a larger share of the world population.
It thus flooded the world markets with its primary products. The skill-intensive products pro-
vided by the North on the other hand became relatively more scarce, and thus fetched higher
prices. The South had to provide more and more primary products to buy the same amount of
high-end products; this served to raise fertility rates by even more. This mechanism, absent in
2We know that Heckscher-Ohlin oriented trade was important during the 19th century since commodity price
convergence induced factor price convergence during this time (O’Rourke and Williamson 1994; O’Rourke, Taylor
and Williamson 1996; O’Rourke and Williamson 1999, Chapter 4). And Mitchener and Yan (2010) suggest that
unskilled-labor abundant China exported more unskilled-labor intensive goods and imported more skill-intensive
goods from 1903 to 1928, consistent with such a trade model.
3See for example Temin (1997) who suggests that the traditional view of the British Industrial Revolution as
a broad change that affected all industries is still in high regard among some economic historians.
7
GM’s work, suggests that productivity growth (and the scale that generated this growth) could
not salvage the South; it in fact contributed to its relative decline.
To further examine the interactions between trade and technologies, we run counter-factual
simulations where either technological growth in both regions is not possible, or trade between
the two regions is not possible. Divergence in either case is minuscule compared with the case
where trade and technological progress interact with each other. Both factors worked in tandem
to generate the massive wealth disparity we see in the world today.
2 Production with Given Technologies and Factors
We now sketch out a model that we will use to describe both a northern economy and a
southern economy (superscripts denoting region are suppressed for the time being).
Total production for a region is given by:
Y=α
2y
σ−1
σ
1+ (1 −α)y
σ−1
σ
2+α
2y
σ−1
σ
3σ
σ−1(1)
where α∈[0,1] and σ≥0. σis the elasticity of substitution among intermediate goods y1,y2,
and y3. The production of these goods are given by:
y1=A1L1(2)
y1=A2Lγ
2H1−γ
2(3)
y3=A3H3(4)
where A1,A2and A3are the technological levels of sectors 1, 2, and 3, respectively.4These
technological levels in turn are represented by a series of sector-specific machines. Specifically,
A1=ZN1
0x1(j)
L1α
dj (5)
A2=ZN2
0x2(j)
Lγ
2H1−γ
2α
dj (6)
A3=ZN3
0x3(j)
H3α
dj (7)
4Thus sectors vary by skill -intensity. While our interest is mainly in the “extreme” sectors (1 and 3), we
require an intermediate sector so that production of intermediate goods are determined both by relative prices
and endowments, and not pre-determined solely by endowments of Land H. This will be important when we
introduce trade to the model.
8
where xi(j) is machine of type jthat can be employed only in sector i. Intermediate producers
choose the amounts of these machines to employ, but the number of types of machines in each
sector is exogenous to producers. Technological progress in sector ican then be represented by
growth in the number of machine-types for the sector, denoted as Ni(we endogenize the growth
of these in the next sections by introducing researchers).
Treating technological coefficients as exogenous for the time being, we can assume that markets
for both the final good and intermediate goods are perfectly competitive. Thus, prices are equal
to unit costs. Solving the cost minimization problems for productions, and normalizing the price
of final output to one, yields the unit cost functions
1 = hα
2σ(p1)1−σ+ (1 −α)σ(p2)1−σ+α
2σ(p3)1−σi
1
1−σ(8)
p1=wl
A1
(9)
p2=1
A2wγ
lw1−γ
h(1 −γ)γ−1γ−γ(10)
p3=wh
A3
(11)
where pidenotes the price for intermediate good yi,wlis the wage paid to Land whis the wage
paid to H.
Full employment of total unskilled labor and total skilled labor implies the following factor-
market clearing conditions:
L=y1
A1
+wγ−1
lw1−γ
h(1 −γ)γ−1γ1−γy2
A2
(12)
H=wγ
lw−γ
h(1 −γ)γγ−γy2
A2
+y3
A3
(13)
Finally, the demands for intermediate goods from final producers can be derived from a
standard C.E.S. objective function.5Specifically, intermediate goods market clearing requires
yi= Υσ
ip−σ
i
α
2σ(p1)1−σ+ (1 −α)σ(p2)1−σ+α
2σ(p3)1−σ!Y(14)
for i= 1,2,3, Υ1= Υ3=α/2, and Υ2= 1 −α.
Provided that we have values for L,H,A1,A2and A3, along with parameter values, this
yields thirteen equations [(1) - (4), (8) - (13), and three versions of (14)] with thirteen unknowns
[Y,p1,p2,p3,y1,y2,y3,wland wh,L1,L2,H2, and H3]. The solution for these variables
5Here demands will be negatively related to own price, will be a function of a price index, and will be
proportional to total product.
9
constitutes the solution for the static model in the case of exogenously determined technological
and demographic variables.
3 Endogenizing Technologies in Both Regions
In this section we describe how innovators in both the North and the South endogenously
develop new technologies. In general, modeling purposive research and development effort is
challenging when prices and factors change over time. This is because it is typically assumed
that the gains from innovation will flow to the innovator throughout his lifetime, and this flow will
often depend on the price of the product being produced and the factors required for production
at each moment in time.6If prices and factors are constantly changing (as they will in any
economy where trade barriers fall gradually or factors evolve endogenously), a calculation of the
true discounted profits from an invention may be impossibly complicated.
To avoid such needless complication but still gain from the insights of endogenous growth
theory, we assume that the gains from innovation last one time period only. More specifically,
technological progress is sector-specific, and comes about though increases in the varieties of
machines employed in each sector. New varieties of machines are developed by profit-maximizing
inventors, who monopolistically produce and sell the machines to competitive producers of the
intermediate goods y1,y2or y3. However, we assume the blueprints to these machines become
public knowledge the time period after the machine is invented, at which point these machines
become old and are competitively produced and sold.7Thus while we need to distinguish between
old and new sector-specific machines, we avoid complicated dynamic programming problems
inherent in multiple-period profit streams.8
Thus, we can re-define sector-specific technological levels given by (5) - (7) as a series of both
old and new machines at time t(once again suppressing region superscripts) as:
A1,t = ZN1,t−1
0
x1,old(j)αdj +ZN1,t
N1,t−1
x1,new(j)αdj!1
L1α
A2,t = ZN2,t−1
0
x2,old(j)αdj +ZN2,t
N2,t−1
x2,new(j)αdj!1
Lγ
2H1−γ
2α
6For example, the seminal Romer (1990) model describes the discounted present value of a new invention as
a positive function of L−LR, where Lis the total workforce and LRare the number of researchers. Calculating
this value function is fairly straight-forward if labor supplies of production workers and researchers are constant.
