Lecture Notes on International Trade Theory and

Industrial Oganization

Karl-Markus Modén
Södertörns Högskola

April 2, 2012
Introduction
International trade is one of the oldest subjects in economics. At the times
of David Hume, Adam Smith and David Ricardo it was claimed by many
(e.g. the so called ”mercantilists”) that a country should export more than
it imports (and actually discourage imports) and accumulate gold (money)
in the process. That was supposed to make the country richer. The classical
economists understood that it is only what people consume that yields util-
ity. Furthermore, David Hume realized that if a country accumulates gold
its price level will rise (the quantity theory of money) and its export-goods
will become too expensive on the world market and its revenue from exports
will therefore decline. Hume realized that there is an automatic adjustment
process which tends to equilibrate imports and exports (”the price-specie
flow”). Hence, a policy of promoting exports and discouraging imports will
be self-defeating.·
Adam Smith basically agreed with Hume, but he was not quite able to
pin down what determines the direction of trade and to prove that free-
trade is generally beneficial to all participating countries. This task was
accomplished by David Ricardo, who developed the theory of comparative
advantage, which is the starting point of the theory of international trade.
Ricardo’s motivation was that he was opposed to a proposal of introduc-
ing tariffs on imported corn. He developed his theory to show that free,
and voluntary, trade is beneficial to both countries involved in the (bilat-
eral) trade. However, the proponents of the tariffs were not foolish (even if
their arguments were faulty), they were generally members of the ”landed
classes” in England, which benefitted from a higher domestic prices of corn,
which would ensue if tariffs were imposed. The losers were the consumers
of bread, which was the common people. The latter did not have any polit-
ical representation so they were not able to oppose the introduction of the
tariffs. Ricardo became their ”champion”. The so called Corn Laws were
repealed in 1846, but this was not because of Ricardo’s efforts, but rather
that the emerging class of industrialists had gained some political influence.
The industrialists benefited from the repeal of the Corn Laws because they
mainly sold their goods to the common people, and they were interested in
improving the real purchasing power of the latter.
This lesson from this historical episode is still important today. We may
in fact start our study of this subject by posing the question: ”What is
different with international trade, as compared to the usual trade between
people, or companies, within a country?” One answer to this question is
that countries are defined by artificial barriers, which changes the nature of
trades between countries compared to within countries. One such artificial
barrier is tariffs. But tariffs is not the only thing that the authorities in the
individual countries can dream up in order to change the pattern of interna-
tional trade. They may for example use their legislative powers to make it

1
illegal for domestic citizens to work abroad (or to emigrate or immigrate), or
to transfer capital abroad. They may also make it illegal for foreign citizens
to own capital, or to work as guestworkers, in the country. Many other ac-
tions are open for countries’ authorities, e.g., to subsidize their own import
competing industries, in various ways. The reasons to do these things are
as politically motivated today, as they were in 1846.
There are of course other dimensions of international trade. One such
dimension is the geographical aspect. Countries are often separated by dis-
tance, which make transport costs important. This is of course only a matter
of degree (geographical distance is a continuous variable), so there is con-
ceptually no difference from any other analysis of regional trade. However,
geographical forces may interact with nationalistic policies, so it is important
to analyze economic geography in the context of traditional trade theory.
This combination is something which has received a lot of interest in recent
years.
Our primary aim is to gain an understanding of the forces driving inter-
national trade and integration and how these forces may have changed over
time. A proper starting point is the classical theories of international trade:
David Ricardo’s theory of comparative advantage and Eli Hecksher’s and
Bertil Ohlin’s theory of relative factor proportions to explain comparative
advantage. However, before that we will look at the data to gain a feel for
the current pattern of international trade and how it has changed and is
currently changing.

The pattern of world trade

Who trades with who?
Table 1 shows the pattern of world trade in terms of the share of total value
of exports and imports of goods and services in the world as an average
over the period 2007-2009. The left-most column shows the share of world
exports eminating from a particular region, e.g. 13% from North-America
etc; the uppermost row shows the share of imports to the same regions, e.g.
17% to North America. Each row inside the table shows the destination of
each region’s export, e.g. 50% of North America’s exports end up in North
America (i.e. regional trade between Canada, Mexico and the USA). Europe
is the largest exporter and importer and 73% of its trade is within Europe.

Some interesting facts to note from this table is the large extent of re-
gional trade and that Africa and South & Central America are very de-
pendent on trade with Europe and North America respectively. The latter
type of trade, which goes from the poor southern hemisphere to the rich
northern hemisphere is called ”north − south” trade. However, most world
trade takes place within the rich northern countries (US-Canada and within

2
Table 1: World trade divided into regional origin of exports and destionation
of imports.
17% 4% 43% 3% 3% 4% 25%
Destination→ North S&C Am. Europe CIS Africa Mid. Asia
Origin Am. East
↓
13% North Am 50 8 18 1 2 3 19
4% S & C Am 28 26 20 1 3 2 26
42% Europe 8 1 73 3 3 3 8
4% CIS 5 1 58 20 1 3 13
3% Africa 20 3 40 0 10 3 21
6% Mid. East 10 1 13 1 5 11 53
28% Asia 18 3 18 2 3 4 50

Europe), this is for obvious reasons called ”north − north” trade. This ta-
ble shows a static picture and trade patterns like this do not change very
rapidly. However, a recent trend is the increasing importance of Asia and in
particular China. So far, however, China’s share of world export is "only"
about 7%.

World trade and world economic growth

Figure 1 shows the long-run trend in world export volume, production and
gross domestic product. All three series starts with an index-value of 100
in 1950. World production and GDP follow each other pretty closely, with
average yearly growth rates for both series of 3.6%. The average growth rate
in world trade volume was 5.8%, which is 61% higher than the growth rate
for GDP. Thus the world has sustained a relatively smooth rate of growth
over the last 60 years (at least) and this has gone together with a much higher
growth rate of world trade (exceptions were in 1974 and 2009). Economic
theory suggests that these two phenomena are dependent on each other,
with increasing openness of markets fuelling economic growth and higher
national incomes increasing demand for import goods in all countries. It is
a dual dependency and it is difficult to say which comes first (or "causes"
the other). Anyway, it is a very clear illustration of what has recently been
called the "globalization" phenomenon.

What is traded?
Table 2 shows the composition of world trade in goods. Agricultural goods
and raw material based goods such as oil and minerals have a significant
part of world trade, but nowadays manufacturing goods as a group is much
more important, comprising almost 3/4 of the value of world trade. Within

3
 35 00

30 00

25 00

20 00

Export volume
Production volume

15 00 G DP

10 00

5 00

Figure 1:

the group of manufactured goods, machinery, which include e.g. cars and
trucks, as well as computers and telecommunication equipment, is the most
important sub-group. Furthermore, it is the case that within the group of
manufactured goods, there are two−way trade to a large extent. This means
that the same type of goods are both imported and exported by the same
countries. Since most trade is between rich countries an educated guess is
that rich countries trade similar goods with each other (a guess that turns
out to be true).

The theory of comparative advantage

Ricardo assumed that there were only two countries, we will assume that
these countries are France (F ) and Germany (G) which produce two goods,
cars (C) and wine (W ). These goods are produced with the help of only
labor (L), and with a very simple (linear) production function (QiC = biC ·Li ,
QiW = biW · Li ). The coefficients biC and biW shows how much more output of
each good that can be produced with one more unit of labor in each country
(i = F, G), or simply the marginal product of labor:
∆Qij
= MPLij = bij , i = F, G; j = C, W.
∆L
Note that since the production functions are linear the marginal and average
products are the same:
Qij
= bij ; i = F, G; j = C, W.
L

4
Table 2: World trade divided into main groups of commodities, 2000 and
2005.
Share Annual %
change
2000 2005 2009 2000-2005
Agriculture 8.8 8.4 9.6 9
of which
food 6.9 6.7 10
raw materials 1.9 1.7 7
Fuels and mining 13.7 17.2 18.6 15
of which
ores and minerals 1.1 1.5 17
fuels 10.7 13.8 14.8 16
metals (not iron) 2 2 10
Manufactures 74.9 72 68.6 9
of which
Iron and steel 2.3 3.1 2.7 17
Chemicals 9.3 10.9 11.9 14
Other semi-manufactures 7.2 7 10
Machinery 42 37.9 8
of which
Office and telecom 15.4 12.6 10.9 6
Transport 13.3 12.8 7.0 9
Textiles 2.5 2 1.7 5
Clothing 3.2 2.7 2.6 7
Other 8.5 8.4 10

5
In trade theory it has, since Ricardo’s time, been a tradition to use so
called labor-input coefficients to describe the production possibilities in each
country and sector. These coefficients tells us how many units of labor
that are required to produce one unit of output and is simply equal to the
reciprocals of the bij coefficients, i.e., the average (and marginal) products of
labor. Let’s denote the labor-input coefficients by aij ; which is the number
of labor hours necessary to produce on unit of good j in country i. Let’s
consider the following technology in F and G,

Cars Wine
France 6 (aFC ) 2 (aFW )
Germany 1 (aGC) 1 (aGW)

Note that Germany has an absolute advantage in the production of both

goods, since it takes less labor hours than in France to produce one unit
of each goods. If absolute advantage is a basis for trade where would be
no gains from trade in this example. But if we start from a position with
no trade initially (so called autarchy), with both countries producing both
goods, and assume that international trade becomes possible, we can show
with a reallocation example that specialization according to comparative
advantage, will increase world production.
Assume that F reduces its production of C by one unit; this will free-up
6 hours of labor time, which is reallocated to the wine industry; increasing
W −production by 6/2 = 3 units. If G reduces its wine production by 1 unit
and channel the released labor toward the C− sector, the latter production
will increase by 1 unit. The total changes is summarized in the next table.

