This action might not be possible to undo. Are you sure you want to continue?

# Munich Personal RePEc Archive

**A Dynamic Chamberlin-Heckscher-Ohlin
**

Model with Endogenous Time

Preferences: A Note

Iwasa, Kazumichi, Kikuchi, Toru and Shimomura, Koji

Kobe University

September 2007

Online at http://mpra.ub.uni-muenchen.de/4981/

MPRA Paper No. 4981, posted 07. November 2007 / 04:21

A Dynamic Chamberlin-Heckscher-Ohlin Model

with Endogenous Time Preferences: A Note

∗

Kazumichi Iwasa

†

Toru Kikuchi

‡

Koji Shimomura

§

September 20, 2007

Abstract

This note formulates a dynamic two-country (developed and developing

countries) Chamberlin-Heckscher-Ohlin model of trade with endogenous

time preferences a la Uzawa (1968). We examine the relationship be-

tween initial factor endowment diﬀerences and trade patterns in the steady

state. In particular, to highlight the integration of developing countries

(e.g., China) into the world trading system, we concentrate on the case of

asymmetric size of two countries (in terms of population). It will be shown

that (i) given that the representative household in each country supplies

∗

We are grateful to the Associate Editor and two annonymous referees for helpful com-

ments. We acknowledge ﬁnancial support from the Ministry of Education, Culture, Sports,

Science and Technology of Japan (the Grant-in-Aid for the 21st Century Center of Excellence

Project ‘Research and Education Center of New Japanese Economic Paradigm’).

†

Graduate School of Economics, Kobe University, 2-1 Rokkodai-cho, Nada-ku, Kobe 657-

8501, Japan

‡

Corresponding author, Graduate School of Economics, Kobe University, 2-1 Rokkodai-

cho, Nada-ku, Kobe 657-8501, Japan; Tel: 81-78-803-6838; Fax: 81-78-803-6838; e-mail:

kikuchi@econ.kobe-u.ac.jp

§

Research Institute for Economics and Business Administration, Kobe University, 2-1

Rokkodai-cho, Nada-ku, Kobe 657-8501, Japan

1

an equal amount of labor, only intra-industry trade occurs in the steady

state irrespective of diﬀerences in the number of representative households

and that (ii) the number of households being equal, the country with less

labor eﬃciency becomes the net exporter of the capital-intensive good.

JEL Classiﬁcation Code: F12

Key Words: dynamic Chamberlin-Heckscher-Ohlin model, developed and

developing countries, trade patterns

2

1 Introduction

In recent decades, many developing countries have opened their economies to

international trade. As an example, China’s integration into the world economy

is one of the most important developments aﬀecting the structure and evolution

of the global trading system at the dawn of the 21st century. How does the

integration of developing countries into the world economy aﬀect world trading

patterns?

It seems to be very important to consider this problem in a dynamic Heckscher-

Ohlin trade model. However, while the static Heckscher-Ohlin theorem holds

even if preferences and technologies are slightly diﬀerent among countries, the

dynamic Heckscher-Ohlin theorem under the assumption of exogenous time pref-

erence that was proved by Chen (1992) holds only if preferences and technologies

are strictly identical among countries. In other words, under exogenous time

preferences, at least one of the two countries should specialize in one of the two

goods and it is very diﬃcult to derive satisfactory results on trade patterns.

1

The state of the art in dynamic trade theory is apparently unsatisfactory. This

seems to suggest that the traditional focus on exogenous time preferences should

be accompanied by a focus on endogenous time preferences.

2

Thus, we address the question of developing countries’ integration in a dy-

namic Chamberlin-Heckscher-Ohlin (CHO) model with endogenous time pref-

erences a la Uzawa (1968), in which there is a monopolistically competitive

‘diﬀerentiated products’ sector, and a perfectly competitive ‘consumable capi-

tal’ sector.

3

Consider the world economy as consisting of one developed country

1

This was pointed out by Stiglitz (1970, p.463).

2

A non-constant time preference rate has been empirically documented through panel data

and cross-country data by Hong (1988), Lawrence (1991) and Ogawa (1993).

3

The static Chamberlin-Heckscher-Ohlin model has been extensively investigated. Help-

man’s (1981) seminal integration of the monopolistic competition trade model into the two-

3

and one developing country. The developed country reached a steady state be-

fore the developing country (which corresponds to China) started the process

of development (i.e., the removal of trade barriers). For simplicity, we call the

former Home and the latter Foreign. Then China’s decision to join the world

trading system represents the opening of trade between Home and Foreign.

Kikuchi and Shimomura (2007) examine a similar problem using a dynamic

two-country Chamberlin-Heckscher-Ohlin model.

45

They assume, however, that

both countries are endowed with an equal number of households. Thus the role

of size diﬀerences in factor endowment is downplayed in the analyses. In the

real world, there is a signiﬁcant size diﬀerence between developed and devel-

oping countries. For example, China’s population is 20 percent of the world

population. To our knowledge, little attention has been given to the relation-

ship between timing of development and the size of developing countries. Thus,

it is important to consider the case of the asymmetric size of countries.

In this note, we extend the analysis of Kikuchi and Shimomura (2007) to the

case of asymmetric size of two countries (in terms of population). We demon-

strate that, given that the representative household in each country supplies an

equal amount of labor, only intra-industry trade occurs in the steady state irre-

spective of diﬀerences in the number of representative households. Even if there

country by two-factor by two-good Heckscher-Ohlin (HO) framework, which was extended

and made popular by Helpman and Krugman (1985), has led to the widely held belief that

HO and Chamberlinian monopolistic competition are complementary in nature.

4

Atkeson and Kehoe (2000) examine a similar problem using a dynamic Heckscher-Ohlin

model composed of a larger number of small open economies.

5

The literature on dynamic two-country models originated in Oniki and Uzawa (1965).

While they assume exogenous saving rate in each trading country, most subsequent contribu-

tions, including Stiglitz (1970), Chen (1992), Shimomura (1992, 1993, 2004), Ventura (1997),

Nishimura and Shimomura (2002, 2006), assume that households maximize their discounted

sum of utility, i.e., saving rates are endogenously determined. Chen, Nishimura and Shimo-

mura (2005) discuss other major problems within dynamic HO models.

4

is a larger amount of labor (in terms of population) in the developing country,

due to catching-up by the developing country, sources of inter-industry trade

based on diﬀerences in the capital-labor ratio vanish and only intra-industry

trade occurs in the steady state.

This note is organized as follows. Section 2 sets up a dynamic CHO model

and Section 3 discusses the existence, uniqueness and local stability of the steady

state. Section 4 derives trade-pattern propositions. Section 5 provides conclud-

ing remarks.

2 The Model

Consider a world economy consisting of two countries, Home and Foreign, that

diﬀer in their factor endowments. There are two types of commodities, diﬀeren-

tiated products (Good 1) and a consumable capital (Good 2), produced using

reproducible capital, k, and a primary and time-invariant factor of production, l

(labor). The consumable capital can be either consumed as a non-durable good

or added to the existing capital stock. Labor is measured in eﬃciency units.

Each Home (resp. Foreign) representative household supplies l (l

∗

) units of ef-

ﬁciency labor. The population of each country is assumed to be constant over

time. The Home (resp. Foreign) population is m (resp. m

∗

). Thus, the Home

(resp. Foreign) household is endowed with ml and mk (resp. m

∗

l

∗

and m

∗

k

∗

)

units of factors of production. Note that Kikuchi and Shimomura (2007)’s case

corresponds to m = m

∗

= 1.

Following the standard trade theory, we assume away international factor

movements. Moreover, in order to focus on international trade, we assume that

there is no international credit market, while there is a competitive domestic

credit market in each country.

5

Each consumer maximizes the discounted sum of utility.