If they are not, however, calculating the true benefits to the inventor may be difficult.
7Here one can assume either that patent protection for intellectual property lasts one time period, or that it
takes one time period for potential competitors to reverse-engineer the blueprints for new machines.
8See Rahman (2009) for more discussion of this simplifying (but arguably more realistic) assumption.
10
A3,t = ZN3,t−1
0
x3,old(j)αdj +ZN3,t
N3,t−1
x3,new(j)αdj!1
H3α
where xi,old are machines invented before t, and xi,new are machines invented at t. Thus in each
sector ithere are Nt−1varieties of old machines that are competitively produced, and there
are Nt−Nt−1varieties of new machines that are monopolistically produced (again, suppressing
country subscripts).
Next, we must describe producers of intermediate goods in each region. These three different
groups of producers each separately solve the following maximization problems:
Sector 1 producers: max[L1,x1(j)] p1y1−wlL1−RN1
0χ1(j)x1(j)dj
Sector 2 producers: max[L2,H2,x2(j)] p2y2−wlL2−whH2−RN2
0χ2(j)x2(j)dj
Sector 3 producers: max[H3,x3(j)] p3y3−whH3−RN3
0χ3(j)x3(j)dj
where χi(j) is the price of machine jemployed in sector i. For each type of producer, solving
the maximization problem with respect to machine jyields a solution for machine demand:
x1(j) = χ1(j)1
α−1(αp1)1
1−αL1(15)
x2(j) = χ2(j)1
α−1(αp2)1
1−αLγ
2H1−γ
2(16)
x3(j) = χ3(j)1
α−1(αp3)1
1−αH3(17)
New machine producers, having the sole right to produce the machine, set the price of their
machines to maximize instantaneous profit. This price will be a constant markup over the
marginal cost of producing a machine. Assuming that the cost of making a machine is unitary
implies that χ1(j) = χ2(j) = χ3(j) = χ= 1/α for new machines. Thus, substituting in this
mark-up price, and realizing that instantaneous profits are (1/α)−1 multiplied by the number
of new machines sold, instantaneous revenues by new machine producers are given by:
π1=1−α
αα2
1−α(p1)1
1−αL1(18)
π2=1−α
αα2
1−α(p2)1
1−αLγ
2H1−γ
2(19)
π3=1−α
αα2
1−α(p3)1
1−αH3(20)
11
Old machines, on the other hand, are competitively produced; competition will drive the price
of all these machines down to marginal cost, so that χ1(j) = χ2(j) = χ3(j) = χ= 1 for all old
machines. Sectoral productivities can then be expressed simply as a combination of old and
new machines demanded by producers. Plugging in the appropriate machine prices into our
machine demand expressions (15) - (17), and plugging these machine demands into our sectoral
productivities, we can express these productivities as:
A1=N1,t−1+αα
1−α(N1,t −N1,t−1)(αp1)α
1−α(21)
A2=N2,t−1+αα
1−α(N2,t −N2,t−1)(αp2)α
1−α(22)
A3=N3,t−1+αα
1−α(N3,t −N3,t−1)(αp3)α
1−α(23)
Thus, if we have given to us the number of old and new machines that can be used in each
sector (the evolution of these are described in section 5.1 , we can simultaneously solve equations
(8) - (14) and (21) - (23) to solve for prices, wages, intermediate goods and technological levels
for a hypothetical economy. Our next goal then is to also endogenize the levels of skilled and
unskilled labor in this hypothetical economy.
4 Endogenizing Population and Labor-Types in Both Re-
gions
We now introduce an over-lapping generations framework, where individuals in each region
live for two time periods. In their youths individuals work as unskilled workers; this income
is consumed by their parents. When they become adults, individuals decide whether or not to
expend a fixed resource cost to become a skilled worker. Adults also decide how many of their
own children to have, who earn unskilled income for the adults. Adults however are required to
forgo some income for child-rearing.
Specifically, an adult i’s objective is to maximize current-period income. If an adult chooses to
remain an unskilled worker (L), she aims to maximize Ilwith respect to her number of children,
where
Il=wl+nlwl−wlλnφ
l(24)
wlis the unskilled labor wage, nlis the number of children that the unskilled adult has, and
λ > 0 and φ > 1 are constant parameters that affect the opportunity costs to child-rearing..
12
If an adult chooses to spend resources to become a skilled worker, she instead maximizes Ih
with respect to her number of children, where
Ih=wh+nhwl−whλnφ
h−τi(25)
whis the skilled labor wage, nhis the number of children that the skilled adult has, and τiis the
resources she must spend to become skilled.
The first order conditions for each of these groups give us the optimal fertility for each group,
n∗
land n∗
h:
n∗
l= (φλ)1
1−φ(26)
n∗
h=wh
wl
φλ1
1−φ
(27)
Note that with wh> wl, the optimal fertility for a skilled worker is always smaller than that
for an unskilled worker (this is simply because the opportunity costs of child-rearing are larger for
skilled workers). Also note that the fertility for unskilled workers is constant, while the fertility
for skilled workers falls with increases in the skill premium.
Finally, assume that τvaries across adults. The resource costs necessary to acquire an educa-
tion can vary across individuals for many reasons, including differing incomes, access to schooling,
or innate abilities. Say τiis uniformly distributed across [0, b], where b > 0. An individual iwho
draws a particular τiwill choose to become a skilled worker only if her optimized income as a
skilled worker will be larger than her optimized income as an unskilled worker. Let us call τ∗
the threshold cost to education; this is the education cost where the adult is indifferent between
becoming a skilled worker or remaining an unskilled worker. Solving for this, we get
τ∗=wh+n∗
hwl−whλn∗φ
h−wl−n∗
lwl+wlλn∗φ
l(28)
Only individuals whose τifall below this level will opt to become skilled.
Figure 4 illustrate optimal fertility rates for two individuals - one with a relatively high τand
one with a relatively low τ. The straight lines illustrate how earnings increase as adults have
more children; the slope of these lines is simply the unskilled wage wl. The earnings line for
a skilled worker is shifted up to show that she earns a premium. Cost curves get steeper with
more children since φ > 1. For skilled individuals, the cost curve is both higher (to illustrate the
resource costs τnecessary to become skilled) and steeper (to illustrate the higher opportunity
cost to have children). Notice then that the only difference between the high-τindividual and
the low-τindividual is that the latter has a lower cost curve. The optimal fertility rates however
are the same for both types of adults. Given these differences in the fixed costs of education, we
can see that the high-τindividual will opt to remain an unskilled worker (and so have a fertility
13
rate of n∗
l), while the low-τindividual will choose to become skilled (and have a fertility rate of
n∗
h).