∆C ∆W
France -1 +3
Germany +1 -1
F+G (”the World”) 0 +2.

Note that if F sends 2 units of W to G, and G sends one unit of C to G (in

exchange for the 2 units of W ); both F and G have the same amount of C
as before but 1 more W. If utility is strictly increasing in the consumption
of W , both countries are strictly better off.
Note that in the previous example we assumed that the two goods were
exchanged according to the following ratio:
2W
=2
1C
i.e., one car is exchange for two units of wine. This ratio is called the
terms − of − trade.
Note also that we can define ”internal” exchange ratios between C and
W , in each country. These ratios show the opportunity costs of increasing

6
C in terms of W foregone.

aFC 6 aG
C
F
= = 3, =1
aW 2 aG
W

aF aG
Since aFC > aGC , G has a comparative advantage in the production of C,
W W
since the opportunity cost is lower than in F. If we invert these ratio we
would get the opportunity costs of increasing W in terms of C foregone, and
we would conclude that F has a comparative advantage in the production
of W, since the opportunity costs (1/3), is lower than in G (1). Note also
that the terms − of − trade of trade will always lie between the opportunity
costs of the individual countries.
We should also note here that since we assume constant opportunity
costs, i.e., linear production possibility curves, the reallocation argument
can be repeated until all labor in F is employed in the W − industry, and all
labor in G works in the C− industry. We get a so called corner solution, or
complete specialization. The division of labor and gain from specialization
is carried out to its extreme in this case.

Example Assume that France has 600 labor hours and Germany has
500. Their respective production possibility curves (or production possibil-
ity frontiers, PPFs, as we will call them) are illustrated in Figure 2, which
also shows the consumption possibilities with trade.(the dashed lines), and
complete specialization. The gains from trade is evident from the fact that
the consumption possibility curves lie everywhere outside the PPFs, which
are also the consumption possibility curves in autarchy.
W ine W ine

Franc e G erm any

600/a F W 500/a TW
= 300 600 Labor units . = 500 500 Labor units

-2

-3 -2 -1
Cars Cars
600/a G C = 100 500/a G C = 500

Figure 2:

7
The shape of the PPF and supply curves. In Ricardo’s case the
slope of each country’s PPFs is constant. This means, as we’ve seen, that
each additional worker is as productive as the previous one hired. If labor
is the only input into the production process and if each worker is paid
the same wage, w, as would be the case in a perfectly competitive labor
market will homogeneous labor, the total cost of production is equal to
L
T C = w·L. The average cost equals AC = w· Q , which is equal to AC = w·a,
since we’ve defined the labor-input coefficient as the number of workers per
unit of output. With a linear P P F the marginal cost (i.e., the cost of
increasing output by one unit) will be equal to average cost, and with perfect
competition on the goods market, price will equal marginal cost, which is
thus equal to average cost in this case. Hence,

p = MC = AC = w · a.

You should remember from introductory microeconomcis that, under perfect

competition, each firm’s supply curve equals its marginal cost curve. In the
Ricardian case marginal cost is constant, and the supply curve is thus a
horizontal line. This means that the position of the demand curve will not
influence the equilibrium price, it is determined by productivity (a) and the
cost of labor (w).

Determination of terms-of-trade
How is terms-of-trade determined? In Figure 2 the dashed line shows (Ger-
many’s) terms-of-trade, or the world relative price ratio (PC /PP ), I just
assumed, for the sake of the argument that it lies between the slopes of the
production possibility sets, i.e., between −1 and −3. To see more clearly the
connection between the goods prices and the slope of production possibility
set we start from the pricing equations for W and C (for example in France):

pFW = wF · aFW , pFC = wF · aFC .

The relative price of W in terms of C in France is given by,

pFW wF · aFW aFW

= = ,
pFC wF · aFC aFC

which is simply the opportunity cost of increasing W in terms of C lost.

When free trade is established, the same price ratio will be established in
both countries (since they now have common goods markets). Exactly where
terms-of-trade will end up depends on the interaction between supply and
demand on both markets simultaneously (a so called general equilibrium)
and of the relative sizes of the two countries. However, it must lie between
the autarchy relative price ratio in the two countries, otherwise one of the

8
countries would not want to participate in international trade, preferring
instead to trade with itself.
We may just briefly look at the condition for a general equilibrium on
the two market simultaneously. According to Walras’ law the value of the
aggregate excess demand (in an closed economy) must always be equal to
zero,
pFW · (DW
F F
− SW ) + pFC · (DC
F
− SCF ) = 0,
there DW F and S F are the demand and supply of wine in France etc. If
W
there is an excess demand for cars, (DC F − S F ) > 0, there must be an
C
F F
excess supply of paper, (DW − SW ) < 0, in autarchy. This means that
the relative price, (pW /pC ), must decline so that the excess demand for C
is reduced, as is the excess supply of wine. In a general equilibrium, in
autarchy, both markets must be in equilibrium at the same time, i.e., the
excess demands (or supplies) must be equal to zero simultaneously. In a
single open economy, a positive excess demand may persist, in for example
the C-market in France, even in equilibrium, since this simply means that
France imports cars. However, Walras’ law must hold but now for the entire
world market (or France and Germany together) and since both prices are
positive, it must mean that France must have a positive excess supply of
wine; this excess supply is exported to Germany. Let’s define.

MF = DCF
− SCF , France’s imports of cars,
F F F
X = SW − DW , France’s export of wine,
−pW X + pC M F
F
= 0
pF X F = MF
MF
pF = ,
XF
where pF = ppWC
is France’s terms-of-trade, i.e., the ratio of the price of
its export good (wine) to its import good (cars). The same analysis for
Germany would give:

−pC X G + pW M G = 0
pG X G = M G
MG
pG = ,
XG
where pG = ppW C
is Germany’s terms-of-trade (note that: pF = p1G and vice
versa). This way of defining a country’s terms-of-trade means that than it
increases the country’s position has improved, i.e., it will be able to buy
more import goods for each unit of export good that it gives up (this is the
interpretion of the ratio M/X). With only two goods and two countries when
one country’s terms-of-trade improves the other automatically worsens.

9
The determination of quantities produced
With linear PPF and perfect competition in both goods markets, the relative
goods price in autarchy is determined only by relative cost of production.
However, exactly how much that is produced and consumed in autarchy is
determined from the demand side. The demand side can be illustrated by
indifference curves, or by the usual demand curve. Figure ?? shows, in the
left-hand diagram, how the position of the community (or aggregate) indif-
ference curves determines exactly how much is produced of wine and cars in
France in autarchy (W ∗ and C ∗ ). Note again the important fact that the rel-
Wine Wine

France France
Wmax = 300 Wmax = 300

(Convex indif f erence (Right angled

curves - imperf ect indif ference curves -
subsitutes) perf ect complement)

W* W* =150

Car Car
C* C* = 75
Cmax = 150 Cmax = 150

Figure 3:

ative price of cars in terms of wine is determined by the relative production

pF aF
costs, or pFC = aFC ; the slope of the PPF. The usual shape of indifference
W W
curves is that they are convex as seen from the origin, which reflects an
assumption that the two goods are (imperfect) substitutes to each other.
In this case it is somewhat difficult to pin down exactly the point on the
PPF where the highest possible indifference curve just touches the budget
line (the PPF). If we assume that the two goods are perfect complements
it will be much easier to determine the exact quantities produced and con-
sumed in autarchy equilibrium. This case is illustrated in the diagram at
the right-hand side in figure ?? where the usually convex indifference curves
have been replaced by indifference curves which are right-angled. When two
goods are perfect complements from the consumers’ perspective they must
be consumed together in order to yield any utility at all to the consumer.
Now, wine and cars do not match very well ("don’t drink and drive"), how-
ever for the sake of the argument we’ll assume that these two goods are
always consumed in the ratio 2 : 1, i.e., two bottles of wine for each unit
of car consumed. If this was indeed the case it would be relatively easy to

10
find the exact point on France’s PPF which is chosen in autarchy. Note
the straight dashed line drawn from the origin through the corners of the
indifference curves. This line has constant slope equal to 2, or it can be
written as the following equation:

W = 2 · C.

The equation for France’s PPF is:

W = 300 − 2 · C.

The equilibrium quantities produced and demanded are determined by the

intersection of these two lines. We set the two equations equal to each other
and solve for C ∗ :

2 · C = 300 − 2 · C
300
C∗ = = 75
4
W ∗ = 2 · C ∗ = 150.