∞

0

uXdt =

∞

0

f [U(V, C

2

)] Xdt, (1)

˙

X = −ρ(u)X, (2)

where V is the quantity index for diﬀerentiated products, C

2

is the consumption

of the consumable capital, and X ≡ exp{−

t

0

ρ(u)dτ} is the discount factor at

time t which depends on the past and present level of utility through the function

ρ.

Following Uzawa (1968), we assume that the variable discount rate ρ(u)

satisﬁes

ρ(0) > 0, ρ

(u) ≡

dρ(u)

du

> 0, ρ

(u) ≡

d

2

ρ(u)

du

2

> 0,

0 < θ

ρ

≡ [uρ

**(u)/ρ(u)] < 1 for any positive u < ∞. (3)
**

It will be assumed that U is linearly homogeneous in its arguments and f

satisﬁes

f(0) = 0, f

(U) > 0, f

(U) < 0. (4)

Quantity index V takes the following Dixit-Stiglitz (1977) form:

V =

¸

N

0

x(i)

(σ−1)/σ

di

¸

σ/(σ−1)

, σ > 1, (5)

where N is the total number of diﬀerentiated products, x(i) is the consumption of

the i-th variety of diﬀerentiated products, and σ is the elasticity of substitution

between varieties.

Solving the static expenditure minimizing problem, we can deﬁne the expen-

diture function as

e(P)ψ(u) ≡ (minPV +C

2

, s.t., u = f [U(V, C

2

)]) , (6)

6

where the consumable capital serves as the numeraire, P ≡

N

0

p(i)

1−σ

di

1/(1−σ)

is the price index for diﬀerentiated products, and ψ(u) is the inverse function

of f, which clearly satisﬁes

ψ(0) = 0, ψ

(u) > 0, ψ

(u) > 0. (7)

Given that the equilibrium is symmetric, that is, p(i) = p and x(i) = x for

∀

i ∈ [0, N], we can obtain the following condition from the envelope theorem,

∂e(P)ψ(u)/∂P = V .

e

[N

1/(1−σ)

p]ψ(u) = N

σ/(σ−1)

x

or N

1/(1−σ)

e

[N

1/(1−σ)

p]ψ(u) = Nx.

Assume that diﬀerentiated products are more capital-intensive than the con-

sumable capital.

6

Diﬀerentiated products are produced by monopolistically

competitive ﬁrms under increasing returns technology, while the consumable

capital is produced by competitive ﬁrms under constant returns technology.

Assume that each ﬁrm in the diﬀerentiated products sector has the homothetic

total cost function c

1

(w, r)φ(y), where y is the output level of each ﬁrm. There

are signiﬁcant economies of scale: φ(y)/y is decreasing over the relevant range

of output levels y. The marginal revenue will be equated to the marginal cost:

p [1 −(1/σ)] = c

1

(w, r)φ

(y).

7

Furthermore, free entry implies that price equals

6

This assumption is just for simpliﬁcation and this capital intensity ranking itself does not

alter the results of this paper.

7

We can obtain this relation as follows. Considering the subutility maximization problem:

max V, s.t.,

R

N

0

p(i)x(i)di ≤ I, we obtain the inverse demand function of i-th variety as

follows: p(i) = [P

(σ−1)

I/x(i)]

1/σ

. Therefore, the revenue of the i-th ﬁrm is given by

π

i

= p(i)x(i) −c

1

(w, r)φ[x(i)]

= [P

(σ−1)

Ix(i)

(σ−1)

]

1/σ

−c

1

(w, r)φ[x(i)].

and the ﬁrst order condition, dπ

i

/dx(i) = 0, yields p(i) [1 −(1/σ)] = c

1

(w, r)φ

[x(i)].

7

average cost: p = [c

1

(w, r)φ(y)]/y. By combining these conditions, one can eas-

ily see that all varieties will have the same output level ¯ y, which is deﬁned

by

8

1 −

1

σ

=

¯ yφ

(¯ y)

φ(¯ y)

.

The constraints on labor and capital within Home are

9

c

1

w

(w, r)φ(¯ y)n +c

2

w

(w, r)y

2

= ml, (8)

c

1

r

(w, r)φ(¯ y)n +c

2

r

(w, r)y

2

= mk, (9)

where n is the number of diﬀerentiated products produced in Home and c

2

(w, r)

and y

2

are the unit cost function and the output of the consumable capital,

respectively.

Then, by deﬁning ξ ≡ ¯ y/φ(¯ y), the zero-proﬁt conditions can be written as

ξp = c

1

(w, r), (10)

1 = c

2

(w, r), (11)

and we can obtain the factor price functions w(ξp) and r(ξp). Utilizing these

factor price functions, the national income is shown as

r(ξp)mk +w(ξp)ml. (12)

The partial derivative of the national income with respect to the price of diﬀer-

8

This result depends crucially on homotheticity in production. See Dixit and Norman

(1980, pp. 284–5). To guarantee the existence and uniqueness of ¯ y, we assume that φ satisﬁes

φ

(0) < ∞, φ

(0) > −∞, lim

y→∞

θ

φ

< 1 −

1

σ

, and

dθ

φ

dy

< 0 for any positive y < ∞,

where θ

φ

≡ [yφ

(y)/φ(y)]. An example of φ(y) is ln(y + 1).

9

As it is clear from these equations, any country’s population size does not aﬀect its relative

factor abundance in the static sense. Our aim is to check whether population size aﬀects long-

run capital accumulation.

8

entiated products, p, is equal to the aggregate national output of those products:

n¯ y = ξr

(ξp)mk +ξw

(ξp)ml. (13)

From (12), we can obtain another condition for each household:

˙

k = r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ(u). (14)

Each household maximizes (1) subject to both (2) and (14). Associated with

this problem is the Hamiltonian

H ≡ uX +λ{r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ(u)} −δρ(u)X, (15)

where λ and δ are the shadow prices of k and X. The necessary conditions for

optimality are

0 = X −λe[N

1/(1−σ)

p]ψ

(u) −δρ

(u)X, (16)

˙

λ = −λr, (17)

˙

δ = ρ(u)δ −u. (18)

Letting Z ≡ λ/X and combining (2) and (17), we can obtain

˙

Z = Z[ρ(u) −r(ξp)]. (19)

Based on the foregoing argument, our dynamic general equilibrium two-

9

country model is described as

˙

k = r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ(u), (20)

˙

k

∗

= r(ξp)k

∗

+w(ξp)l

∗

−e[N

1/(1−σ)

p]ψ(u

∗

), (21)

˙

Z = Z[ρ(u) −r(ξp)], (22)

˙

Z

∗

= Z

∗

[ρ(u

∗

) −r(ξp)], (23)

˙

δ = ρ(u)δ −u, (24)

˙

δ

∗

= ρ(u

∗

)δ

∗

−u

∗

, (25)

0 = 1 −Ze[N

1/(1−σ)

p]ψ

(u) −δρ

(u), (26)

0 = 1 −Z

∗

e[N

1/(1−σ)

p]ψ

(u

∗

) −δ

∗

ρ

(u

∗

), (27)

0 = N¯ y −ξ[r

(ξp)(mk +m

∗

k

∗

) +w

(ξp)(ml +m

∗

l

∗

)], (28)

0 = e

[N

1/(1−σ)

p]N

1/(1−σ)

[mψ(u) +m

∗

ψ(u

∗

)] −N¯ y. (29)

The system determines the equilibrium path of two state variables, k and k

∗

,

and eight jump variables, Z, Z

∗

, δ, δ

∗

, u, u

∗

, p, and N.