With this we can describe aggregate supplies of skilled and unskilled labor (demands for these
labor types are described by full employment conditions (12) and (13)), fertility and education.
Given a total adult population equal to pop, we can describe these variables as:
H=τ∗
bpop (29)
L=1−τ∗
bpop +npop (30)
n=1−τ∗
bn∗
l+τ∗
bn∗
h(31)
e=τ∗
b(32)
where His the number of skilled workers (comprised strictly of old workers), Lis the number
of unskilled workers (comprised of both old and young workers), nis aggregate fertility, eis the
fraction of the workforce that gets an education, and n∗
l,n∗
h, and τ∗are the optimized fertility
rates and threshold education cost given respectively by (26), (27) and (28).
This completes the description of the static one-country model. The next section uses this
model to describe two economies that endogenously develop technologies and trade with each
other to motivate a story of world economic history.
5 The Roles of Trade and Technological Growth in the
Great Divergence
In this section we show how the interactions between the growth of trade and biased tech-
nologies contributed to the great divergence of the late 19th Century. To do this we perform
a thought experiment by simulating two economies. The above model describes a hypothetical
country - now we will use it to describe both a “northern” economy and a “southern” economy,
where the southern economy is relatively more unskilled labor-endowed.
The simulations demonstrate a number of things. Early industrialization in both regions
was unskilled labor intensive (O’Rourke et al 2008). Trade between the two regions generates
some income convergence early on - specialization induces the North to devote R&D resources
to the skill-intensive sector and the South to devote resources to the unskilled-intensive sector.
Because the skilled sector is so much smaller than the unskilled sector, the South is able to grow
relatively faster at first. But the dynamic effects of these growth paths (through fertility and
education changes) ultimately reverses income convergence. The reinforcing interactions between
14
Figure 4: Optimal Fertility Rates for High and Low τIndividuals (for given wland wh)
Optimal Fertility Rates for High-τ Individual
Optimal Fertility Rates for Low-τ Individual
fertility nl*
nh*
Costs for
Unskille
d
Earnings for
Unskille
d
Costs and
Benefits
Earnings for
Skille
d
Costs for
Skille
d
fertility nl*
nh*
Costs for
Unskille
d
Earnings for
Unskille
d
Costs and
Benefits
Earnings for
Skille
d
Costs for
Skille
d
15
technological growth and intercontinental commerce help produce dramatic divergence between
the incomes of northern and southern economies.
To simulate this tale however, we will first need to endogenize the time paths of technologies
and trade volumes.
5.1 A Dynamic Model - The Evolution of Technology and Trade
How do technologies grow in each region? Recall that equations (18) - (20) describe one-
period revenues for innovation. There must also be some resource costs to research. For this,
we can assume that these costs are rising in N(“applied” knowledge, specific to each sector
and to each country), and falling in some measure of “general” knowledge, given by B(basic
knowledge, common across all sectors and countries). Thus, a no-arbitrage (free entry) condition
for potential researchers in each region can be described as:
πi≤cNi
B(33)
Specifically, we can assume the following functional form for these research costs:
cNi
B=Ni,t+1
Btν
(34)
for i= 1,3 (for convenience we assume no research occurs in sector 2. This way technological
growth is unambiguously factor-biased), and ν > 0. Given some level of basic knowledge (which
we can assume grows at some exogenous rate) and number of existing machines, we can determine
the resource costs of research. When basic knowledge is low relative to the number of available
machine-types used in sector i, the costs of inventing a new machine in sector iis high (see
O’Rourke et al. 2008 for a fuller discussion). Thus from (33) and (18) - (20) we see that
innovation in sector ibecomes more attractive when basic knowledge is large, when the number
of machine-types in sector iis low, when then price of good iis high, and when the employment
in sector iis high.9
Note that if πi> c(Ni/B), there are potential profits from research in sector i. However, this
will induce research activity, increasing the number of new machines, and hence costs of research,
up. We assume in fact that Niadjusts upward such that costs of research just offset the revenues
of new machine production. Thus increases in Bare matched by increases in levels of Nisuch
that the no-arbitrage condition holds with equality whenever technological growth in the sector
occurs.
9For ease of analysis we assume that the numbers of blueprints in each region are independent of each other.
We relax this assumption in section 5.4.
16
We must also specify how trade technologies evolve. Here we use an amended version of (1),
where production for each region is given by
Yn=α
2(yn
1+aZ1)σ−1
σ+ (1 −α) (yn
2)σ−1
σ+α
2(yn
3−Z3)σ−1
σσ
σ−1(35)
Ys=α
2(ys
1−Z1)σ−1
σ+ (1 −α) (ys
2)σ−1
σ+α
2(ys
3+aZ3)σ−1
σσ
σ−1(36)
Z1is the amount of good 1 that is exported by the South, Z3is the amount of good 3 that is
exported by the North, and 0 < a < 1 is an iceberg factor for traded goods (i.e. the proportion
of exports not lost in transit). Thus the North imports only fraction aof southern exports, and
the South imports only fraction aof northern exports.10 Intermediate goods production is still
described by (2) - (4). To capture improvements in transport technologies over the course of
the 18th and 19th centuries, we simply have agrow exogenously each time period, such that it
reaches the limiting value of 1 by the end of the simulation.
Note that we assume that there is no trade in y2- because this is produced using both Land
H, differences in p2are very small between the North and the South, and thus the assumption
is not very restrictive or important.11 12
5.2 Evolution of the World Economy
General equilibrium is a thirty-six equation system that, given changes in the number of
machine blueprints and the iceberg costs, solves for prices, wages, fertility, education, labor-
types, intermediate goods, employment, trade, and sectoral productivity levels for both the
North and the South. We impose only one parameter difference between the two regions - bn<
bs(this means that there is a bigger range of resource costs for education in the South, so that
the South begins with relatively more unskilled labor than skilled labor). All other parameters
are the same in both regions. Fertility is normalized to one in the beginning for each country, so
that population is stable. The equilibrium is described in more detail in the appendix.
Because the model contains so many moving parts, we can only solve for general equilibrium
numerically. Specifically, we assume that both basic technology (Bin eq 34) and trade technology
10The case where the North specializes in and exports the unskilled-intensive good and the South specializes
in and exports the skilled-intensive good is ruled out due to the North’s relative abundance of skilled labor. The
North would have to have very high levels of unskilled-biased technology compared to the South to reverse its
comparative advantage in skill-intensive production.