Concave PPFs
A linear PPF is very unrealistic. It is much more plausible that it becomes
increasingly more difficult to increase production of either good the more
of it that is produced initially. The PPF will have a concave shape in this
case. This means that the labor input coefficients are not constant but are
increasing with the volume of cars or wine produced. It also means that
the opportunity cost of increasing the production of wine, i.e., the ratio
aC
aW is not constant but is quite small in absolute number then only wine
is produced and increases as we "slide down" the PPF, towards a higher
production of cars. In autarchy the quantity produced and consumed and
the relative price ratio is determined by the price ratio which is consistent
with a general equilibrium on both markets at the same time. Hence, in this
case we need both the supply and demand side to determine the relative
price ratio. Let’s assume that the demand side, as represented by the shape
of the indifference curves, are very similar in France and Germany. The
supply side, represented by the PPFs are quite dissimilar as before. In
particular the French PPF is much steeper then much wine and few cars are
produced in either country, reflecting a higher opportunity cost of increasing
car production in France than in Germany. This makes it likely that the
relative price of cars in autarchy in France is much higher than in Germany,
and since if is the autarchy relative price that determines the direction of
trade, France will import cars and export wine, as before. However, France
will not specialize completely in wine (and Germany not completely in cars),
but will specialize in these goods in the sense that they will produce (supply)

11
more of them than they consume themselves (have an excess supply). They
will keep producing some of the other good (cars in France and wine in
Germany) but will produce less than they consume (have a positive excess
demand).
In a free-trade equilbrium the relative price ratio has adjusted so the
supply equals demand on the world market, but France will import cars
in exchange for wine. The gains from trade can still be seen from the fact
that the consumption possibilities are expanded. Figure ?? shows both
France’s and Germany’s PPFs in the same diagram. Since Germany is
assumed to be a larger economy, due to higher average productivity, its PPF
lies outside France’s. The autarchy points are marked by A(F ) and A(G),
respectively. The relative prices in autarchy are not shown in order not to
clutter the picture with too many lines, but is given by the slope of the PPF
at the tangency point with the highest possible indifference curve. France
would produce in autarchy at a point on its PPF there the slope is steeper
than for Germany. With free international trade the relative (and absolute)
prices would be the same in both countries, illustrated by the dashed lines
(which have the same slopes). The production points in both countries
are at points there the relative price lines is tangent to the PPFs, and the
consumption points there we relative price lines are tangent to the highest
possible indifference curves. Note that the production point in France moves
from A(F ) up along France’s PPF to Prod(F), which means more wine is
produced and less cars. The consumption of cars are increased in France,
which is due to a lower relative price of imported German cars. The vertical
difference between the amount of wine produced and consumed in France,
is France exports of wine. In Germany the car-sector expands and the wine-
industry declines, i.e., the production point moves down along Germany’s
PPF from A(G) to Prod(G). On the consumption side, more wine and less
cars are consumed. (We should note here that it is not necessarily the case
that the consumption of the good whose relative price has increased will
decline. Relative specialization according to comparative advantage makes
the country richer and the income effect may overwhelm the substitution
effect so that the net effect is an increase in demand for both goods.)

A piecewise linear "World" PPF

We now return to the case with linear PPFs. In Figure ?? France and
Germany’s PPFs have been combined into a "world" PPF. If we start at
the intercept on the vertical axis, 800 units of wine is the maximum that can
be produced if both countries specialize in wine. If some "global" central
planner now tell Germany to start producing cars, but France still only wine,
we will start sliding down Germany’s PPF, exchanging one unit of cars for
one unit of wine. This goes on until Germany is completely specialized in
cars and France in wine, at point A. If the central planner now tells France to

12
 Wine

Germany

Cons(G
)
A(G)

Prod(F)
Prod(G)

A(F) Cons(F)

France

Car

Figure 4:

start producing cars as well, we will start sliding down France’s PPF giving
up three units of wine for each additional unit of cars produced. If the central
planner had gotten the countries’ comparative advantage wrong he would
have told France to start producing cars first and we would have traced out
the dashed world PPF, which lies everywhere inside the one consistent with
an efficient pattern of specialization (except at the intercept points).
An important point that we should note from Figure ?? is that with two
countries complete specialization according to comparative advantage would
only occur if the aggregate (world) indifference curve touches the world PPF
at the corner (point A). In this case any relative price between the autarchy
relative prices (−3 to −1) would support such an equilibrium. However, if
the world taste for cars vs. wine is like the dashed indifference curve which
is tangent to the world PPF at point B, this is not consistent with complete
specialization. In this case France would be completely specialized in wine
but Germany would produce both cars and wine. The world relative price
ratio would be equal to −1 in this case, i.e., equal to Germany’s autarchy
relative price. Only France would gain from trade, Germany would be in-
different.

More goods and countries

Diagrammatical exposition limits us to consider the case of only two goods.
However, with many goods (N ) and still two countries we may choose one
good (n) to compare all other with and produce rankings of ratios of labor

13
 W in e

800

500

A
300

Ca r s
150 500 650

Figure 5:

input coefficients. I.e., for goods 1 to n − 1 France has a lower opportunity

cost in terms of the number of good n that it must give up than is the case
in Germany:

F
G


a1 a1 an−1 F an−1 G
< ,..., < .
an an an an
For goods n + 1 to N France has a higher opportunity cost than Germany:



F
G
an+1 F an+1 G aN aN
> ,..., > .
an an an an
Note that we have ranked the good so that the n − 1 first ones are those for
which France has a comparative advantage and N − n − 1 last ones are the
one for which Germany has a comparative advantage.
If we go back to the case with only two goods, but now consider more
countries, we can again construct an efficient world PPF by combining the
individual countries’ PPFs. Assume for example that we add Italy and
Sweden to Germany and France. Italy’s opportunity cost of increasing car
production in terms of wine lost is −5 while Sweden’s is −0.5. In Figure 6
we may again start from the intercept on the vertical axis, which shows the
maximum amount of wine that can be produced if all four countries only
produces wine. This would occur if the price of cars, in terms of wine, on
the world market is less than 0.5 (Sweden’s opportunity cost of producing

14
cars). At such a low price of cars no producer (in any country) would find it
worthwhile to start producing cars. However, if the relative price increases
to lie between −0.5 and −1, such as the slope of straight line that touches
point A on the world PPF, Sweden will specialize completely in cars and
produce no wine. If the price of cars increases further to a relative price ratio
between −1 and −3 (at point B), Sweden and Germany would produce only
cars and France and Italy only wine. At point C only Italy will produce
wine.
Wine

Wmax
Sw eden

Germany

France

Italy

Cars

Figure 6:

Neoclassical production theory

The Ricardian theory is too simple because it only assumes one factor of
production. The Heckscher-Ohlin (HO) model, which we’ll come to soon,
considers two, or more, factors. Furthermore, the HO-model specifies a
general production structure which allows substitution possibilities between
factors of production (they in fact rely on the so called neoclassical theory
of production). As we’ll see later, relative factor prices will generally be
different in different countries and this leads firms in different countries to
use factors of production in different proportions.
A natural starting point is the concept of a production function, e.g., for
W and C as before:
QW = F (K, L), QC = G(K, L),

15
where K is ”capital” and L is labor as before, and F and G stands for the
functional relationship between K and L and Qi (the quantity of output of
good i). A simple, and useful, functional form is the so called Cobb-Douglas
function, so let’s use that:

QW = K αW L1−αW
QC = K αC L1−αC ,

where αW and αC etc. are parameters which determines the shape of the
production function and the relative importance of capital and labor in the
production processes of wine and cars.
To illustrate the production possibilities in a diagram with the two inputs
on the axes, we can trace out a set of isoquants, i.e., combinations of capital
and labor which can produce the same quantity of the output good. Let’s
focus on wine production and try to trace out the isoquant for one unit of
wine (QW = 1) :
QW = K αW L1−αW = 1
next we solve for K αW :
1
K αW = = LαW −1 ,
L1−αW
next we try to get rid of the exponent of K by raising both sides to the
power of α1W :

αW · α1 (αW −1)· α1
K W = L W
αW −1
K(L, 1) = L αW
.

K is now expressed as a "function" of L and 1 (the quantity of W ). Two plots

of this function are shown in the figure below, for the case where αW = 1/3
and αW = 1/2 (the former is the steeper line).
K6
5

0
1 2 3 4
x

Both isoquants have convex shapes as seen from the origin, which reflects
the fact that K and L are substitutes in the production of wine. This means

16
that it is possible, to some extent to use more capital and less labor (or vice
versa) but still produce the same amount of output. The convex shape
also means that the isoquants become flatter and flatter the further towards
either of the axes we move, i.e. it becomes increasingly difficult to replace
capital with labor the less capital we have initially (or vice versa). The
actual choice of how much of each factor to use depends on the relative
factor cost, i.e., the cost of using capital compared to the cost of hiring
labor. The cost of capital is a "rental rate" per unit of capital and time,
which we denote r; and the cost of labor is the market wage rate, w.

Cost minimization
A profit-maximizing firm, operating on perfectly competitive input- and
output-markets, will choose the combination of the two inputs which maxi-
mizes profits, given the targeted level of output. The total cost of production
is equal to the amount used, times the cost per unit, of each factor:

C̄ = w · L + r · K.

A firm that wants to maximize its profit can as well be thought of as min-
imizing its cost of production. Taking w and r as given by the markets, it
will choose L and K to minimize total cost subject to the constraint that it
can produce at least its target level of output. If we take this target to be
Q = 1, we write the minimization problem formally as:

min w · L + r · K s.t. Q = K α L1−α ≥ 1.