10

3 The Steady State

The steady state is the solution for the system of equations

0 = r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ(u), (30)

0 = r(ξp)k

∗

+w(ξp)l

∗

−e[N

1/(1−σ)

p]ψ(u

∗

), (31)

0 = ρ(u) −r(ξp), (32)

0 = ρ(u

∗

) −r(ξp), (33)

0 = ρ(u)δ −u, (34)

0 = ρ(u

∗

)δ

∗

−u

∗

, (35)

0 = 1 −Ze[N

1/(1−σ)

p]ψ

(u) −δρ

(u), (36)

0 = 1 −Z

∗

e[N

1/(1−σ)

p]ψ

(u

∗

) −δ

∗

ρ

(u

∗

), (37)

0 = N¯ y −ξ[r

(ξp)(mk +m

∗

k

∗

) +w

(ξp)(ml +m

∗

l

∗

)], (38)

0 = e

[N

1/(1−σ)

p]N

1/(1−σ)

[mψ(u) +m

∗

ψ(u

∗

)] −N¯ y. (39)

For a given p, if

ρ(0) < r(ξp),

then there exists a unique and positive u such that

ρ(u) = r(ξp).

Let u(·) be the inverse function of ρ(·).

10

Since the shadow prices, Z, Z

∗

, δ, δ

∗

,

are derived once the above system of equations determines p, k, k

∗

, N, we see

10

As is clear from (32) and (33), u = u

∗

holds at the steady state in which both countries

are incompletely specialized.

11

that the main system consists of the four equations:

0 = r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ[u(r(ξp))], (40)

0 = r(ξp)k

∗

+w(ξp)l

∗

−e[N

1/(1−σ)

p]ψ[u(r(ξp))], (41)

N¯ y = ξ[r

(ξp)(mk +m

∗

k

∗

) +w

(ξp)(ml +m

∗

l

∗

)], (42)

N¯ y = e

[N

1/(1−σ)

p]N

1/(1−σ)

{mψ[u(r(ξp))] +m

∗

ψ[u(r(ξp))]}. (43)

Now, we can restate Kikuchi and Shimomura (2007)’s result.

Proposition 1: Suppose that diﬀerences in initial factor endowments between

Home and Foreign are not very large and that both the preference of each house-

hold and production technologies take the Cobb-Douglas form. Then there exists

a unique steady state which is saddle-point stable. In the steady state both coun-

tries produce both goods.

Proof: See Appendix.

4 Trade-Pattern Propositions

Let us focus on the Home (gross) excess demand for diﬀerentiated products in

the steady state,

11

ED

1

≡ m{e

[N

1/(1−σ)

p]N

1/(1−σ)

ψ(u) −ξ[r

(ξp)k +w

(ξp)l]}.

Considering the steady-state Home budget constraint, (40), we obtain

k =

eψ −wl

r

.

11

As we will see later, even if ED

1

= 0 holds, there is an incentive for trade due to product

diﬀerentiation. Thus, ED

1

(pED

1

) can be interpreted as the Home gross excess demand for

diﬀerentiated products (the Home excess supply of the consumable capital).

12

Substituting this into the Home excess demand and rearranging, we obtain the

following condition:

ED

1

= (m/p)[eψ(θ

e

−θ

r

) +wl(θ

r

−θ

w

)], (44)

where θ

e

≡ [pe

N

1/(1−σ)

/e], θ

r

≡ (ξpr

/r), and θ

w

≡ (ξpw

/w), respectively.

Following the same procedure, we can obtain the Foreign excess demand for

diﬀerentiated products in the steady state, ED

∗

1

:

ED

∗

1

= (m

∗

/p)[eψ(θ

e

−θ

r

) +wl

∗

(θ

r

−θ

w

)]. (45)

From these excess demand functions, we see that

ED

1

−ED

∗

1

= (1/p)[eψ(θ

e

−θ

r

)(m−m

∗

) +w(θ

r

−θ

w

)(ml −m

∗

l

∗

)]. (46)

Since diﬀerentiated products are assumed to be capital intensive,

θ

r

> 1 > θ

e

> 0 > θ

w

holds. Let us examine the following two cases.

4.1 Case A: m < m

∗

and l = l

∗

If the representative household in each country supplies an equal amount of

labor (l = l

∗

), the gross excess demands for diﬀerentiated products have the

same sign in both countries (see (44) and (45)). Since demands have to add up

to zero, this implies that both of them have to be zero and, therefore, there is no

net trade (the value of imports equals the value of exports) in the diﬀerentiated

products sector. This also implies that there is no incentive for inter-industry

trade (i.e., the exchange of diﬀerentiated products for the consumable capital).

Still, since each country specializes in a diﬀerent range of diﬀerentiated products,

an incentive for intra-industry trade remains. We obtain our main proposition

on the patterns of intra-industry trade.

13

Proposition 2: Suppose that the representative household in each country sup-

plies an equal amount of labor. Then, in the steady state, only intra-industry

trade of diﬀerentiated products between countries occurs irrespective of diﬀer-

ences in the number of households.

This case provides a complementary view for the existence of intra-industry

trade between developed and developing countries. We implicitly assume that

Foreign (the developing country) started the process of development late (i.e.,

its capital stock is relatively low initially). Then, Proposition 1 and Proposition

2 state that Foreign accumulates capital until its capital-labor ratio equals that

of Home.

12

Therefore, due to catching-up by the developing country, sources of

inter-industry trade based on diﬀerences in the capital-labor ratio vanish and

only intra-industry trade occurs in the steady state. Furthermore, since Foreign

has a larger amount of labor, that is, m < m

∗

, its share of diﬀerentiated products

in the world market also becomes larger than Home. Note that, in the steady

state, the share of Foreign varieties [n

∗

/(n+n

∗

)] is equal to the share of Foreign

households [m

∗

/(m+m

∗

)]. Our dynamic model reinforces the role of increasing

returns and monopolistic competition as determinants of intra-industry trade:

the importance of intra-industry trade remains in the dynamic setting while

that of inter-industry trade is downplayed.

4.2 Case B: m = m

∗

and l > l

∗

In this case, each Foreign household is relatively less eﬃcient in providing la-

bor. And also, assume that capital-labor endowment ratio is lower in Foreign

12

This point contrasts sharply with Atkeson and Kehoe (2000), in which the developing

country accumulates capital until its capital-labor ratio equals the ratio used in the rest of

the world to produce the labor-intensive good: the developing country never catches up in this

setting.

14

initially. Since we assume that each household in both countries has the same

instantaneous discount function, u = u

∗

holds at the steady state (see (32) and

(33)). Therefore, from (30) and (31), each Foreign household accumulates more

capital (i.e., k < k

∗

). Then, ED

1

−ED

∗

1

> 0 holds and Foreign becomes a net

exporter of diﬀerentiated products (i.e., capital intensive products) although it

is a labor-rich country at the initial moment.

13

Proposition 3: If the number of households is equal, the country with lower

labor eﬃciency becomes the net exporter of the capital-intensive good.

This case highlights that the source of inter-industry trade crucially depends

on the eﬃciency of each household, not on the number of households. It also

highlights the importance of capital accumulation in dynamic trade patterns.

Again, in this case, Foreign’s share of diﬀerentiated products in the world market

becomes larger than Home’s.

5 Concluding Remarks

Based on the two-sector Chamberlin-Heckscher-Ohlin (CHO) framework, this

note has formulated a dynamic model of international trade by introducing the

Uzawa (1968) endogenous time preferences. Also, in contrast to Kikuchi and

Shimomura (2007), the diﬀerence in the number of households has been em-

phasized. We have shown that there exists a unique and saddlepoint-stable

steady state that is independent of the initial international distribution of capi-

tal. In that steady state production in both countries is incompletely specialized

(Proposition 1). Making use of the new dynamic trade model, we have shown

that, (i) given that the representative household in each country supplies an

13

The case of m < m

∗

and ml > m

∗

l

∗

can be analyzed in a similar way.