11Indeed, trade in all three goods would produce an analytical problem. It is well known among trade economists
that when there are more traded goods than factors of production, country-specific production levels, and hence
trade volumes, are indeterminate. See Melvin (1968) for a thorough discussion.
12One can conceive of y2as the technologically-stagnant and non-tradeable “service” sector. Thus each labor-
type can work either in manufacturing or in services.
17
(ain eqs 35 and 36) start low enough so that neither technological progress nor trade are possible.
We allow however for exogenous growth in basic knowledge and trade technologies, and solve
for the endogenous variables each period. Let us first summarize the evolution of these two
economies with a few propositions, starting with the nature of early industrialization in the
world.
Proposition 1 If N1=N3,L > H , and σ > 1, initial technological growth will be unskilled-labor
biased.
From (18)-(20) we can see that revenues from innovation rise both in the price of the intermediate
good (the “price effect”) and in the scale of sectoral employment (the “market-size effect). If
intermediate goods are grossly substitutable, market-size effects will outweigh price effects (see
Acemoglu 2002 for more discussion of this).
Thus as basic knowledge exogenously grows, sector 1 will be the first to modernize. The
logical implication of this is that early industrialization around the world (provided there are
intellectual property rights in these countries) will be unskilled labor intensive (O’Rourke et al.
2008).
Proposition 2 If pn
3
ps
3·ps
1
pn
1> a2,Z1=Z3= 0.
If transport costs are large (that is, if ais small) relative to cross-country price differences, no
trade occurs. As mentioned above, we will assume that early on transport technologies are not
advanced enough to permit trade. That is, with a small value of a, each country can produce
more under autarky than by trading. Once areaches this threshold level, trade becomes possible,
and further increases in aallows Z1and Z3to rise as well.
Proposition 3 For certain ranges of factors and technologies, the trade equilibrium implies that
ys
3= 0. For other ranges of technologies and factors, the trade equilibrium implies that yn
1= 0.
As trade technologies improve, economies specialize more and more. And divergent technological
growth paths can help reinforce this specialization. There is indeed a point where the North no
longer needs to produce any y1(they just import it from the South), and the South no longer needs
to produce any y3(they just import it from the North). This case we will call the “specialized
trade equilibrium” (described in more detail in the Appendix).
Both trade and technological changes will change factor payments. The final proposition
states how these changes can affect the factors of production themselves.
Proposition 4 If φ > 1, any increase in wl(keeping whconstant) will induce a decrease in e
and an increase in n; furthermore, so long as φis “big enough,” any increase in wh(keeping wl
constant) will induce an increase in eand a decrease in n.
18
Proof.
Substituting our expressions for n∗
land n∗
h, given by (26) and (27), into our expression for τ∗,
given by (28), and rearranging terms a bit, we get the following expression:
τ∗= (wh−wl)−wlλ1
1−φφ1
1−φ−φφ
1−φ+wl φ
φ−1wh 1
1−φλ1
1−φφ1
1−φ−φφ
1−φ
First we must have the condition ∂ τ ∗
∂wl<0 hold. Solving for this and rearranging yields
wl
wh1
φ−1
<1 + 1
λ1
1−φφ1
1−φ−φφ
1−φ
Since the inverse of the skill-premium is always less than one, this expression always holds for
any φ > 1. Next we show what condition must hold in order to have the expression ∂τ ∗
∂wh>0 be
true. Solving and rearranging gives us
λ1
φφ > wl
wh
Thus for a given value of λ,φneeds to be large enough for this condition to hold. Finally, our
expression for total fertility, (31), can be slightly rearranged as
n=n∗
l+ (n∗
h−n∗
l)τ∗
b
From (26) and (27) we know that the second term is always negative, and that n∗
lis constant.
So any increase in education from wage changes will lower aggregate fertility, and any decrease
in education from wage changes will increase aggregate fertility.
5.3 Simulations13
Here we simulate the model described above to analyze the potential sources of North-South
divergence in history. Basic knowledge Band trade technology aare set such that neither
technological growth nor trade is possible at first; each however exogenously rises over time. We
run the simulation for 100 time periods to roughly capture major economic trends from around
1700 to the turn of the twentieth century.
13The parameter values used in the simulations are as follows: σ= 3, α= 0.5, γ= 0.5, λ= 0.5, φ= 10, ν= 2.
These values ensure that Propositions 1 and 4 hold - beyond that, our qualitative findings are not sensitive to
specific parameter values. We also set bn= 2, bs= 6, and pop = 2; this gives us initial factor endowments of
Ln= 3.14, Ls= 3.48, Hn= 0.86, Hs= 0.52. Initial machine blueprints for both countries are set to be N1= 10,
N2= 15, N3= 10; initial trade technology is set to be a= 0.85, and grows linearly such that a= 1 100 periods
later; initial Bis set high enough so that growth in at least one sector is possible early in the simulation; Bgrows
2 percent each time period.
19
Figure 5: The Market for Technologies in the North
10 20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
3
3.5 x 10
-3
Market for Unskilled-Bias Technologies (sector 1) in the North
value
cost
10 20 30 40 50 60 70 80 90 100
1
1.5
2
2.5
3
3.5 x 10
-3
time
Market for Skilled-Bias Technologies (sector 3) in the North
value
cost
Note that because Ln > Hn (combined with the fact that σ>1), the revenues of new research in the unskilled
sector is initially larger than those for the skilled sector. Consequently, unskilled knowledge is the first to
grow (at t = 3). Trade (which becomes possible at t = 20) spurs innovation in the skilled sector since it raises
both the price and employment level in sector 3, pulling research effort away from sector 1.
20
Figure 6: The Market for Technologies in the South
10 20 30 40 50 60 70 80 90 100
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5 x 10
-3
Market for Unskilled-Bias Technologies (sector 1) in the South
value
cost
10 20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
3
3.5
4x 10
-3
time
Market for Skilled-Bias Technologies (sector 3) in the South
value
cost
Just like the North, Ls > Hs, so unskilled technologies grow early in the simulation for the South as well. But
unlike the North, trade reinforces this unskilled intensive growth. Trade also destroys the South’s potential
to develop skill- intensive technologies, since it helps eliminate sector 3 altogether! Consequently the South
never obtains “balanced” technological growth.