K,L

This problem can be illustrated graphically in an isoquant diagram by solv-

ing the expression of total cost for K :
C̄ w
K= − · L.
r r
C̄
r is the intercept on the vertical axis and − wr is the slope of the straight
line. This straight line shows combinations of K and L giving exactly C̄ in
total cost, it is called an iso-cost line. By pushing this line as far towards
the origin as possible but at the same time not losing contact with the target
isoquant, gives the cost-minimizing solution at the point there the isoquant
and the isocost are tangent to each other. The solution is illustrated in
Figure 7. In the left-hand side diagram the factor price ratio is w/r and the
optimal factor combination can be read off the two axes. Note that since
we defined the target output level to be Q = 1, we can define the optimal
factor levels in terms of input-coefficients, as we did in the Ricardian model.
However now we define aK and aL the number of units of capital and labor,
respectively, necessary to produce exactly one unit of output. In the right-
hand side diagram we show what happens as the factor price ratio increases

17
 K K

C'/r

C/r

a'K
aK

Q=1 Q=1
(K/L)' -(w /r)'
K/L -(w /r)
L L
aL a'L

Figure 7:

to (w/r)′ , i.e., the cost of labor has gone up relative to the cost capital. It
will be optimal (cost minimizing) to use less labor and more capital and we
move to the left up the isoquant. Note that we now have a new optimal
capital-labor ratio (K/L)′ and a′K (a′L ) is higher (lower) than initially. This
means that in the neoclassical model of trade (the HO-theory) the unit
input coefficients are not constant as they were in the Ricardian model, but
depend on the relative factor prices. We therefore write them as functions
of w and r : aK (w, r) and aL (w, r).
The unit input coefficients that we get as a solution to the cost min-
imization problem can also be thought of as demand functions for capital
and labor by firms. It is somewhat difficult to solve mathematically for these
demand functions, so I state the result directly for the type of production
function that we considered in the beginning of this section:


α w 1−α
aK (w, r) =
1−α r


α w −α
aL (w, r) = .
1−α r

Since we assume that 0 < α < 1, it follows that the demand for capital
is increasing in the factor price ratio, w/r, and the demand for labor is
decreasing, just like the diagram above showed.
The unit input demand functions can now be substituted into the defi-

18
nition of total cost:
C̄(w, r) = w · aL (w, r) + r · aK (w, r)



α w −α α w 1−α
=w· +r·
(1 − α) r (1 − α) r

−α
1−α
α 1−α α
= w α
·r + w1−α · rα
1−α 1−α

−α
1−α 
α α
= + w1−α · rα
1−α 1−α
  
A
1−α α
=A·w ·r .
The last line gives us the unit (or average) cost-function, the smallest pos-
sible cost of producing exactly one unit of output given the factor prices,
w and r. Note that average cost increases in factor prices, but the increase
in cost is more sensitive to an increase in the cost of capital (r) the more
important capital is in the production process (i.e., the higher α is).

Example 1 Assume that w = 1 and r = 1, and α = 1/4. The factor

demands are, given an output target of one:
− 1
1 4

L∗ (1, 1) =
4 1 = 1.316,
1− 4
1− 1
1 4
4
K ∗ (1, 1) =
 = 0.439.
1 − 14

Assume now that the wage rate increases to w′ = 2, the factor demands
become:
− 1
1 4

L∗ (2, 1) =
4 1 · 2 = 1.11,
1− 4
1− 1
1 4

K ∗ (2, 1) =
4 1 · 2 = 0.74.
1− 4
The minimum cost of producing one unit at these prices are:
 − 1 1− 1 
1 4 1 4

C ∗ (1, 1) = 
4 1  +
4 1  = 1.75,
1− 4 1− 4
  1  1
1 −4 1 1− 4
∗
C (2, 1) = 
4
+
4   · 21− 14 = 2.95.
1 − 14 1 − 14

19
Combining prices to average cost
Given the assumption of perfect competition on both output markets the
output prices will be equal to the unit (average) cost:

PW = AW · w1−αw · rαw ,
PC = AC · w1−αC · rαC .

Note that these unit cost (obviously) are functions of factor prices and we
can actually define new types of iso-cost curves, now in the {w, r}-space.
The equation for the unit cost iso-cost curves can be found by setting the
prices equal to 1, and solving for w :
1 αw
−
w = AW1−αw · r− 1−αw ,
1 α
− 1−α C
− 1−α
w = AC C
·r C .

Example 2 Assume that αW = 1/4, and αC = 1/2. This implies that the
1 1
− − 1−α
constants are: AW1−αw = 0.474, AC C
= 0.25, and we can rewrite the
above equations as:
1
w = 0.474 · r− 3 ,
w = 0.25 · r−1 .

These equations are plotted in Figure 8, where the steeper curve denotes
the unit cost isocost in the car sector and the flatter the same in the wine
industry. If both prices were equal to 1 initially, the wage and rental rates
are found by the intersection of the two curves.

The Heckscher-Ohlin — model

The Heckscher-Ohlin (H-O), or the factor proportion, model, tries to ex-
plain why certain countries have comparative advantages for certain goods.
It does so by explicitly assuming that there are no differences in technolog-
ical knowledge in the two countries. This means that given the same inputs
the same amount of final goods could be produced in both countries (the
production functions, or the isoquant maps, are the same in each country).
Hence, technological knowledge is ruled out as determinants of comparative
advantage. Instead H-O focus on differences in endowments of factors of pro-
duction. For example, Sweden has obviously a vast supply of trees suitable
for making paper, and this is the basis for our comparative advantage in this
particular business. Likewise, China and India have comparative advantages
in production of goods which require a lot of (unskilled) labor inputs. In the
H-O model, like in the Ricardian model, factors of production cannot move

20
 1,4

PC = 1
1,2

PW= 1

0,8
w

0,6

0,4

0,2

0
0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1
r

Figure 8:

across country borders, and differences in factor proportion tends to persists

therefore. However, we will see that, in the H-O model, trade in goods is a
substitute to migration of factors. Additional important assumptions are:

• Constant returns to scale in both sectors,

• factors are completely mobile between sectors within a country,
• all factors are inelastically supplied,
• there are no imperfections in any market so factors are paid (the value
of) their marginal products,
• consumers preferences are identical and homothetic in both countries1 .
(This assumption is made in order to rule out taste differences as a
basis for trade.)

Factor proportions We will assume that there are two factors only, cap-
ital (K) and labor (L). If industry W uses less capital per laborer than
industry C, at some given factor price ratio (w/r), we say that C is capital-
intensive relative to W. I.e.,



K K
> .
L C L W
Likewise, of course, W is labor-intensive relative to C. This is illustrated in
Figure 9, where the solutions to the cost minimization problem for firms in
1
Homothetics preferences implies that the income − consumption curve is a straight
line from the origin.This implies, in turn, that the income-elasticities of the two goods are
equal to one and constant.

21
the car industry is at point AC and that for firms in the wine industyr at
point AW . The straight lines drawn from the origin to these two points show
the optimal capital labor ratios in the two industries.
KW

(K/L) C
QC = 1

C/r

aK C (K/L) W
AC

aK W AW

QW = 1
LW
a LC a LW

Figure 9:

Factor supplies are assumed to be given at a point in time (inelastic

supply) and may differ between countries. If G is endowed with more capital
per labor compared to F, we say that G is relatively capital-abundant:



K K
> .
L G L F

As long as the capital-labor ratios differ between sectors the production

possibility curves will be concave, even if we have constant returns to scale
in both sectors. Since the two countries differ in their factor-proportions,
the production possibility curves will generally differ in shape, but due to
the assumption of constant returns to scale they will retain their shapes
as the total size of the countries varies. The assumption of identical and
homothetic preferences implies that we can apply the same indifference map
to both countries. In Figure 10, we’ve used these assumptions to illustrate
the equilibrium price ratios (p = pW /pC ) in both countries in autarchy (the
dashed lines). Note that the relative price ratio is steeper in G than in
F in autarchy, reflecting the fact that F is relative labor-abundant and W
is relative labor intensive (or that G is relative capital-abundant and C is
relative capital-intensive).
When free-trade is allowed the price ratio will end up somewhere be-
tween the autarchy price ratios, this means that wine will be relatively
cheaper in Germany and cars relatively cheaper in France (and cars rel-
atively more expensive in Germany etc.). Germany will now produce less,

22
 W ine

Franc e's PP F
AS

AG

P AG
P AS
Germ any's PP F

Cars

Figure 10:

but consume more wine, than in autarchy, and imports the difference from
France. France, in turn decreases its production of cars, but increases its
consumption, importing cars from Germany in exchange for wine. These
moves are illustrated in Figure 11, where the A− points stands for autarchy,
P - points for production under free-trade, and C− points stand for con-
sumption under free-trade.
W ine

PS

AS CS

CG

AG

PG
Franc e's P P F
p*
G erm any 's P P F

Cars

Figure 11:

With the help of our graphical approach we can now state the Heckscher-
Ohlin theorem:

”Given the assumptions of the model, a country will export

23
 the commodity that intensively uses its relatively abundant fac-
tor.”

Factor-price equalization When trade starts the production of C in G

and W in F increases. Since C is relatively capital-intensive and W relatively
labor-abundant, where will be a shift in demand in favor of K in G and in
favor of L in F, and since there is no supply responses, by assumption, the
price of capital must increase relative to the price of labor in G, and vice
versa in F. In general, there is a functional relationship between the relative
price of W and C and the w/r− ratio. We’ve earlier defined the unit cost
of producing W and C is as:

PW = C̄W (w, r) = w · aLW (w, r) + r · aKW (w, r),

PC = C̄C (w, r) = w · aLC (w, r) + r · aKC (w, r).

Now, since aLW (w, r) > aLC (w, r), and aKW (w, r) < aKC (w, r), but w and
r is the same within each country, we can define a functional relationship
between the goods price ratio, and the factor-price ratio: PPW
C
= G( wr ). This
relationship is illustrated in Figure 12, there we can see that a move to
free-trade decreases the relative factor-price ratio in G and increases it in
F. The implication of this is that workers gain in France and capital owners
gain in Germany. Furthermore, under certain conditions it is possible that
factor prices become equalized across the countries, i.e., not only the ratio
is equal but also the individual factor prices. This follows since prices are
determined from the cost side and the prices are the same in both countries.
We now state the so called factor-price-equalization theorem (FPE):

P W /P C

G( w/r )
pG

*
p

pF

w/r
*
(w/r) F (w/r) (w/r) G

Figure 12:

24
 ”Under identical constant-returns-to-scale production tech-
nologies, free trade in commodities will equalize relative factor
prices through the equalization of relative commodity prices, so
long as both countries produce both goods.”

Example 3 We’ve earlier found the unit cost equations, taking the ratios
of these gives:

PW AW · w1−αw · rαw
=
PC AC · w1−αC · rαC
= A∗ · wαC −αw · rαw −αC

Or using our earlier assumption that αW = 1/4, and αC = 1/2, we get,

PW w 1
4
= A∗ ,
PC r

where A∗ = 1.7548
2 = 0.877, hence this price ratio is obviously increasing in
w/r (since W is the labor intensive industry).