15

equal amount of labor (l = l

∗

), only intra-industry trade occurs in the steady

state irrespective of diﬀerences in the number of households (Proposition 2), (ii)

if the number of households is equal, the country with higher labor eﬃciency

becomes the net exporter of the labor-intensive good (Proposition 3). Proposi-

tions 2 and 3 highlight the dominance of the developing country in the world

economy: although its capital-labor ratio is lower than that of the developed

country, capital accumulation makes it a major exporter of diﬀerentiated prod-

ucts. Although our result depends critically on several restrictive assumptions

(e.g., Uzawa’s endogenous time preferences), it establishes a link between the

workhorse model of monopolistic competition and the size of labor endowment.

Hopefully this analysis provides a useful paradigm for considering how the la-

bor endowment of developing countries (e.g., China) works as a determinant of

world trade patterns.

16

References

[1] Atkeson, A. and P. Kehoe (2000) ‘Paths of Development for Early- and Late-

Bloomers in a Dynamic Heckscher-Ohlin Model,’ Research Department Staﬀ

Report 256, Federal Reserve Bank of Minneapolis.

[2] Chen, B.-L., K. Nishimura, and K. Shimomura (2005) ‘An Uzawa-Oniki-

Uzawa Dynamic Two-Country Model of International Trade,’ manuscript,

Kobe University.

[3] Chen, Z. (1992) ‘Long-run Equilibria in a Dynamic Heckscher-Ohlin Model,’

Canadian Journal of Economics 23, 923–943.

[4] Dixit, A. K., and V. Norman (1980) Theory of International Trade, Cam-

bridge University Press.

[5] Dixit, A. K., and J. E. Stiglitz (1977) ‘Monopolistic Competition and Opti-

mum Product Diversity,’ American Economic Review 67, 297–308.

[6] Grossman, G., and E. Helpman (1991) Innovation and Growth in the Global

Economy, The MIT Press: Cambridge MA.

[7] Helpman, E. (1981) ‘International Trade in the Presence of Product Diﬀeren-

tiation, Economies of Scale and Monopolistic Competition: A Chamberlin-

Heckscher-Ohlin Approach,’ Journal of International Economics 11, 305–

340.

[8] Helpman, E., and P. R. Krugman (1985) Market Structure and Foreign

Trade, The MIT Press: Cambridge MA.

[9] Hong, W (1988) ‘Time Preference in Dynamic Trade Models: An Empirical

Critique,’ Economic Development and Cultural Change 36, 741–751.

17

[10] Kikuchi, T., and K. Shimomura (2007) ‘A New Dynamic Trade Model with

Increasing Returns and Monopolistic Competition,’ Review of Development

Economics 11, 232–241.

[11] Lawrence, E.C. (1991) ‘Poverty and the Rate of Time Preference: Evidence

from Panel Data,’ Journal of Political Economy 99, 54–77.

[12] Nishimura, K., and K. Shimomura (2002) ‘Trade and Indeterminacy in a

Dynamic General Equilibrium Model,’ Journal of Economic Theory 105,

244–259.

[13] Nishimura, K., and K. Shimomura (2006) ‘Indeterminacy in a Dynamic

Two-Country Model,’ Economic Theory 29, 307–324.

[14] Ogawa, K. (1993) ‘Economic Development and Time Preference Schedule:

The Case of Japan and East Asian NICs,’ Journal of Development Eco-

nomics 42, 175–195.

[15] Oniki, H., and H. Uzawa (1965) ‘Patterns of Trade and Investment in a

Dynamic Model of International Trade,’ Review of Economic Studies 32,

15–38.

[16] Shimomura, K. (1992) ‘A Two-Sector Dynamic General Equilibrium Model

of Distribution,’ in G. Feichtinger, ed., Dynamic Economic Models and Op-

timal Control, North-Holland, 105-123.

[17] Shimomura, K. (1993) ‘Durable Consumption Goods and the Pattern of

International Trade,’ in H. Herberg and N. V. Long, eds., Trade, Welfare,

and Economic Policies: Essays in Honor of Murray C. Kemp, Michigan

University Press, 103–112.

[18] Shimomura, K. (2004) ‘Indeterminacy in a Dynamic General Equilibrium

Model of International Trade,’ in M. Boldrin, B.-L. Chen and P. Wang, eds.,

18

The Development Process of Rapidly Growing Economies: From Theory to

Empirics, Cheltenham, UK, Edward Elgar Publishing Inc. Chapter 7, 153–

167.

[19] Stiglitz, J. E. (1970) ‘Factor Price Equalization in a Dynamic Economy,’

Journal of Political Economy 78, 456–488.

[20] Uzawa, H. (1968) ‘Time Preferences, the Consumption Function, and Opti-

mal Asset Holdings,’ in J. N. Wolfe, ed., Value, Capital and Growth: Papers

in Honour of Sir John Hicks, University of Edinburgh Press: Edinburgh,

485–504.

[21] Ventura, J. (1997) ‘Growth and Interdependence,’ Quarterly Journal of

Economics 112, 57–84.

19

6 Appendix: Existence, Uniqueness and Stabil-

ity of the Steady State with Incomplete Spe-

cialization in Both Countries

Here, we shall prove the existence, uniqueness and stability of the steady state

with incomplete specialization in the present two-country dynamic general equi-

librium model. We shall focus on the symmetric case where preferences, tech-

nologies, and initial factor endowments are common between Home and Foreign

(m = m

∗

= 1, l = l

∗

). As we shall show later, the determinant of the Jacobian

at a symmetrical steady state is not zero, which implies that as long as the

international diﬀerences in those economic fundamentals are not very large, the

existence, uniqueness and stability are guaranteed.

6.1 Existence

Let us consider the existence of the steady state. Since we assume l = l

∗

, it

is clear from (30)-(33) that k = k

∗

holds at the steady state. Therefore, the

system of equations which describes the steady-state k, p, and N becomes

0 = r(ξp)k +w(ξp)l −e[N

1/(1−σ)

p]ψ[u(r(ξp))], (47)

N¯ y = 2ξ[r

(ξp)k +w

(ξp)l], (48)

N¯ y = 2e

[N

1/(1−σ)

p]N

1/(1−σ)

ψ[u(r(ξp))]. (49)

From (47),

k =

eψ −wl

r

(50)

holds. Combining (48) −(49), one can obtain

0 = e

N

1/(1−σ)

ψ −ξ(r

k +w

l).

20

Substituting (50) into this, one can obtain

0 = e

N

1/(1−σ)

ψ −ξ

¸

r

r

(eψ −wl) +w

l

.

Multiplying p and rewriting this in terms of elasticity,

0 = θ

e

eψ −[θ

r

(eψ −wl) +θ

w

wl],

where θ

e

= [pe

N

1/(1−σ)

/e], θ

r

= (ξpr

/r), and θ

w

= (ξpw

/w), respectively.

Rearranging this, we obtain:

eψ = Θwl, (51)

where Θ ≡ [(θ

r

−θ

w

)/(θ

r

−θ

e

)], which is greater than 1.

Next, multiplying p to (49), one can obtain

pN¯ y = 2eψ

pe

N

1/(1−σ)

e

or N =

2θ

e

p¯ y

eψ. (52)

Substituting (51) into (52), one can obtain

N =

2θ

e

p¯ y

Θw(ξp)l. (53)

In terms of proportional change, we obtain the ﬁrst relationship between N and

p:

14

ˆ

N

ˆ p

= θ

w

−1. (54)

Since the diﬀerentiated products are capital-intensive, that is, θ

w

< 0, (54)

implies that N is decreasing in p: we can depict (53) as Curve AA in Figure

1.

15

14

Note that θ

e

(θ

r

and θ

w

) is constant when the preference (the production technologies)

takes the Cobb-Douglas form.

15

It can be easily shown that the right-hand side of (53) goes to ∞ (0) when p goes to 0

(∞).

21

Now, let us turn to the other condition. From (51),

e[N

1/(1−σ)

p] =

Θw(ξp)l

ψ[u(r(ξp))]

.

Let e

−1

≡ β, then we can obtain:

N

1/(1−σ)

p = β

Θw(ξp)l

ψ[u(r(ξp))]

.