21
Figure 7: Factor Productivities
10 20 30 40 50 60 70 80 90 100
0
1
2
3
4
5
6
7
8
9
Total Factor Productivities in the North
A1
A2
A3
10 20 30 40 50 60 70 80 90 100
0
1
2
3
4
5
6
7
Total Factor Productivities in the South
time
A1
A2
A3
22
Figures 5 through 7 summarize the evolution of technologies in both regions. In the beginning
the costs of research are prohibitively high everywhere, so technologies are stagnant. But growth
in basic knowledge allows us to see the implications of Proposition 1 - because there is a greater
abundance of unskilled labor relative to skilled labor in both the North and the South, the costs
of research first catch up to revenues in sector 1 in both regions.
This growth in unskilled labor intensive technologies lowers the relative returns to unskilled
labor in both regions, inciting fertility increases and educational decreases (Proposition 4). We
can see this manifest itself in the North by the increases in the revenues generated by innovation
- as population rises in the North, the market-size effects caused by fertility increases raises the
value of such innovation. Still, because skilled labor remains in relative scarce supply, the cost
of innovation exceeds the benefits in sector 3 for the beginning of the simulation.
But evidently, looking at these figures, some event happens at t= 20 that in the North seems
to pull resources away from sector 1 (so that potential revenues to research in the sector fall)
and push resources into sector 3 (so that potential revenues in the sector rise). The opposite
seems to happen in the South - resources are pulled from sector 3 to sector 1. Soon after this the
North begins to develop new machine blueprints for skilled labor intensive sector 3. The South
on the other hand increases its development of unskilled-intensive production, while abandoning
its production of the sector 3 good altogether. What happened?
The answer is that at t= 20 the trade technology parameter abecomes large enough so
that commerce between the two regions becomes possible (Proposition 2). At this point the
South starts exporting some of its production of y1and the North starts exporting some of its
production of y3. Once such trade occurs, both price and market-size effects rise in sector 3, and
innovation in the sector in the North begins.
The opposite happens in the South. Producing very little of good 3 even in autarky, the South
finds itself importing all of the good from the North once the North raises its productivity in
the sector (Proposition 3). Of course, this ultimately means that it will not be able to produce
any skill-intensive innovations even with high Bvalues (note that in the absence of trade such
technological growth would have been possible at around t≈50).
Figures 8 and 9 chart the evolution of fertility, education, trade, and incomes per person in
both regions. Early industrialization without any trade is characterized by rising fertility and
falling education. Initial trade generates the demographic divergence that we would expect - we
see continued increases in fertility and declines in education in the South, and a demographic
transition of falling fertility and rising education in the North. This however produces some slight
convergence in incomes per capita between the two regions. The reason, as mentioned earlier,
is that sector 3 is smaller than sector 1 in both economies. Sector 3 suddenly becomes the
modernizing sector in the North, while sector 1 continues to grow technologically in the South.
Through scale effects, the South benefits more from this specialization; the growing sector in the
23
Figure 8: Rates of Fertility and Education
10 20 30 40 50 60 70 80 90
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1
1.01
1.02
1.03
Fertility
Northern Fertility
Southern Fertility
010 20 30 40 50 60 70 80 90
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
time
Education
Northern Educatio n
Southern Education
24
Figure 9: Trade and Output
10 20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Trade Volumes
Fraction of Export That Makes it as Import
(exogenous linear growth)
Trade of x1 (from South to North)
Trade of x3 (from North to South)
10 20 30 40 50 60 70 80 90 100
1
1.5
2
2.5
3
3.5
4
4.5
Output per Person
Northern Output per Person
Southern Output per Person
10 20 30 40 50 60 70 80 90 100
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
Relative Income (yn/ys)
time
25
North on the other hand is fairly small and cannot generate as much aggregate growth.
This trend towards convergence ultimately reverses with continued growth in trade and tech-
nologies. Specialization becomes so pronounced that the South quickly abandons production of
good 3 altogether. This seals the South’s fate forever as an unskilled-intensive producer. No
matter how large basic knowledge grows, it can never generate new skill-intensive technologies,
for the skill-intensive sector no longer exists for them. And the North eventually abandons its
production of the unskilled good, reinforcing even further the South’s energies in the production
of this good.
Notice in figure 9 that once specialized trade occurs, the South’s terms of trade begins to
rapidly deteriorate - the South has to give up more and more y1for the same amount of y3.
Why is this? Because of its rapid population growth, the South increasingly makes up a larger
share of the world population. It thus begins to flood the market with its primary products.
The skill-intensive product on the other hand becomes relatively more scarce, and thus fetches
a higher and higher price. Of course, all the y1that the South produces in order to acquire
increasingly precious quantities of y3raises fertility even more, further eating away at per capita
incomes.
Ultimately the increasing ability to trade mixed with biased technological growths in each
country foster the divergence. There is rapid fertility declines in the North due to its rising
production of good 3; the South on the other hand retains its high fertility due to the low skill
premia generated by its specialization of good 1. The demographic transition of the North, and
the lack thereof in the South, ultimately creates a great divergence in output per person between
the two regions.14
5.4 Some Counterfactuals
The above simulation stresses that both trade and technological forces interacting together
created the divergence in living standards around the world. Here we perform some counter-
factual experiments to further support this idea. We ask, what if we have an endogenous biased
technological growth model, but do not allow for any trade between the North and the South?
On the other hand, what if transport costs fall (just like our main simulation) so that trade
eventually becomes possible, but technological progress can never occur?
Figure 10 displays the simulation results of these exercises. We can see that each factor
by itself creates very modest divergence in living standards (measured by changes in yn/ys).
Trade alone can not generate such divergence; the South’s ability to develop unskilled-intensive
technologies allows its absolute income level to grow, but through even more rapid population
14In fact, the demographic transition is so dramatic in the North that the shrinking population induces a drop
in the scale of the market for inventors, slowing down Northern technological growth. In reality, immigration and
other factors kept populations in northern countries from dropping so dramatically.