Extra: Diversification and FPE. An important condition for FPE to hold

is that factor endowments are not too different; if they are, complete
specialization may ensue, and then the move towards equalization of
factor-prices stops. However, free-trade still tends to equalize factor
prices. In Figure 13, we illustrate the problem of minimizing the total
cost subject to the constraint of producing exactly one unit of output
of two representative firms, one in the wine-industry and one in the
car-industry. Since C is relatively capital-intensive its unit isoquant
(QC = 1), lies closer to the vertical axis than the unit isoquant for W
(QW = 1). The capital-intensity is given by the slope of the straight
lines from the origin to the tangency points (K/L), it is also given by
the unit input coefficients (aKC > aKW , and aLW > aLC ). Note that
the two straight lines form a cone, with its point (or vertex) at the
origin. This cone is called the cone of diversification. This cone gives
the limits between which each country’s capital-labor ratios must lie,
in order for it to produce both goods in a free-trade equilibrium. To
understand this, note that for both goods to be produced, it must be
the case that the average capital-labor ratio for the two commodities
lies between (K/L)C and (K/L)W . This average must also be equal
to the endowment ratio for the entire country.

A cost side approach to FPE We’ve earlier looked at a single firm’s

cost-minimization problem with K and L on the axes, however we can also
analyze the same problem in "price-space" i.e., with the factor prices (w and
r) on the eaxes. Figure 14 depicts unit iso-cost curves in price-space. It can

25
 K

(K/L) M
Q C= 1

1/r

a KC AC (K/L) W

a KW AW

QW = 1
L
a LC a LW 1/w

Figure 13:

be shown (which we will not do) that the slope of an iso-cost curve is equal
to the capital-labor (K/L). ratio employed in the same industry.
If both goods are produced in equilibrium, we must have that PW = CW
and PC = CC , and in this case w and r must be given by the intersection of
the two isocost curves. We noted above that the slope of an iso-cost curve
at the cost-minimizing point shows the capital-labor ratio employed in that
industry. Note also that we must have: K̄ = KW + KC , and L̄ = LW + LC ,
so that we have full employment of both capital and labor. The aggregate
(or economy wide) capital-labor ratio is therefore: (K̄/L̄), which can be
written as,



K̄ LW KW LC KC aKW aKM
= + = λLW + λLM ,
L̄ LW + LM LW LW + LC LC aLW aLM

where λLW = LW /L̄, the fraction of the total labor force used in the wine
industry (and analogously for λLC ). Obviously, λLW + λLC = 1, so the ex-
pression above says that the economy-wide capital-labor ratio is a weighted
average of the two factor intensities.
If K̄/L̄ is not a weighted average of the capital-labor ratios in the two
industries at point A in Figure 14 it will not be possible (or an equilibrium)
that both goods are produced. For exampel, if K̄/L̄ is larger than the slope
of both curves at that point only the capital intensive good (C) will be
produced, if it is smaller, only W will be produced. This means that if the
two goods differ very much in their factor intensities it is likely that the
country will become fully specialized in the good which is intensive in the
factor of which the country is richly endowed.

26
 w

CW =P W
CC =P C
r

Figure 14:

Factor-Price Equalization If both countries produce both goods they

must both be at point A in Figure 14 . With free-trade goods prices are
equalized but since goods prices with perfect competition in both industries
are equal to average cost, they must also have identical factor prices. If
one or both countries specialize, they cannot both be at A and must have
distinct factor prices. Whether or not both countries are at A depends upon
whether both have factor-endowments ratios between the slopes of the two
isocost curves at A.

The Stolper-Samuelson Theorem Both goods- and factor-prices change

as countries move from autarchy to free trade. Since it is the real wage and
real return to capital that counts for workers and capital owners, we have to
take into account how these price change in relation to each other, in order
to evaluate which factor owner will gain or lose. This is what the so called
Stolper − Samuelson theorem tells us:

”An increase in the relative price of the labor-intensive good,

will increase the wage rate relative to both commodity prices
and reduce the rent relative to both commodity prices”.

Note that since the nominal wage increases relative to both goods prices
we do not have to take into account the workers’ consumption pattern in
order to conclude that workers gain from an increase in the relative price

27
of the labor-intensive good. If it had been the capital-intensive good whose
price had increased, it would have been the capital owner that had gained
(unambiguously).
To illustrate the theorem graphically, suppose that G produces both
goods, and that the initial equilibrium is at point A, in Figure ??. Now,
suppose that the price of cars increases from PC to PC′ , this shifts the isocost
curve, for C, outward in proportion of the price rise. At point B the CC curve
has the same slope as at A, and w, r and PC are all greater by an equal
percentage (AB/A0), so their relative magnitudes are unaltered.PW has not

A'
CW = P W
C C = P 'C
CC = P C
r

Figure 15:

changed, so the new equilibrium is at A′ . At this point w is lower than at A

and r is greater than at B. This implies that r has increased by a greater
proportion than has PC . Thus r rises relative to the prices of both goods,
and the wage falls relative to the prices of both goods.

The Rybczynski Theorem Factor endowments have so far been as-

sumed to be constant in both countries. However, they may very well (and
have in most countries) change over time, and the K/L− ratios may there-
fore also change. Such changes will of course upset the earlier free-trade
equilibrium, causing changes in prices, production and consumption. The
Rybczynski theorem explains how output of both industries change, as the
factor endowment change:
”At constant prices, an increase in one factor endowment will
increase by a greater proportion the output of the good intensive
in that factor and will reduce the output of the other good.”
To illustrate this theorem we assume that G:s capital stock is increased,
ceteris paribus. In Figure 16 this is shown as an outward shift of G:s pro-

28
duction possibility curves; note that it does not shifts out proportionally,
but more along the vertical axis, since C is relative capital-intensive. Out-
put changes from A′ to A, along the output expansion line, indicated in the
figure. It is clear from the figure that the production of C increases and
the production of W decreases, as the theorem states. Note that the world
relative price is assumed to be unchanged, which means that the change in
factor supply in G is not big enough to affect the world market price.If we
C
O u t p u t e x p a n s io n lin e

A'

Figure 16:

define K̂ as the rate of change in the capital stock, L̂ as the rate of change
in the supply of labor, and Q̂W the rate of change in W − output etc, we
must have that:
K̂ = λKW Q̂W + λKC Q̂C ,
L̂ = λLW Q̂W + λLC Q̂C ,
where λKW = aKWK·QW , the fraction of the nation’s capital stock used in the
W − industry, and analogously for the other λ′ s. Note that λKW + λKM = 1
so that the first equation says that K̂ is a weighted average of Q̂W and Q̂C ,
and similarly for the second equation. It follows that,
Q̂C > K̂ > L̂ > Q̂W
when K̂ > L̂, and
Q̂W > L̂ > K̂ > Q̂C
when L̂ > K̂.

Intersectoral factor immobility — The specific factor model

The HO-model is best interpreted as a model of the (very) long-run, since
it assumes that factors are effortlessly transferred between sectors in a each

29
country. In reality real capital can only be reallocated between sectors in-
directly, by profits from the declining sector firms being distributed to their
owners instead of plowed back into the firms and used to replace assets as
they depreciate. These profits are then invested in positive net investments
in the advancing sector. Obviously this process takes time. Similarly, as
workers in one sector becomes unemployed, they often have to go back to
school and retrain, before they are able to get a job in the expanding sector.
The specif ic − factor model takes a much shorter time period into account
and assumes that, for all practical purposes, factors can only be used in one
sector and not in the other, i.e., they are immobile between sectors.
Even though, as we pointed out, labor as well as capital can be immobile
in the short-run, we will assume here that labor is in fact mobile between
sectors, while capital can only be used in one sector. Effectively, then, we
have a model with two goods, two countries and three factors. This means
that we will also have three factor-prices, one common wage rate, and two
returns to capital (rC ) and (rW ). We will start with a simple approach first,
and look at a more advanced treatment later.

The specific-factor model - a simple approach In Figure 17 we’ve

put together two diagrams showing the value of the marginal product of
labor in the wine industry and in the car industry. We have the wage rate
on the vertical axes and labor on the horizontal axis. The length of the
horizontal axis is determined by the fixed amount of labor available; note
also that we’ve reversed the direction of the car diagram, which goes from
right to left. Intitially we have an equilibrium on the labor market with a
wage w0 and L0W number of workers in the wine industry (L0C = L̄ − L0W
in the car industry). If the demand for wine increases, and the price of
wine therefore goes up, the value of the marginal product of labor in the
wine industry increases for each level of labor force employed. The wage
level increases to w1 and workers move from the car to the wine industry,
induced by the higher wages during the transition to the new equilibrium.
To understand which groups (workers or capital owners) gains or loses
from an increase in the price of wine, we can look at the part of Figure 17
which shows the MPLW -curve in isolation. This part is shown in Figure 18.
Note that the total wage bill paid to workers within the Wine industry is
equal to w·LW , since we still assume perfect competition the value of output
in the wine industry must be equal to the total compensation paid out to
factor owners, or,
PW · QW = w · LW + rW · KW ,
In Figure 18 the total area under the MPLW up to L0W is equal to PW 0 · Q0 ,
W
and it divides as shown between compensation to workers and capital owners.
If the price of wine goes up by x%, it is clear from Figure 17 that the
wage rate goes up by less, hence owners of wine-specific capital will definitely

30
 w

w1

w0

P 0 W *M P W L P 0 C *M P C L P 1 C *M P C L

LC L0C L1C LW

Figure 17:

gain in real terms (the ratio rW /PW increases). The owners of car-specific
capital will definitely lose, however, because the wage rate has increased
while the price of cars is unchanged. The area showing the total rent to
owners of car-specific capital, in a similar diagram as for the car industry,
has shrunk. What about workers, have they gained or lost? Well, we cannot
actually say because while their wage has increased by between 0 and x%,
the price of wine has increased by more (by x%), and if they consume a lot
of wine their real wage may very well have declined. Similarly, the owners of
car-specific capital have lost, both because their rent has declined and also
because goods have become more expensive (their purchasing power have
declined).