Rearranging this, one can obtain

N =

¸

p

β(Θw(ξp)l/ψ[u(r(ξp))])

σ−1

. (55)

In terms of proportional change, we obtain the second relationship between N

and p.

ˆ

N

ˆ p

= (σ −1)

¸

1 −

1

θ

e

θ

w

−

θ

r

θ

ψ

θ

ρ

, (56)

where θ

ψ

≡ [uψ

(u)/ψ(u)] and θ

ρ

= [uρ

(u)/ρ(u)]. Since θ

ψ

and θ

ρ

are positive,

(56) implies that N is increasing in p: we can depict (55) as Curve BB in Figure

1.

16

Based on the foregoing argument, one can conclude as follows.

17

LEMMA A1: There uniquely exists a steady state in which production is

incompletely specialized.

16

Let us deﬁne p as the solution of u(r(ξp)) = 0 ⇐⇒ r(ξp) = ρ(0). Then,

lim

p→p

β(Θw(ξp)l/ψ[u(r(ξp))]) = ∞, which implies that the right-hand side of (55) goes

to 0 when p goes to p. On the other hand, the right-hand side of (55) goes to ∞ when p goes

to ∞.

17

It is apparent from (50) and (51) that the steady-state capital stock is positive.

22

OOO

p

N

p

p

N

A

A

B

B

Figure 1

6.2 The Non-Existence of the Steady State with Complete

Specialization in Home and/or Foreign

Now, what remains to be argued concerning uniqueness is to exclude a steady

state where at least one country is completely specialized. For this purpose, let

us consider the whole GDP function. In the case where the diﬀerentiated prod-

ucts are more capital-intensive than the homogeneous good, it can be expressed

as follows.

F(k, ξp) =

f

2

(k, l), 0 < k < k

2

(ξp),

r(ξp)k +w(ξp)l, k

2

(ξp) < k < k

1

(ξp),

pf

1

(k, l, ¯ y), k > k

1

(ξp),

where k

i

(ξp) ≡ l{c

i

r

[w(ξp), r(ξp)]/c

i

w

[w(ξp), r(ξp)]}, and f

1

(k, l, ¯ y) ≡ ¯ yn(k, l, ¯ y).

18

Making use of the above GDP function, we can express the steady-state Home

18

For the derivation of the monopolistically competitive industry’s implicit production func-

tion, f

1

, see Helpman and Krugman (1985, p. 139).

23

and Foreign budget constraints as

0 = F(k, ξp) −e[N

1/(1−σ)

p]ψ[u(F

k

(k, ξp))], (57)

0 = F(k

∗

, ξp) −e[N

1/(1−σ)

p]ψ[u(F

k

(k

∗

, ξp))]. (58)

If k > k

∗

, then both F(k, ξp) > F(k

∗

, ξp) and F

k

(k, ξp) ≤ F

k

(k

∗

, ξp) hold from

properties of the GDP function and vice versa. Thus, (57) and (58) together

imply that there is no steady state such that k = k

∗

holds. Therefore, we can

conclude as follows.

LEMMA A2: When the two countries are suﬃciently close in terms of factor

endowment ratio, no country can specialize in producing only one good in the

steady state.

6.3 Local Saddlepoint-Stability

Let us assume that the two countries are identical. Let us consider the Jacobian

matrix of the steady state,

ρ 0 0 0 0 0 −eψ

0

p¯ y

2(σ−1)

0

0 ρ 0 0 0 0 0 −eψ

p¯ y

2(σ−1)

0

0 0 0 0 0 0 Zρ

0 0 −Zξr

0 0 0 0 0 0 0 Zρ

0 −Zξr

0 0 0 0 ρ 0 −Zeψ

0 0 0

0 0 0 0 0 ρ 0 −Zeψ

0 0

0 0 −eψ

0 −ρ

0

−Zeψ

−δρ

0

Zp¯ yψ

2(σ−1)ψ

−

Z¯ yNψ

2ψ

0 0 0 −eψ

0 −ρ

0

−Zeψ

−δρ

Zp¯ yψ

2(σ−1)ψ

−

Z¯ yNψ

2ψ

−ξr

−ξr

0 0 0 0 0 0 ¯ y −2ξ

2

(r

k +w

l)

0 0 0 0 0 0

¯ yNψ

2ψ

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

¯ y 2ψe

N

2/(1−σ)

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

¸

.

24

Denote the above matrix by J, and the corresponding eigenvalue as x. Then

x is determined by the characteristic equation Ω(x) = |J − xI| = 0, where

I ≡

I

6

0

0 O

4

¸

¸

¸.

Let us make the following calculations to obtain the above determinant.

First, let us add both the ﬁrst row multiplied by ξr

**/(ρ−x) and the second row
**

multiplied by ξr

**/(ρ−x) to the 9th row. Next, the 7th row minus the third row
**

multiplied by eψ

/x, and the 8th row minus the 4th row multiplied by eψ

/x.

Finally, we add the 5th row multiplied by ρ

**/(ρ − x) to the 7th row, and add
**

the 6th row multiplied by ρ

**/(ρ −x) to the 8th row. Then, we see that
**

Ω(x) = (ρ −x)

4

x

2

Ξ(x)

x(ρ−x)

0

pZ¯ yψ

2(σ−1)ψ

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

0

Ξ(x)

x(ρ−x)

pZ¯ yψ

2(σ−1)ψ

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

−

ξr

eψ

ρ−x

−

ξr

eψ

ρ−x

¯ y +

pξr

¯ y

(ρ−x)(σ−1)

−2ξ

2

(r

k +w

l)

¯ yNψ

2ψ

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

¯ y 2ψe

N

2/(1−σ)

= (ρ −x)

×

Ξ(x) 0

(ρ−x)xpZ¯ yψ

2(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

0 Ξ(x)

(ρ−x)xpZ¯ yψ

2(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

−ξr

eψ

−ξr

eψ

(ρ −x)¯ y +

pξr

¯ y

σ−1

−2(ρ −x)ξ

2

(r

k +w

l)

¯ yNψ

2ψ

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

¯ y 2ψe

N

2/(1−σ)

,

where Ξ(x) ≡ (Zeψ

+δρ

)x

2

−ρ(Zeψ

+δρ

)x −Zeψ

ρ

ρ.

When the ﬁrst column is subtracted from the second column, we obtain

Ω(x) = (ρ −x)Ξ(x)

×

1 0

(ρ−x)xpZ¯ yψ

2(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

−1 Ξ(x)

(ρ−x)xpZ¯ yψ

2(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

0 −ξr

eψ

(ρ −x)¯ y +

pξr

¯ y

σ−1

−2(ρ −x)ξ

2

(r

k +w

l)

0

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

¯ y 2ψe

N

2/(1−σ)

.

For Ω(x) = 0, we have x

1

= ρ > 0. Furthermore, from Ξ(x) = 0, we have

25

x

2

> 0 > x

3

such that Ω(x

2

) = Ω(x

3

) = 0, since all of the ﬁrst and second

derivatives of ψ and ρ are positive.

Next, when the 1st row is added to the 2nd row, we obtain

Ω(x) = (ρ −x)Ξ(x)

×

1 0

(ρ−x)xpZ¯ yψ

2(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

2ψ

+

eψ

Zξr

x

0 Ξ(x)

(ρ−x)xpZ¯ yψ

(σ−1)ψ

(ρ −x)x

−

Z¯ yNψ

ψ

+

2eψ

Zξr

x

0 −ξr

eψ

(ρ −x)¯ y +

pξr

¯ y

σ−1

−2(ρ −x)ξ

2

(r

k +w

l)

0

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

¯ y 2ψe

N

2/(1−σ)

= (ρ −x)Ξ(x)¯ y

×

Ξ(x)

(ρ−x)xpZψ

(σ−1)ψ

(ρ −x)

−

xZ¯ yNψ

ψ

+ 2eψ

Zξr

−ξr

eψ

ρ −x +

pξr

σ−1

−2(ρ −x)ξ

2

(r

k +w

l)

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

2ψe

N

2/(1−σ)

≡ (ρ −x)Ξ(x)¯ yA(x).