26
Figure 10: Simulations without Technology-Trade Interaction
No Trade
050 100
2
3
4
5
6x 10
-3
Sector 1 Tech (North)
val ue
cost
050 100
1
2
3
4x 10
-3
Sector 3 Tech (North)
val ue
cost
050 100
0
5
10 Factor P roductivities (North)
A1
A2
A3
050 100
3
3.5
4
4.5
5x 10
-3
Sector 1 Tech (South)
val ue
cost
050 100
0
1
2
3
4x 10
-3
Sector 3 Tech (South)
val ue
cost
050 100
0
2
4
6
8Factor Productivities (South)
A1
A2
A3
050 100
1
1.01
1.02
1.03 Fertility
Northern Ferti lity
Southern Fert ility
050 100
0.2
0.3
0.4
0.5 Education
Northern Educ ation
Southern Education
050 100
1.15
1.2
1.25
1.3 Relative Income (yn/ys)
050 100
0
0.5
1
1.5 Trade Volumes
Fraction of Export That Makes it as Import
Trade of x1
Trade of x3
No Technological Growth
050 100
0
0.1
0.2
0.3
0.4 Sector 1 Tech (North)
val ue
cost
050 100
0
0.1
0.2
0.3
0.4 Sect or 3 Tech (North)
val ue
cost
050 100
1
2
3
4Factor P roductivities (North)
A1
A2
A3
050 100
0
0.1
0.2
0.3
0.4 Sec tor 1 Tech (Sout h)
val ue
cost
050 100
0
0.1
0.2
0.3
0.4 Sector 3 Tech (South)
val ue
cost
050 100
0
1
2
3
4Factor Productivities (South)
A1
A2
A3
050 100
0.99
0.995
1
1.005 Fertility
Northern Ferti lity
Southern Fertilit y
050 100
0.2
0.4
0.6
0.8 Education
Northern Educ ation
Southern Education
050 100
1.2
1.22
1.24
1.26 Relative Income (yn/ys)
050 100
0
0.5
1Trade Volumes
Fraction of Export That M akes it as Import
Trade of x1
Trade of x3
27
Figure 11: Simulations with Perfect Northern Technological Diffusion to the South
Technological Growth in North with Perfect Diffusion to the South; Regular Trade
020 40 60 80 100
0
1
2
3
4x 10
-3
Sector 1 Tech (North)
val u e
cost
020 40 60 80 100
1
2
3
4x 10
-3
Sect or 3 Tech (North)
val ue
cost
020 40 60 80 100
0
2
4
6
8Factor Produc tivities (North)
A1
A2
A3
020 40 60 80 100
0
2
4
6Factor Productivities (South)
A1
A2
A3
020 40 60 80 100
0.96
0.98
1
1.02
1.04 Fertility
Northern Fert ility
Southern Fertility
020 40 60 80 100
0.2
0.4
0.6
0.8
1Educati on
Northern Education
Southern Education
020 40 60 80 100
1.3
1.4
1.5
1.6 Relative Income (yn/ys)
020 40 60 80 100
0
0.5
1
1.5
2Trade Volumes
Fraction of Ex port That Makes it as Import
Trade of x1
Trade of x3
Technological Growth in North with Perfect Diffusion to the South; No Trade
020 40 60 80 100
2
3
4
5
6x 10
-3
Sector 1 Tech (North)
val ue
cost
020 40 60 80 100
1
2
3
4x 10
-3
Sector 3 Tech (North)
val ue
cost
020 40 60 80 100
0
5
10 Factor Productivities (North)
A1
A2
A3
020 40 60 80 100
0
5
10 Factor Productivities (South)
A1
A2
A3
020 40 60 80 100
1
1.01
1.02
1.03 Fertility
Northern Fertility
Southern Fertility
020 40 60 80 100
0.2
0.3
0.4
0.5 Educ ation
Northern Educ ation
Southern Education
020 40 60 80 100
1.1
1.15
1.2
1.25 Relative Income (yn/ys)
020 40 60 80 100
0
0.5
1
1.5 Trade Volumes
Fracti on of Export That M akes it as Import
Trade of x1
Trade of x3
28
growth keeps its income per capita level anchored. Thus unlike the GM approach that has the
South specialize in the inherently technologically-lagging sector, we generate a divergence even
as the South endogenously experiences rapid technological progress.
Another hypothetical - what if the South could not develop technologies themselves, but
instead relied on the diffusion of technologies from the North? Some could argue that this is a
more realistic scenario, since the new technologies of the Industrial Revolution could be exported
mechanically with relative ease to most of the world. After all, while developing new knowledge
was an arduous task, copying this knowledge was much easier. This was particularly true of the
technologies of early industrialization; since they were not very sophisticated, they were quickly
transmitted to, and easily adopted by, much of the world (Mokyr 1999; Clark 2007, Chapter 15).
Figure 11 displays the results of simulations where the South instantly inherits the machine
blueprints of the North (where Nn
1,t =Ns
1,t,Nn
2,t =Ns
2,t, and Nn
3,t =Ns
3,t ∀t). Again, we can see
that the interaction between trade and technological developments creates the divergence (top
figure); technological diffusion with no trade allowed actually generates convergence (bottom
figure). Divergence in this case happens for a slightly different reason - the North develops skill-
intensive technologies that the South can not use (it stops producing the skill-intensive good early
in the simulation). Technological diffusion in this case is “inappropriate” - the receiving country
does not have the appropriate factors needed to exploit such knowledge (see Basu and Weil 1998;
Acemoglu and Zilibotti 2001). Notice however that without trade, the North never switches
to skill-intensive production; consequently technological diffusion is always of the “appropriate”
variety for the South (that is, unskilled intensive), and so income convergence occurs. Once
again, we see trade is important in the story of divergence, but it is its interaction with biased
technological developments that truly ripped open the chasm in living standards around the
world.
References
Acemoglu, Daron. 2002. “Directed Technical Change,” Review of Economic Studies 69: 781-809.
Acemoglu, Daron. forthcoming. “When Does Labor Scarcity Encourage Innovation?” Journal of
Political Economy.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2001. “The Colonial Origins of Comparative
Development: An Empirical Investigation.” American Economic Review, 91(5): 1369-1401.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2005. “The Rise of Europe: Atlantic Trade,
Institutional Change, and Economic Growth.” American Economic Review, 95(3): 546-79.
Acemoglu, Daron and Fabrizio Zilibotti. 2001. “Productivity Differences,” Quarterly Journal of Eco-
nomics 116: 563-606.
Aghion, Philippe and Peter Howitt. 1992. “A Model of Growth through Creative Destruction,” Econo-
29
metrica, 60: (323-351).
Basu, Susanto and David N. Weil. 1998. “Appropriate Technology and Growth,” Quarterly Journal of
Economics 113: 1025-1054.
Bekar, Clive and Richard Lipsey. 1997. “A Structuralist View of Technical Change and Economic
Growth.” Reprinted in Microeconomics, Growth and Political Economy: The Selected Essays of
Richard Lipsey Vol. One. Edward Elgar: Cheltenham, UK.
Clark, Gregory. 2007. “A Review of Avner Grief ’s Institutions and the Path to the Modern Economy:
Lessons from Medieval Trade.” Journal of Economic Literature, XLV: 727-743.
Clark, Gregory. 2007. A Farewell to Alms: A Brief Economic History of the World. Princeton and
Oxford: Princeton University Press.