The specific-factor model - a more advanced approach We now use

the isocost framework to analyze the specific-factor model. In Figure 19, an
initial long-run equilibrium (HO-equilibrium) would occur at A. However,
we let’s assume that an initial, short-run, equilibrium is at points D and
E, for the wine- and car-industries, respectively. The common wage rate
is the distance w = 0B, the return to car-specific capital is rC = BE, and
the return to wine-specific capital is: rW = BD. Assume now that PC rises
in the proportion EM/0E (= BJ/0B) to PC′ . If labor, like capital, did not
move between sectors, the car equilibrium would switch from E to M, where
w and rC , increases in proportion to PC while the wine industry stays at
D. But this is impossible because the car wage, 0J, would exceed the wine
wage, 0B, and labor is mobile even in the short-run. What will happens
is that labor will move from the wine-industry to the car industry, until a
common wage rate has been established at an intermediate position, such
as 0F. The new short-run equilibrium is now G (wine) and H (cars).

31
 w

T o t a l re n t t o c a p it a l o w n e rs

W a g e b ill

w0

P 0C *M P C L

LC L 0C

Figure 18:

Because G is necessarily to the left of D, rW has fallen; because H is

to the right of M , rC has risen relative to PC ; because F is between J and
B, w has risen relative to the unchanged PW , but fallen relative to PM . In
summary,
r̂C > P̂C > ŵ > P̂W > r̂W .
We may summarize this conclusion as follows:

”A relative price increase of a good benefits the specific factor

used in that industry, reduces the real income of the other specific
factor, and has an ambiguous effect on the mobile factor.”

Imperfect competition
“Pro-competitive” gains from trade
Monopoly All previous models assume constant returns to scale and per-
fect competition. However, we know that many industries are characterized
by imperfect competition, due to various degrees of increasing returns to
scale. In these industries it is possible that there are gains from trade, even
in the absence of any comparative advantages.
Even if imperfect competition, and in particular a monopolistic market
structure, is usually associated with increasing returns to scale, it is con-
venient to separate these phenomena when analyzing international trade.
First we note that in our basic 2 × 2 × 2 HO-model, with constant returns
to scale, perfect competition and different factor intensities, the so called

32
 w

M
J
G H
F
B E
D

A'
A C W =P W

C C =P 'C
C C =P C
r C, r W

Figure 19:

first-order profit maximizing condition in a general equilibrium in autarchy

is,
PW MCW
= = MRTCW ,
PC MCC
which is the tangency condition between the relative price line and the pro-
duction possibility set. This follows from each individual firm’s profit max-
imizing decision which leads it to set the marginal cost equal to the given
price. We can also remind ourself that each (atomistic) consumer will set,
PW MUW
= = MRSCW ,
PC MUC
and hence the general equilibrium entails a tangency between the production
possibility curve and the highest attainable (community) indifference curve.
Assume now that the wine industry is monopolized (for example in F ).
The wine-monopoly will set the marginal revenue equal to marginal cost,
which we know from previous courses can be expressed as,
 
1
MRW = PW 1 − = MCW ,
εW

where εW = − dQ W pW
dpW QW , is (negative of) the price-elasticity of demand.
Now if the car industry is still perfectly competitive we have that,
 
1
PW 1 − εW MCW PW
= = M RTCW < = MRSCW .
PC MCC PC
Hence, the tangency condition which guarantees overall efficiency of the
competitive allocation does not longer hold. All these things are illustrated

33
in Figure 20. The autarchy equilibrium is at point Am , where MRTCW <
PW
PC = MRSCW . If we assume that a move to free trade implies a move
from monopoly to perfect competition, and if both countries are assumed to
be identical there will be no actual trade taking place, but the possibility
of trade breaks down the monopoly power of the domestic wine-producing
firm. The increase in utility is the pro − competitive gain from trade.
C

Am

p*
Pm

Figure 20:

Increasing returns to scale

It was assumed in the previous example that there were imperfect com-
petition in autarchy, but still constant returns to scale in production. This
combination is in general incompatible though, and it was only done in order
to emphasize the potential gains from trade through increased competition.
Since a monopolistic market structure normally is due to the presence of
scale economics, it is not certain that more firms will lead to lower prices, if
production has to be done at a suboptimal scale. Another reason for gains
from trade was suggested already by Adam Smith, in a famous passage he
said the ”the gains from division of labor is limited by the extent of the
market”. The quickest way of expanding a market is to remove trade bar-
riers, and that may permit a greater extent of specialization, through an
international division of labor, which could lead to lower prices, even if
there is no increase in competition.
Theoretically, it is difficult to incorporate increasing returns to scale into
general equilibrium models, which typically assumes perfect competition and
price taking behavior, which is inconsistent with (large) scale economies.
One way of reconciling scale economies with price taking is by assuming
that the scale economies are external to the firm.

34
External economies of scale This is not a boneheaded idea. Larger
geographical markets makes it possible for more service industries to arise
around some core industry. In the geographical models we’ll look at later
this will be a reason for agglomerations, i.e., it pays to be located close to
each other. However, in pure trade models without any real geographical
dimension, a larger market will lower the cost of doing business since it
will be cheaper to buy inputs, such as technical consultancy, transport-
and handling services, marketing and legal services, as well as, to specialize
in R&D. The important theoretical point is that these lowering of total
costs associated with a higher total output refers to the entire industry, not
to individual firms, which still take prices as given. I.e., act like perfect
competitors.
The gains to trade with external scale economies is illustrated in Figure
21. We assume here that there are external economies of scale in both the
car and the wine industries in both Germany and France, which implies a
convex production frontier (the thick line). We also assume that these two
countries are identical, i.e., they are of the same size and have the same factor
proportions, and also the same tastes. With constant returns to scale, which
in this case implies linear production frontiers, there would be no reasons
for trade.
We assume that the autarchy equilibrium is a point A, with utility equal
to UA in both countries. If Germany now specializes in cars and France in
wine (produce at points QM G and QW F , respectively), and if they exchange
these goods along the straight line between these points, they might reach
point T, with equal consumption in both countries and utility UT . Clearly a
welfare improvement in both countries.A problem with the previous example
C

QCG

E UG

A
UA

W
QWF

Figure 21:

is that it is not very likely that the consumption point will be at T, which

35
requires identical and homothetic preferences. If, for example wine, is a
luxury good its proportion of the consumers budgets will increase as income
increases, this is shown by the curved income-consumption curve from A
through E. However, E cannot be a trade equilibrium because G wants
to import more wine than F wants to export (or G wants to export more
cars than F wants to import). This implies that there is an excess demand
(supply) of wine (cars), and the relative price of wine in terms of cars (the
slope of the straight line between QCG and QW F ) must increase to clear the
markets.Figure 22 shows a trade-equilibrium, where G consumes at point
C

QCG

E XP C G

EF
EG

IM P C A

QWF
W

E XP W F

Figure 22:

EG on the income-consumption curve extending from A, and F consumes

at EF . Here, G:s desired exports of cars is equal to F :s desired imports, etc.
There are several things to note: First, since F happens to specialize in the
good with the higher income elasticity it will gain more from trade than G
does. Second, the relative price ratio, PPW
C
, has increased (the solid lines are
steeper than the dashed autarchy price-line), which implies that at QCG :
PW M CW PW M CW
PC < M CC , and at QW F : PC > M CC , and firms in both countries want
to specialize in their respective goods, and the equilibrium is stable. Third,
in autarchy the factor prices are equalized, but with trade they are driven
apart; the reverse of the HO-result.
The last point can be seen by considering that since both countries have
the same technologies and the same factor endowments, they both allocate
the same bundles of factors to the goods they produce. But if factor inten-
sities are not equal in the wine- and car industries, the factor prices which
clear the factor markets, must be unequal. In each country, the factor price
ratio, w/r, will equal the slope of the isoquant of its good through the en-
dowment point. This implies that if F specializes in W, it will have a higher
w/r ratio than G, because if W is labor intensive it’s isoquant is steeper than

36
that for M. Each country will therefore have relatively high factor price for
the factor used intensively in its export industry. It should be noted that in
contrast to the Stolper−Samuelson result, the factor which ”loses” (capital
in F, labor in G) are not necessarily worse of in this case, since the capture
of scale economies may increase the absolute productivities of both factors,
and they will both be better off with trade.
One may wonder how specialization will occur in this model? Since both
countries are identical there is no automatic mechanism (different autarchy
prices) which determine the pattern of specialization, and it is not possible
to predict the specialization pattern. Historical accidents will decide which
country happens to be first in specializing in a particular good. An om-
niscient central planner would, in the example above, want the domestic
industry to specialize in wine, since that will lead to a higher final welfare.
If governments in both countries are intelligent enough to realize this, they
would both like to induce (e.g. through subsidies) a concentration towards
the wine industry. These efforts might easily frustrate each other.