According to A(x), the term on x

3

becomes

−(Zeψ

+δρ

)2ψe

N

2/(1−σ)

−

¯ yNψ

2ψ

·

pZψ

(σ −1)ψ

2ξ

2

(r

k +w

l)

+

¯ yNψ

2ψ

·

¯ yNZψ

ψ

+ (Zeψ

+δρ

)2ξ

2

(r

k +w

l)

θ

e

σ −1

+ 1

,

where only the second term is negative, since the term, r

k + w

l, is positive

due to the convexity of the GDP function with respect to p. Therefore, the term

on x

3

becomes positive if σ is suﬃciently large, which we assume.

19

Then, if

19

Indeed, it can be shown that the sum of the second term and the third one becomes

positive if σ > [(θ

r

r

k + θ

w

w

l)/(r

k + w

l)].

26

A(0) > 0 holds, A(x) = 0 has one negative root x

4

.

A(0) = 2eψ

ρ

×

−Zρ

0 Zξr

−ξr

eψ

ρ +

pξr

σ−1

−ρξ

2

(r

k +w

l)

¯ yNψ

2ψ

−

θ

e

σ−1

+ 1

ψe

N

2/(1−σ)

= 2eψ

ρZ

×

−ρ

ρ

1 +

θ

r

σ−1

ψe

N

2/(1−σ)

+ (ξr

)

2

eψ

θ

e

σ−1

+ 1

−ξr

ρ

1 +

θ

r

σ−1

e

N

1/(1−σ)

ψ

+ ρ

ρξ

2

(r

k +w

l)

θ

e

σ−1

+ 1

¸

¸

¸,

which is positive, since the sum of the second term and the third one in the

square brackets becomes

ξr

eψ

ρ

p

¸

pξr

r

θ

e

σ −1

+ 1

−

pe

N

1/(1−σ)

e

1 +

θ

r

σ −1

=

ξr

eψ

ρ

p

(θ

r

−θ

e

)

> 0.

Therefore, there are two negative characteristic roots, i.e., x

3

and x

4

. Since

there are two state variables, k and k

∗

, it follows that the steady state is a

saddle point.

LEMMA A3: When the two countries are suﬃciently close, the steady state

with both countries being incompletely specialized is locally saddlepoint-stable.

27

**A Dynamic Chamberlin-Heckscher-Ohlin Model with Endogenous Time Preferences: A Note∗
**

Kazumichi Iwasa† Toru Kikuchi‡ September 20, 2007 Koji Shimomura§

Abstract This note formulates a dynamic two-country (developed and developing countries) Chamberlin-Heckscher-Ohlin model of trade with endogenous time preferences a la Uzawa (1968). We examine the relationship between initial factor endowment diﬀerences and trade patterns in the steady state. In particular, to highlight the integration of developing countries (e.g., China) into the world trading system, we concentrate on the case of asymmetric size of two countries (in terms of population). It will be shown that (i) given that the representative household in each country supplies

∗ We

are grateful to the Associate Editor and two annonymous referees for helpful com-

ments. We acknowledge ﬁnancial support from the Ministry of Education, Culture, Sports, Science and Technology of Japan (the Grant-in-Aid for the 21st Century Center of Excellence Project ‘Research and Education Center of New Japanese Economic Paradigm’). † Graduate School of Economics, Kobe University, 2-1 Rokkodai-cho, Nada-ku, Kobe 6578501, Japan ‡ Corresponding author, Graduate School of Economics, Kobe University, 2-1 Rokkodaicho, Nada-ku, Kobe 657-8501, Japan; Tel: 81-78-803-6838; Fax: 81-78-803-6838; e-mail: kikuchi@econ.kobe-u.ac.jp § Research Institute for Economics and Business Administration, Kobe University, 2-1 Rokkodai-cho, Nada-ku, Kobe 657-8501, Japan

1

only intra-industry trade occurs in the steady state irrespective of diﬀerences in the number of representative households and that (ii) the number of households being equal. the country with less labor eﬃciency becomes the net exporter of the capital-intensive good. trade patterns 2 . developed and developing countries.an equal amount of labor. JEL Classiﬁcation Code: F12 Key Words: dynamic Chamberlin-Heckscher-Ohlin model.

in which there is a monopolistically competitive ‘diﬀerentiated products’ sector. Lawrence (1991) and Ogawa (1993).3 Consider the world economy as consisting of one developed country 1 This 2A was pointed out by Stiglitz (1970.1 The state of the art in dynamic trade theory is apparently unsatisfactory. How does the integration of developing countries into the world economy aﬀect world trading patterns? It seems to be very important to consider this problem in a dynamic HeckscherOhlin trade model. at least one of the two countries should specialize in one of the two goods and it is very diﬃcult to derive satisfactory results on trade patterns.2 Thus. 3 The static Chamberlin-Heckscher-Ohlin model has been extensively investigated. In other words. However. As an example. and a perfectly competitive ‘consumable capital’ sector. p. we address the question of developing countries’ integration in a dynamic Chamberlin-Heckscher-Ohlin (CHO) model with endogenous time preferences a la Uzawa (1968). under exogenous time preferences.463). many developing countries have opened their economies to international trade. the dynamic Heckscher-Ohlin theorem under the assumption of exogenous time preference that was proved by Chen (1992) holds only if preferences and technologies are strictly identical among countries. while the static Heckscher-Ohlin theorem holds even if preferences and technologies are slightly diﬀerent among countries. Helpman’s (1981) seminal integration of the monopolistic competition trade model into the two- 3 .1 Introduction In recent decades. This seems to suggest that the traditional focus on exogenous time preferences should be accompanied by a focus on endogenous time preferences. China’s integration into the world economy is one of the most important developments aﬀecting the structure and evolution of the global trading system at the dawn of the 21st century. non-constant time preference rate has been empirically documented through panel data and cross-country data by Hong (1988).

4 . Even if there country by two-factor by two-good Heckscher-Ohlin (HO) framework. saving rates are endogenously determined. In this note.e. Ventura (1997). including Stiglitz (1970). China’s population is 20 percent of the world population. Then China’s decision to join the world trading system represents the opening of trade between Home and Foreign.. it is important to consider the case of the asymmetric size of countries. Thus the role of size diﬀerences in factor endowment is downplayed in the analyses. there is a signiﬁcant size diﬀerence between developed and developing countries. most subsequent contributions. 5 The literature on dynamic two-country models originated in Oniki and Uzawa (1965). given that the representative household in each country supplies an equal amount of labor.e. 1993. 4 Atkeson and Kehoe (2000) examine a similar problem using a dynamic Heckscher-Ohlin model composed of a larger number of small open economies. 2004). For example. only intra-industry trade occurs in the steady state irrespective of diﬀerences in the number of representative households. The developed country reached a steady state before the developing country (which corresponds to China) started the process of development (i. which was extended and made popular by Helpman and Krugman (1985). Nishimura and Shimomura (2002. We demonstrate that. In the real world. we extend the analysis of Kikuchi and Shimomura (2007) to the case of asymmetric size of two countries (in terms of population). Thus. i. Shimomura (1992.. that both countries are endowed with an equal number of households. assume that households maximize their discounted sum of utility. has led to the widely held belief that HO and Chamberlinian monopolistic competition are complementary in nature. the removal of trade barriers). To our knowledge. Nishimura and Shimomura (2005) discuss other major problems within dynamic HO models. little attention has been given to the relationship between timing of development and the size of developing countries. we call the former Home and the latter Foreign. Chen. While they assume exogenous saving rate in each trading country. Kikuchi and Shimomura (2007) examine a similar problem using a dynamic two-country Chamberlin-Heckscher-Ohlin model. Chen (1992).45 They assume. 2006). however. For simplicity.and one developing country.