Clark, Gregory, and Robert C. Feenstra. 2001. “Technology in the Great Divergence,” in NBER book
Globalization in Historical Perspective.
Clark, Gregory and Susan Wolcott. 2003. “One Polity, Many Countries: Economic Growth in India,
1873-2000.” in Dani Rodrik (ed.), Frontiers of Economic Growth. Princeton: Princeton University
Press
Crafts, N.F.R., and C.K. Harley. 1992. “Output Growth and the Industrial Revolution: A Restatement
of the Crafts-Harley View,” Economic History Review 45(4): 703-30.
Findlay, Ronald. 1996. “Modeling Global Interdependence: Centers, Peripheries, and Frontiers,” Amer-
ican Economic Review 86(2): 47-51.
Galor, Oded. 2005. “From Stagnation to Growth: Unified Growth Theory,” in Handbook of Economic
Growth, Vol. 1, part 1, edited by P. Aghion and S.N. Durlauf, pp. 171-293. Amsterdam: North
Holland.
Galor, Oded, and Andrew Mountford. 2002. “Why are one Third of People Indian and Chinese? Trade,
Industrialization and Demographic Transition,” manuscript.
Galor, Oded and Andrew Mountford. 2006. “Trade and the Great Divergence: The Family Connection,”
American Economic Review Papers and Proceedings, 96: 299-303.
Galor, Oded and Andrew Mountford. 2008. “Trade and the Great Divergence: Theory and Evi-
dence,”Review of Economic Studies, 75: 1143-1179.
Grossman, Gene M. and Elhanan Helpmann. 1991. Innovation and Growth in the Global Economy.
Cambridge, MIT Press.
Howitt, Peter. 1999. “Steady Endogenous Growth with Population and RD Inputs Growing,” Journal
of Political Economy, 107: 715-730.
Jones, Charles I. 1995. “R&D-Based Models of Economic Growth,” Journal of Political Economics,
103: 759-784.
Melvin, James R. 1968. “Production and Trade with Two Factors and Three Goods,” American Eco-
nomic Review 58: 1249-1268.
30
Mitchener, Kris J., and Se Yan. 2010. “Globalization, Trade and Wages: What Does History Tell Us
about China?” NBER working paper no. 15679.
Mokyr, Joel. 1999. The British Industrial Revolution: An Economic Perspective, 2nd edition. Westview
Press: Boulder, CO.
Mokyr, Joel. 2002. The Gifts of Athena. Princeton, N.J.: Princeton University Press.
North, Douglass C., and Robert Paul Thomas. 1973. The Rise of the Western World: A New Economic
History. Cambridge: Cambridge University Press.
O’Rourke, Kevin H., Ahmed S. Rahman and Alan M. Taylor. 2008. “Luddites and the Demographic
Transition,” NBER working paper no. 14484.
ORourke, Kevin H., Alan M. Taylor and J. G. Williamson. 1996. “Factor Price Convergence in the
Late Nineteenth Century,” International Economic Review 37: 499-530.
O’Rourke, Kevin H. and Jeffrey G. Williamson. 1994. “Late 19th Century Anglo-American Factor
Price Convergence: Were Heckscher and Ohlin Right?” Journal of Economic History 54: 892-916.
O’Rourke, Kevin H. and Jeffrey G. Williamson. 1999. Globalization and History. Cambridge, MA.:
MIT Press.
Pomeranz, Kenneth. 2000. The Great Divergence: China, Europe, and the Making of the Modern
World Economy. Princeton Economic History of the Western World series. Princeton and Oxford:
Princeton University Press.
Rahman, Ahmed S. 2009. “The Road Not Taken - What Is The ‘Appropriate’ Path to Development
When Growth Is Unbalanced,” United States Naval Academy working paper no. 26.
Romer, Paul M. 1986. “Increasing Returns and Long-Run Growth,” Journal of Political Economy, 94:
1002-1037.
Romer, Paul M. 1990. “Endogenous Technological Change,” Journal of Political Economy 98(5): S71-
S102.
Segerstrom, Paul S., T. C. A. Anant, and Elias Dinopoulos. 1990. “A Schumpeterian Model of the
Product Life Cycle,” American Economic Review, 80: 1077-1092.
Temin, Peter. 1997. “Two Views of the British Industrial Revolution,” Journal of Economic History
57(1): 63-82.
Young, Alwyn. 1998. “Growth without Scale Effects.” Journal of Political Economy, 106: 41-63.
31
A Diversified Trade Equilibrium
With trade of goods y1and y3between the North and the South, productions in each region are
given by (35) and (36).
For each region c∈n, s, the following conditions characterize the diversified trade equilibrium.
ps
1=ws
l
As
1
(37)
pc
2=1
Ac
2(wc
l)γ(wc
h)1−γ(1 −γ)γ−1γ−γ(38)
pn
3=wn
h
An
3
(39)
1
Ac
1yc
1+1
Ac
2(wc
l)γ−1(wc
h)1−γ(1 −γ)γ−1γ1−γyc
2=Lc(40)
1
Ac
2(wc
l)γ(wc
h)−γ(1 −γ)γγ−γyc
2+1
Ac
3yc
3=Hc(41)
yn
1+a1Z1= α
2σ(pn
1)−σ
α
2σ(pn
1)1−σ+ (1 −α)σ(pn
2)1−σ+α
2σ(pn
3)1−σ!·Yn(42)
ys