Internal economies of scale We’ve already discussed the gains from

trade through the pro − competitive effect. However, in that analysis it was
(implicitly) assumed that the technology was characterized by constant-
returns to scale. With increasing returns to scale, internal to firms, there
will be further gains from trade, through decreasing average cost. To see
this effect we can decompose the change in total cost as,

∆C(x) = ACx ∆x + x∆ACx ,

dividing by the change in output, ∆x,

 
∆C(x) ∆ACx
= MCx = ACx + x .
∆x ∆x

Now, if p exceeds marginal cost, an expansion of output leads to an increase

in welfare through the pro-competitive effect, which is approximately equal
to,
(p − MCx ) ∆x,
or, 
∆ACx
(p − MCx ) ∆x = (p − ACx ) ∆x − x ∆x.
∆x
The first term on the right-side of this equation can be called the prof it
ef fect. If price exceeds average cost an increase in output generates a surplus
of price over average cost on the additional output. The second term is
the decreasing-average-cost effect. Note that, with increasing  returns,
 the
change in average cost with respect to output is negative: ∆AC ∆x
x
< 0, and
hence welfare is increased.

37
 In an industry with some degree of scale economies, new firms may enter
until the profit is equal to zero for the last entrant. However, each individual
firm may not operate at minimum average cost. With trade the ”marginal”
firms in each country may have to exit and the other firms can expand their
output. Price and average cost will decrease, as outlined above. However,
where may also be fixed cost of starting production, for each firms. With
fewer firms, each producing at a larger scale, there will be a savings of real
resources since less fixed costs are incurred. This is called the f irm exit
ef fect. Note that we may thus observe, firm exits as trade is expanded, but
this is not a sign of less competition.

Gains from product diversity

Wine and cars are two distinct goods, they are substitutes in the sense
that they compete for the consumers’ money. However, we often think of
substitutes as two goods which can fulfill a certain need, or craving, that
the consumer has. E.g., wine and beer may be substitutes as dinner drinks,
in fact there are various types of wine (red, white, dry, sweet, etc.) which
are close, but not perfect, substitutes. Now, for some meals red wine is the
best, but for others white wine is the better choice. These sort of preferences
are said to exhibit a love of variety. Another possibility is that there is a
fraction of the population which always want red wine, and the others prefer
white wine to every meal. People with such preferences has an ideal variety,
of each good.
If the production of red and white wine are both associated with a (sep-
arate) fixed cost of the same amount, while the marginal cost is equal for
both, it follows that the average cost will always be greater if both varieties
of wine are produced in a single country. If G and F are again identical,
and if they specialize in one variety (G in white wine and F in red wine),
there will be savings of one fixed cost in each country, and the average cost
will consequently be lower. Opening trade now means that consumers with
love-of-variety preferences will benefit from lower prices, but still consume
both types of wine. If everything is symmetrical (red and white wine gives
the same utility to each person, the cost of production and factor endow-
ments are the same, the countries are equally big), red and white wine will
be exchanged at the ratio one-to-one. The gain in welfare is illustrated in
Figure 23, where both types of wine are assumed to be produced with in-
creasing returns to scale. We should note that there is no pro-competitive
effect in a move from autarchy to free trade, if both goods were produced in
autarchy.
If different consumer groups in each country has different ideal varieties,
and there are increasing returns in the wine industry, the gains from trade
can be illustrated in yet another way (Figure 24). Here it is assumed that
the fixed cost is so large that in autarchy each country is producing only

38
 White w ine

Fixed
cost

UT

A
UA

Red w ine
Fixed cost

Figure 23:

one variety; let’s assume that F produces red wine and G white wine. In
autarchy the french people which prefers white wine only reach utility level
UW W F (the indifference curve passing through point B, with the steepest
slope), and Germans which prefer red wine only reaches utility level URW G
(the indifference curve passing through point C with the flattest slope).
With trade these groups can increase their utility and reach the utility levels
as the compatriots in the other country (UW W F → UW W G , and URW G →
URW F ). There is no change, in production, nor in average cost, and the
utility for the group which prefer the variety that their country produced
in autarchy does not change. The move to free trade is Pareto-optimal,
though.

Monopolistic competition - The Dixit-Stiglitz model

The love-of-variety model was initially suggested by Chamberlin (1933), but
it was further developed by Dixit and Stiglitz (1977) to be consistent with
general equilibrium. To illustrate some of the traits of their model, we
assume the following, ”representative” consumer’s, utility function,
N  ρ1
 ρ
U= ci , 0 < ρ < 1,
i=1

where ci is the quantity consumed of variety i of the a differentiated con-

sumption good, which is available in N varieties; ρ is a parameter in the util-
ity function chosen by Dixit and Stiglitz, which indicates how close substi-
tutes the different varieties are (the closer to 1, the closer to being ”perfect”

39
 White w ine

C
UWWF

UWWG

B
Red w ine
URWG URWF

Figure 24:

substitutes). This functional form is characterized by a constant elasticity

of substitution (CES) between each pair of the N varieties. If we make the
further assumption that the consumer consumes each variety in the same
quantity (so that ci = c, for all i), we can rewrite the utility function as,
1 1 1
U = (N · cρ ) ρ = N ρ · c = N ρ −1 · N c.
1
The term N ρ −1 shows the ”love-of-variety” effect on utility; since ρ < 1
utility is increasing in N, but at a decreasing rate. The term N c shows the
total quantity produced of the differentiated good; or the size of the market.
It is a positive externality because a larger market will support a bigger
number of varieties which directly benefit the consumer.
This representative consumer maximizes this utility function subject to
the budget constraint,
N

pi · ci ≤ I,
i=1
where I is the consumer’s money income, and pi is the price of good i.

Decreasing average cost and the zero profit condition The original
monopolistic competition model assumed that there were increasing returns
to scale, at the plant level, which meant that the average cost of production
falls with a higher scale of production. A neat (and realistic) way of mod-
elling this assumption is to assume that there is a fixed cost (fi ) of setting up
a production plant, but that the marginal production cost is constant at m.

40
This implies that the total cost of production is equal to Ci (xi ) = fi + m · xi .
A not so realistic assumption is to assume that the only factor of produc-
tion is labor, the total amount of labor necessary to start production and to
produce at the scale xi , is then,

Li = fi + mi · xi ,

Each of the N potentially active firms chooses the scale of production of its
own variety and determines a price, pi , to maximize profit,

max [(pi − w · m) xi − w · fi ] ,
pi

where w is the wage rate. The solution is familiar (MR = MC),

1
pi (1 − ) = w · mi .
εi
1
Note that, since εi = 1−ρ (or, ρ = 1 − 1ε ), the equilibrium price is,

w · mi
pi = .
ρ

This shows that the lower ρ is (i.e., the less substitutable the goods are) the
higher is the price, and vice versa.
The monopolistic competition model allows free entry of firms, and we
have to add a zero profit condition, which for the representative firm is,


w · mi
− w · mi xi = w · fi ,
ρ


1
w · mi − 1 xi = w · fi ,
ρ
fi · (ε − 1)
xi = ,
mi

since, 1ρ − 1 = ε−11
. The interpretation is that the output of each firm is
fixed in equilibrium, it does not depend on the aggregate income (or the size
of the economy). As the economy grows the number of varities grows; the
number of varieties depends on the available labor force and the size of the
fixed cost and the parameter ε. We can calculate this by noting first that
the amount of labor needed to produce the given amount of output is,


fi · (ε − 1)
Li = fi + mi ·
mi
= fi + fi · (ε − 1) = fi · ε.

41
This gives the labor requirement for each firm; if there are L̄ units of labor
available, the maximum number of varieties that can be produced (subject
to the zero-profit constraint) is,

L̄
N∗ = .
f ·ε
The higher the fixed cost and the higher the elasticity of demand, the lower
will be the number of varieties in a monopolistic competitive equilibrium à
la Dixit-Stiglitz.

The Dixit-Stiglitz model and gains from Intra-industry trade If

we consider two neighboring countries of exactly the same size, preferences,
technological knowledge and labor endowments, we would conclude that
there would be no reason for, or gain from, trade in the HO-world. In the
Dixit-Stiglitz world, however, a move from autarky to a common market
would double the number varieties. If everything is symmetric all consumers
(in both countries) would consume each variety and half the output of each
country would be exported, and since each variety has exactly the same price,
the value of trade would be exactly balanced. Since we’ve seen above that
the scale of production for each firm is fixed there are no gains from capture
of economies of scale, and the price of each variety is therefore unchanged,
as are the wage rate and hence national income. The gain from trade will
come entirely from the expansion of the number of varieties available, which,
according to the assumed utility function, leads to an increase in utility.
Note, that if we introduced trade costs the effects would have been muted.

Taste differences as determinants of trade

Up to now we’ve assumed that consumers in different countries have iden-
tical preferences, or if there are different groups of consumer with different
preferences these groups are represented equally in both countries. How-
ever, it is obvious that taste differ and these differences are by themselves
sufficient determinants of trade. This should be no surprise since we started
with deriving excess demand and -supply functions in each country, with
different autarchy equilibrium relative prices; when trade started these dif-
ferent relative prices determined the direction of trade. While the Ricardian
and HO-models focused on differences in technology, or cost of production
due to different factor endowments, it is of course equally plausible that
differences in autarchy relative prices are due to different demand functions.
We will focus on two reasons why there are differences in demand be-
tween countries: i) genuine differences in consumers’ taste, and ii) non-
homogenous preferences, i.e., the income elasticity of demand is not con-
stant but varies with households’ income level. The first case is illustrated
in Figure 25. The left-hand diagram shows the autarchy situation in F and

42
G, assuming that they have the same production frontier (same technology
and factor endowments), but different tastes. Consumers in F has prefer-
ences biased towards wine, and consumers in G towards cars. This implies
that the autarchy price ratio ( PPW
C
) is higher in F than in G (since the french
demands more wine at a every price than the Germans do). The right-hand
diagram show a trade equilibrium where PT is the common price line, and
Q is the production point which is the same in both countries (since they
have the same technology and factor endowments). F consumes more (less)
wine (cars) than she produces and imports (exports) the difference; G con-
sumes more (less) cars (wine) than it produces and imports (exports) the
difference.
C C

PG CG
AG
AG

Q
CF
AF
AF

PT
PF
W W

Figure 25:

Factor prices will be the equalized through trade, as in the HO-model.