Home and Foreign. m∗ ). Foreign) population is m (resp. Labor is measured in eﬃciency units. Foreign) representative household supplies l (l∗ ) units of efﬁciency labor. diﬀerentiated products (Good 1) and a consumable capital (Good 2). we assume that there is no international credit market. due to catching-up by the developing country. 5 . the Home (resp. This note is organized as follows. Section 2 sets up a dynamic CHO model and Section 3 discusses the existence. Moreover. Following the standard trade theory. and a primary and time-invariant factor of production. The population of each country is assumed to be constant over time. The Home (resp. while there is a competitive domestic credit market in each country. l (labor). Note that Kikuchi and Shimomura (2007)’s case corresponds to m = m∗ = 1. sources of inter-industry trade based on diﬀerences in the capital-labor ratio vanish and only intra-industry trade occurs in the steady state. k. The consumable capital can be either consumed as a non-durable good or added to the existing capital stock. 2 The Model Consider a world economy consisting of two countries. Thus.is a larger amount of labor (in terms of population) in the developing country. Foreign) household is endowed with ml and mk (resp. produced using reproducible capital. Section 5 provides concluding remarks. uniqueness and local stability of the steady state. m∗ l∗ and m∗ k ∗ ) units of factors of production. There are two types of commodities. in order to focus on international trade. that diﬀer in their factor endowments. we assume away international factor movements. Each Home (resp. Section 4 derives trade-pattern propositions.

du du2 (3) 0 < θρ ≡ [uρ (u)/ρ(u)] < 1 for any positive u < ∞. s. we assume that the variable discount rate ρ(u) satisﬁes ρ(0) > 0. Solving the static expenditure minimizing problem. ρ (u) ≡ > 0. and σ is the elasticity of substitution between varieties. σ > 1. ∞ ∞ uXdt 0 = 0 f [U (V. ρ (u) ≡ dρ(u) d2 ρ(u) > 0. we can deﬁne the expenditure function as e(P )ψ(u) ≡ (minP V + C2 .t. Quantity index V takes the following Dixit-Stiglitz (1977) form: N σ/(σ−1) (4) V = 0 x(i) (σ−1)/σ di . u = f [U (V. (6) 6 . where V is the quantity index for diﬀerentiated products. x(i) is the consumption of the i-th variety of diﬀerentiated products.. f (U ) < 0. Following Uzawa (1968).Each consumer maximizes the discounted sum of utility. (5) where N is the total number of diﬀerentiated products. (1) (2) ˙ X = −ρ(u)X. C2 is the consumption of the consumable capital. It will be assumed that U is linearly homogeneous in its arguments and f satisﬁes f (0) = 0. and X ≡ exp{− t 0 ρ(u)dτ } is the discount factor at time t which depends on the past and present level of utility through the function ρ. f (U ) > 0. C2 )] Xdt. C2 )]) .

N ]. ∂e(P )ψ(u)/∂P = V . 7 . and the ﬁrst order condition. r)φ [x(i)]. Assume that diﬀerentiated products are more capital-intensive than the consumable capital. s.t. where y is the output level of each ﬁrm. p(i) = p and x(i) = x for ∀ i ∈ [0. r)φ[x(i)]. and ψ(u) is the inverse function of f . Considering the subutility maximization problem: RN max V. yields p(i) [1 − (1/σ)] = c1 (w. 7 We can obtain this relation as follows. Therefore. 0 p(i)x(i)di ≤ I.where the consumable capital serves as the numeraire. we can obtain the following condition from the envelope theorem.7 Furthermore. Assume that each ﬁrm in the diﬀerentiated products sector has the homothetic total cost function c1 (w. There are signiﬁcant economies of scale: φ(y)/y is decreasing over the relevant range of output levels y. r)φ (y). that is. r)φ[x(i)] [P (σ−1) Ix(i)(σ−1) ] 1/σ − c1 (w.. while the consumable capital is produced by competitive ﬁrms under constant returns technology. ψ (u) > 0. ψ (u) > 0. free entry implies that price equals 6 This assumption is just for simpliﬁcation and this capital intensity ranking itself does not alter the results of this paper.6 Diﬀerentiated products are produced by monopolistically competitive ﬁrms under increasing returns technology. we obtain the inverse demand function of i-th variety as follows: p(i) = [P (σ−1) I/x(i)] πi = = 1/σ . the revenue of the i-th ﬁrm is given by p(i)x(i) − c1 (w. dπi /dx(i) = 0. P ≡ N 0 p(i) 1−σ di 1/(1−σ) is the price index for diﬀerentiated products. e [N 1/(1−σ) p]ψ(u) = N σ/(σ−1) x or N 1/(1−σ) e [N 1/(1−σ) p]ψ(u) = N x. which clearly satisﬁes ψ(0) = 0. The marginal revenue will be equated to the marginal cost: p [1 − (1/σ)] = c1 (w. r)φ(y). (7) Given that the equilibrium is symmetric.

σ φ(¯) y The constraints on labor and capital within Home are9 c1 (w. See Dixit and Norman (1980. (10) (11) 1 = and we can obtain the factor price functions w(ξp) and r(ξp). the zero-proﬁt conditions can be written as ¯ y ξp = c1 (w. σ dy where θφ ≡ [yφ (y)/φ(y)]. 284–5). which is deﬁned ¯ by8 1− 1 y φ (¯) ¯ y = .average cost: p = [c1 (w. By combining these conditions. To guarantee the existence and uniqueness of y . An example of φ(y) is ln(y + 1). any country’s population size does not aﬀect its relative factor abundance in the static sense. r) and y2 are the unit cost function and the output of the consumable capital. (12) The partial derivative of the national income with respect to the price of diﬀer- 8 This result depends crucially on homotheticity in production. r)y2 y w w c1 (w. c2 (w. r). by deﬁning ξ ≡ y /φ(¯). r)φ(¯)n + c2 (w. Utilizing these factor price functions. r)φ(y)]/y. respectively. 9 As it is clear from these equations. and < 0 for any positive y < ∞. mk. 8 . pp. we assume that φ satisﬁes ¯ φ (0) < ∞. r)φ(¯)n + c2 (w. one can easily see that all varieties will have the same output level y . r)y2 y r r = = ml. (8) (9) where n is the number of diﬀerentiated products produced in Home and c2 (w. r). φ (0) > −∞. Our aim is to check whether population size aﬀects longrun capital accumulation. the national income is shown as r(ξp)mk + w(ξp)ml. Then. lim θφ < 1 − y→∞ dθφ 1 .

−λr. we can obtain another condition for each household: ˙ k = r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ(u). is equal to the aggregate national output of those products: n¯ = ξr (ξp)mk + ξw (ξp)ml. we can obtain ˙ Z = Z[ρ(u) − r(ξp)]. ρ(u)δ − u. our dynamic general equilibrium two- 9 . Associated with this problem is the Hamiltonian H ≡ uX + λ{r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ(u)} − δρ(u)X. (16) (17) (18) Letting Z ≡ λ/X and combining (2) and (17). (14) (13) Each household maximizes (1) subject to both (2) and (14). p. (19) Based on the foregoing argument.entiated products. (15) where λ and δ are the shadow prices of k and X. y From (12). The necessary conditions for optimality are 0 = ˙ λ ˙ δ = = X − λe[N 1/(1−σ) p]ψ (u) − δρ (u)X.

u. δ ∗ .country model is described as ˙ k ˙ k∗ ˙ Z ˙ Z∗ ˙ δ δ˙∗ = = = = = = r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ(u). 10 . Z ∗ . 1 − Z ∗ e[N 1/(1−σ) p]ψ (u∗ ) − δ ∗ ρ (u∗ ). p. N y − ξ[r (ξp)(mk + m∗ k ∗ ) + w (ξp)(ml + m∗ l∗ )]. ¯ (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) 0 = 0 = 0 = 0 = The system determines the equilibrium path of two state variables. k and k ∗ . ¯ e [N 1/(1−σ) p]N 1/(1−σ) [mψ(u) + m∗ ψ(u∗ )] − N y . and N . Z. ρ(u∗ )δ ∗ − u∗ . Z ∗ [ρ(u∗ ) − r(ξp)]. 1 − Ze[N 1/(1−σ) p]ψ (u) − δρ (u). and eight jump variables. Z[ρ(u) − r(ξp)]. ρ(u)δ − u. δ. u∗ . r(ξp)k ∗ + w(ξp)l∗ − e[N 1/(1−σ) p]ψ(u∗ ).