1−Z1= α
2σ(ps
1)−σ
α
2σ(ps
1)1−σ+ (1 −α)σ(ps
2)1−σ+α
2σ(ps
3)1−σ!·Ys(43)
yc
2= (1 −α)σ(pc
2)−σ
α
2σ(pc
1)1−σ+ (1 −α)σ(pc
2)1−σ+α
2σ(pc
3)1−σ!·Yc(44)
yn
3−Z3= α
2σ(pn
3)−σ
α
2σ(pn
1)1−σ+ (1 −α)σ(pn
2)1−σ+α
2σ(pn
3)1−σ!·Yn(45)
ys
3+a3Z3= α
2σ(ps
3)−σ
α
2σ(ps
1)1−σ+ (1 −α)σ(ps
2)1−σ+α
2σ(ps
3)1−σ!·Ys(46)
An
1(An
1Ln
1+a1Z1)−1
σ=2(1 −α)γ
αAn
2
σ−1
σ(Ln−Ln
1)−γ−σ+σγ (Hn−Hn
3)γ+σ−σγ−1(47)
An
3(An
3Hn
3−Z3)−1
σ=2(1 −α)(1 −γ)
αAn
2
σ−1
σ(Ln−Ln
1)−γ+σγ (Hn−Hn
3)γ−σγ−1(48)
32
As
1(As
1Ls
1−Z1)−1
σ=2(1 −α)γ
αAs
2
σ−1
σ(Ls−Ls
1)−γ−σ+σγ (Hs−Hs
3)γ+σ−σγ−1(49)
As
3(As
3Hs
3+a3Z3)−1
σ=2(1 −α)(1 −γ)
αAs
2
σ−1
σ(Ls−Ls
1)−γ+σγ (Hs−Hs
3)γ−σγ−1(50)
Ac
1=Nc
1,t−1+αα
1−αNc
1,t −Nc
1,t−1(αpc
1)α
1−α(51)
Ac
2=Nc
2,t−1+αα
1−αNc
2,t −Nc
2,t−1(αpc
2)α
1−α(52)
Ac
3=Nc
3,t−1+αα
1−αNc
3,t −Nc
3,t−1(αpc
3)α
1−α(53)
Hc=τ∗c
bcpopc(54)
Lc=1−τ∗c
bcpopc+ncpopc(55)
nc=1−τ∗c
bcn∗c
l+τ∗c
bcn∗c
h(56)
ec=τ∗c
bc(57)
pn
1
pn
3
=Z3
aZ1
(58)
ps
1
ps
3
=aZ3
Z1
(59)
Equations (37) - (39) are unit cost functions, (40) and (41) are full employment conditions, (42) -
(46) denote regional goods clearance conditions, (47) - (50) equate the marginal products of raw factors,
(51) - (53) describe sector-specific technologies, , (54) - (63) describe fertility, education and labor-types
for each region, and (64) and (65) describe the balance of payments for each region. Solving this system
for the unknowns pn
1,ps
1,pn
2,ps
2,pn
3,ps
3,yn
1,ys
1,yn
2,ys
2,yn
3,ys
3,wn
l,ws
l,wn
h,ws
h,Ln
1,Ls
1,Hn
3,Hs
3,An
1,
An
2,An
3,As
1,As
2,As
3,Ln,Ls,Hn,Hs,nn,ns,en,es,Z1and Z3constitutes the static partial trade
equilibrium.
Population growth for each region is given simply by
popc
t=nc
t−1popc
t−1
33
Each region will produce all three goods so long as factors and technologies are “similar enough.”
If factors of production or technological levels sufficiently differ, the North produces only goods 2 and
3, while the South produces only goods 1 and 2. No other specialization scenario is possible for the
following reasons: first, given that both the North and South have positive levels of Land H, full
employment of resources implies that they cannot specialize completely in good 1 or good 3. Second,
specialization solely in good 2 is not possible either, since a region with a comparative advantage in
this good would also have a comparative advantage in either of the other goods. This implies that
each country must produce at least two goods. Further, in such a scenario we cannot have one region
producing goods 1 and 3: with different factor prices across regions, a region cannot have a comparative
advantage in the production of both of these goods, regardless of the technological differences between
the two regions. See Cunat and Maffezzoli (2002) for a fuller discussion.
B Specialized Trade Equilibrium
The specialized equilibrium is one where the North does not produce any good 1 and the South does
not produce any good 3. Productions in each region are then given by
Yn=α
2(aZ1)σ−1
σ+ (1 −α) (yn
2)σ−1
σ+α
2(yn
3−Z3)σ−1
σσ
σ−1(60)
Ys=α
2(ys
1−Z1)σ−1
σ+ (1 −α) (ys
2)σ−1
σ+α
2(aZ3)σ−1
σσ
σ−1(61)
Once again, we do not permit any trade of good 2. For each region c∈n, s, the following conditions
characterize this equilibrium.
ps
1=ws
l
As
1
(62)
pc
2=1
Ac
2(wc
l)γ(wc
h)1−γ(1 −γ)γ−1γ−γ(63)
pn
3=wn
h
An
3
(64)
1
An
2(wn
l)γ−1(wn
h)1−γ(1 −γ)γ−1γ1−γyn
2=Ln(65)
1
An
2(wn
l)γ(wn
h)−γ(1 −γ)γγ−γyn
2+1
An
3yn
3=Hn(66)
1
As
1ys
1+1
As
2(ws
l)γ−1(ws
h)1−γ(1 −γ)γ−1γ1−γys
2=Ls(67)
34
1
As
2(ws
l)γ(ws
h)−γ(1 −γ)γγ−γys
2=Hs(68)
a1Z1= α
2σ(pn
1)−σ
α
2σ(pn
1)1−σ+ (1 −α)σ(pn
2)1−σ+α
2σ(pn
3)1−σ!·Yn(69)
ys
1−Z1= α
2σ(ps
1)−σ
α
2σ(ps
1)1−σ+ (1 −α)σ(ps
2)1−σ+α
2σ(ps
3)1−σ!·Ys(70)
yc
2= (1 −α)σ(pc
2)−σ
α
2σ(pc
1)1−σ+ (1 −α)σ(pc
2)1−σ+α
2σ(pc
3)1−σ!·Yc(71)
yn
3−Z3= α
2σ(pn
3)−σ
α
2σ(pn
1)1−σ+ (1 −α)σ(pn
2)1−σ+α
2σ(pn
3)1−σ!·Yn(72)
a3Z3= α
2σ(ps
3)−σ
α
2σ(ps
1)1−σ+ (1 −α)σ(ps
2)1−σ+α
2σ(ps
3)1−σ!·Ys(73)
An
3(An
3Hn
3−Z3)−1
σ=2(1 −α)(1 −γ)
αAn
2
σ−1
σ(Ln)−γ+σγ (Hn−Hn
3)γ−σγ−1(74)
As
1(As
1Ls
1−Z1)−1
σ=2(1 −α)γ
αAs
2
σ−1
σ(Ls−Ls
1)−γ−σ+σγ (Hs)γ+σ−σγ −1(75)
As
1=Ns
1,t−1+αα
1−αNs
1,t −Ns
1,t−1(αps
1)α
1−α(76)
Ac
2=Nc
2,t−1+αα
1−αNc
2,t −Nc
2,t−1(αpc
2)α
1−α(77)
An
3=Nn
3,t−1+αα
1−αNn
3,t −Nn
3,t−1(αpc
n)α
1−α(78)
Hc=τ∗c
bcpopc(79)
Lc=1−τ∗c
bcpopc+ncpopc(80)
nc=1−τ∗c
bcn∗c
l+τ∗c
bcn∗c
h(81)
35
ec=τ∗c
bc(82)
pn
1
pn
3
=Z3
aZ1
(83)
ps
1
ps
3
=aZ3
Z1
(84)
36