The Stolper-Samuelson theorem also holds: In F , for example, the price of
wine will decline and if wine production is relatively labor intensive, the
wage-rental ratio will fall, the real return to labor will fall, and the real
return to capital will rise. The opposite changes occur in G.
If preferences are non-homothetic, the autarchy prices will differ even if
consumers’ in the two countries have the same preferences, but if they differ
in factor endowments and/or technology. Let’s imagine that cars stands for
some basic ”subsistence” good, which must be consumed at some minimum
quantity, and wine is a relative luxury, which will be consumed only when
incomes have increased enough. This implies that the income-consumption
curve does not emanates from the origin, but has a positive intercept along
the C− axis. This is illustrated in Figure 26, where CC0 stands for the
subsistence level of cars in both countries.
By assuming that the income—consumption curve is linear, instead of
starting from origin but immediately curling down towards the W − axis,

43
 M

Engel c urv e
QG
CG
CF

CM0
QF P

P
W

Figure 26:

it is theoretically possible to aggregate individual preferences into one set

of aggregate preferences. The shape of the income—consumption curve in
Figure 26 is therefore assumed due to theoretically consistency. Now, let’s
assume that F is poorer than G; we may for example return to the Ricar-
dian world, where there are differences in technological know-how (absolute
factor-productivity), which implies that the G:s production frontier lies fur-
ther out from the origin compared to F :s. Otherwise the two production
frontiers have the same shape, implying that given the same price-ratio, the
two goods will be produced in the same proportions in both countries. In
autarchy consumers in F spend a higher proportion of their budgets on C,
than does consumers in G. This implies that the relative price-ratio, PPW C
,
is greater in G than in F. When trade starts the same prices will be es-
tablished in both countries and they will therefore produce at points QF
and QG , while consumption takes place at points CF and CG , respectively.
Hence, the poor country (F ) will import the subsistence good (C) and export
the luxury good (W ), and the rich country (G) will export the subsistence
good and import the luxury good. Alternatively we can predict that if dif-
ferences in demand, due solely on differences in per capita income, is the
only determinant of trade, the poorer country imports the good with the
low income elasticity (e.g., food).

Types of trade: Inter- & intra-industry , North-

South & North-North
The theories of the determinants of trade which rests on some inherent dif-
ferences between countries, such as differences in technological knowledge,

44
factor endowments, preferences, or per capita income, all predict that coun-
tries will tend to exchange relatively different goods. This type of trade is
called inter−industry trade. The dominant model, the HO-model, predicts,
in particular, that it is differences in factor endowments which determines
the direction of trade. Countries which are relatively labor intensive will
tend to export labor intensive goods, such as textiles, and countries which
are relatively capital abundant will export capital intensive goods, such as
automobiles. This type of inter-industry trade is also called north − south
trade, since the labor intensive (and therefore poor) countries, are mostly
to be found in the southern hemisphere. However, most of the interna-
tional trade is actually taking place between the rich countries in the north
(north−north) trade, furthermore this trade is to a large extent in the form
of intra − industry trade, i.e., trade with goods within the same industrial
classification.
There are several explanation for intra-industry trade. One, very plausi-
ble reason, is product differentiation due to love-of-variety preferences, which
we’ve already have discussed. Increasing returns to scale is another reason
for trade between similar countries. But, in our discussion of scale economies
as a basis for trade we assumed (implicitly) that the two goods were quite
different. However, a large part of the intra-industry trade is actually in
intermediate products. The presence of scale economies will lead to lower
average costs of inputs if their production is concentrated in one country.
The next step in the production process is located in another country, and
the intermediate products are imported. Automobiles is one case in point,
intermediate parts, of for example a Volvo, are imported to a large extent,
assembled and then exported. Now, assembly of cars and production of car
parts are classified into the same industry, hence trade in intermediate parts
is an example of intra-industry trade.
We should note that the different stages of production, within the same
industry, may require quite different factor proportions and it is therefore
perfectly possible that the HO reasons for trade does apply to trade in in-
termediates; scale economies is not the only possible explanation. However,
in this case it is really the artificiality, or crudeness, of the industrial classi-
fication system which produces this apparent negation of the HO-theory.
Transport costs is an important variable which we have neglected so far.
Since there is a greater distance between north and south than between
north and north, much trade north-south trade is simply not profitable,
hence the great proportion of north-north trade in the statistics. If trade is
thus geographically limited it is possible that there are more trade within
a region comprising two or more countries, than between regions within
one country. If scale economies are important, we’ve seen that historical
accidents may determine the location of production between identical, or
similar, countries. It is therefore possible that one region in for example
France has come to specialize in cars and trade mostly cars for German wine

45
with a nearby German region, while another region in France specializes in
wine and trade wine for German cars.
Finally we should note that if we combine the HO-model with the last of
models of trade determined by differences demand due to differences in per
capita income, we get another explanation for why the North-South trade is
not so prevalent as the HO-model alone would suggest. If, for example food,
is a labor intensive good, the HO-model predicts that it will be exported
from the labor-abundant South. However, food is also the most typical
subsistence good with a very low income elasticity, and for this reason poor
countries, which are usually also labor-abundant, will tend to import food.
These two effects might cancel and there would be very little trade between
rich and poor countries. Furthermore, the latter would not be able to reap
any gains from trade, due to division of labor and scale economies, nor will
they get the benefit of positive knowledge spillovers often associated with
trade. Consequently, the may stay poor and the North and the South may
continue to diverge.

The Linder hypothesis, Product cycles & Technol-

ogy cycles
The models we’ve looked at so far are inherently static in nature. There are a
given set of goods, factors of production and technological know-how. Even
though growth can easily be introduced through growth in factor endow-
ments, we know from growth theory that it is technological progress which
is the most important determinants of economicd growth. We may think
of technological progress mostly in form of better techniques for producing
already existing goods (so called process-innovation), however, the inven-
tion and introduction (product-innovation) of new products is basically the
same thing. Process- and product innovation does come for free but are the
fruits of investment in research & development. R&D can be considered as
a particular service industry, even if it is not tied down to a particular prod-
uct, but takes place in most traditional industries. It is usually an activity
which is vertically integrated within an organization (firm) which produces
final goods. It also requires highly skilled workers, e.g. workers with an
advanced university education, and is said to be intensive in human capital.
Staffan Burenstam-Linder should be credited with the insight that new
goods, which requires much R&D, are usually introduced in the home mar-
ket first, and later exported to other countries. Newly developed goods may
also be quite expensive and can only be sold in markets where consumers (on
average) have a relatively low price elasticity of demand and are relatively
rich (since new products usually have a high income elasticity).
Burenstam-Linders idea is close in spirit to the so called product cycle
model of international trade, introduced first by Vernon. This model as-

46
sumes that new products go through three stages in their ”life-cycles”. The
first stage is the innovative stage, which takes place in the rich countries
which are abundant in human capital. Initially the goods can only be sold
in the rich home country, where the price elasticity is low. It is also usu-
ally imperative to be in close contact with customers in the early stages of
the introduction of a new product, since its characteristics may have to be
adapted to consumers’ need, and/or the technology must be adjusted. This
also lead firm to start introducing their product close to home. The sec-
ond stage is the maturing stage, which occurs when the technological and
product characteristics have been figured out. At this stage the good may
start to be exported to medium rich countries; the larger market permits ex-
ploitation of possible scale economies and lower prices, which in themselves
help speed up the market penetration. Economies of scale may also include
learning − by − doing, standardization of parts and division of the produc-
tion process. Some part of the production process, or the entire process,
may in the third stage by moved to a country rich in unskilled labor. The
standardization of the processes makes it possible to employ such workers,
and the low wage cost therefore becomes of paramount importance.
The movement of production to poorer countries entails also a trans-
fer of technological knowledge, i.e., the knowledge how to organize pro-
duction in the best way. In the HO-model this knowledge is assumed to
be available to all countries equally. This could be a fair description in a
long-run (static) equilibrium. However, in a world where economic growth
is pushed along through continuous introduction of new products and tech-
niques, there will always be a certain knowledge-gap between the R&D− and
human-capital-intensive countries, and the more (unskilled) labor-intensive
countries. Furthermore, the fruits of R&D, and acquired organizational
knowledge is something which firms want to keep for themselves. This may
not be completely possible, but it may be a long time during which the
innovating firm is the sole proprietor of the new knowledge. The patent
institution also give firms a rather long (25 years) monopoly position in the
use of the new technique, or the production of the new product. Sometimes
these firms choose to exploit this monopoly position by licensing the right
of production to other firms, sometimes it chooses to produce it themselves.
One possible explanation for this choice, between licensing or own produc-
tion, is the fear that licensing will result in too much (unwanted) transfer of
technological knowledge. The product life-cycle model suggests that at the
third stage firms will move production in response to low wages. For many
poor countries, production of relatively advance goods through licensing,
or the entry of a foreign multinational firm, could be their best chance of
closing the technological knowledge-gap with the richer countries.
If one distinguishes between process- and product-innovation, one may
also talk about a technology-cycle model. We’ve basically already explained
the dynamics of this cycle, but we should also note that firms in high-wage

47
countries have a further incentive for labor-savings process-innovations. This
technology is initially not transferred to low-wage countries since that it is
not the cost-minimizing technology at the relative factor-prices in those
countries. Eventually, when this technology has become ”old” it may be
exported to the low-wage countries, in particular if these countries’ income
has increased (and relative factor prices changed) to make it appropriate to
use this particular technique.

48