3 The Steady State The steady state is the solution for the system of equations 0 = 0 = 0 = 0 = 0 = 0 = r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ(u). N y − ξ[r (ξp)(mk + m∗ k ∗ ) + w (ξp)(ml + m∗ l∗ )]. are derived once the above system of equations determines p. ¯ Let u(·) be the inverse function of ρ(·). Z ∗ . we see 10 As is clear from (32) and (33). ρ(u∗ ) − r(ξp). ρ(u∗ )δ ∗ − u∗ . k ∗ . (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) 0 = 1 − Ze[N 1/(1−σ) p]ψ (u) − δρ (u).10 Since the shadow prices. if ρ(0) < r(ξp). Z. 0 = 1 − Z ∗ e[N 1/(1−σ) p]ψ (u∗ ) − δ ∗ ρ (u∗ ). ¯ e [N 1/(1−σ) p]N 1/(1−σ) [mψ(u) + m∗ ψ(u∗ )] − N y . then there exists a unique and positive u such that ρ(u) = r(ξp). N . r(ξp)k ∗ + w(ξp)l∗ − e[N 1/(1−σ) p]ψ(u∗ ). u = u∗ holds at the steady state in which both countries are incompletely specialized. ρ(u) − r(ξp). ρ(u)δ − u. δ ∗ . 11 . 0 = 0 = For a given p. δ. k.

12 . (40) (41) (42) (43) N y = ξ[r (ξp)(mk + m∗ k ∗ ) + w (ξp)(ml + m∗ l∗ )].11 ED1 ≡ m{e [N 1/(1−σ) p]N 1/(1−σ) ψ(u) − ξ[r (ξp)k + w (ξp)l]}. there is an incentive for trade due to product diﬀerentiation. Proposition 1: Suppose that diﬀerences in initial factor endowments between Home and Foreign are not very large and that both the preference of each household and production technologies take the Cobb-Douglas form. ¯ Now. (40). r(ξp)k ∗ + w(ξp)l∗ − e[N 1/(1−σ) p]ψ[u(r(ξp))]. r we will see later. Proof: See Appendix. ED1 (pED1 ) can be interpreted as the Home gross excess demand for diﬀerentiated products (the Home excess supply of the consumable capital).that the main system consists of the four equations: 0 = 0 = r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ[u(r(ξp))]. we can restate Kikuchi and Shimomura (2007)’s result. 4 Trade-Pattern Propositions Let us focus on the Home (gross) excess demand for diﬀerentiated products in the steady state. Then there exists a unique steady state which is saddle-point stable. Considering the steady-state Home budget constraint. even if ED1 = 0 holds. we obtain k= 11 As eψ − wl . In the steady state both countries produce both goods. Thus. ¯ N y = e [N 1/(1−σ) p]N 1/(1−σ) {mψ[u(r(ξp))] + m∗ ψ[u(r(ξp))]}.

13 . respectively. Still. (46) Since diﬀerentiated products are assumed to be capital intensive. θr > 1 > θe > 0 > θw holds.. (45) From these excess demand functions.1 Case A: m < m∗ and l = l∗ If the representative household in each country supplies an equal amount of labor (l = l∗ ). (44) where θe ≡ [pe N 1/(1−σ) /e]. therefore. the exchange of diﬀerentiated products for the consumable capital). Following the same procedure. This also implies that there is no incentive for inter-industry trade (i. we can obtain the Foreign excess demand for ∗ diﬀerentiated products in the steady state. ED1 : ∗ ED1 = (m∗ /p)[eψ(θe − θr ) + wl∗ (θr − θw )]. θr ≡ (ξpr /r).Substituting this into the Home excess demand and rearranging. there is no net trade (the value of imports equals the value of exports) in the diﬀerentiated products sector.e. Since demands have to add up to zero. 4. we obtain the following condition: ED1 = (m/p)[eψ(θe − θr ) + wl(θr − θw )]. the gross excess demands for diﬀerentiated products have the same sign in both countries (see (44) and (45)). since each country specializes in a diﬀerent range of diﬀerentiated products. this implies that both of them have to be zero and. we see that ∗ ED1 − ED1 = (1/p)[eψ(θe − θr )(m − m∗ ) + w(θr − θw )(ml − m∗ l∗ )]. an incentive for intra-industry trade remains. We obtain our main proposition on the patterns of intra-industry trade. and θw ≡ (ξpw /w). Let us examine the following two cases.

e. in the steady state. in the steady state. Furthermore. 4. Proposition 1 and Proposition 2 state that Foreign accumulates capital until its capital-labor ratio equals that of Home. only intra-industry trade of diﬀerentiated products between countries occurs irrespective of diﬀerences in the number of households.. 14 . m < m∗ . This case provides a complementary view for the existence of intra-industry trade between developed and developing countries. And also. each Foreign household is relatively less eﬃcient in providing labor. Our dynamic model reinforces the role of increasing returns and monopolistic competition as determinants of intra-industry trade: the importance of intra-industry trade remains in the dynamic setting while that of inter-industry trade is downplayed. the share of Foreign varieties [n∗ /(n + n∗ )] is equal to the share of Foreign households [m∗ /(m + m∗ )]. its capital stock is relatively low initially). that is. Then.12 Therefore. assume that capital-labor endowment ratio is lower in Foreign 12 This point contrasts sharply with Atkeson and Kehoe (2000). due to catching-up by the developing country. since Foreign has a larger amount of labor.2 Case B: m = m∗ and l > l∗ In this case. in which the developing country accumulates capital until its capital-labor ratio equals the ratio used in the rest of the world to produce the labor-intensive good: the developing country never catches up in this setting. its share of diﬀerentiated products in the world market also becomes larger than Home.Proposition 2: Suppose that the representative household in each country supplies an equal amount of labor. We implicitly assume that Foreign (the developing country) started the process of development late (i. Note that. sources of inter-industry trade based on diﬀerences in the capital-labor ratio vanish and only intra-industry trade occurs in the steady state. Then.

. We have shown that there exists a unique and saddlepoint-stable steady state that is independent of the initial international distribution of capital. 5 Concluding Remarks Based on the two-sector Chamberlin-Heckscher-Ohlin (CHO) framework. we have shown that. Also. (i) given that the representative household in each country supplies an 13 The case of m < m∗ and ml > m∗ l∗ can be analyzed in a similar way.e. k < k ∗ ). It also highlights the importance of capital accumulation in dynamic trade patterns. the diﬀerence in the number of households has been emphasized. 15 . this note has formulated a dynamic model of international trade by introducing the Uzawa (1968) endogenous time preferences.13 Proposition 3: If the number of households is equal. This case highlights that the source of inter-industry trade crucially depends on the eﬃciency of each household. each Foreign household accumulates more ∗ capital (i.e. the country with lower labor eﬃciency becomes the net exporter of the capital-intensive good. Again. Making use of the new dynamic trade model. Foreign’s share of diﬀerentiated products in the world market becomes larger than Home’s. from (30) and (31). in contrast to Kikuchi and Shimomura (2007).initially.. Then. not on the number of households. in this case. capital intensive products) although it is a labor-rich country at the initial moment. Since we assume that each household in both countries has the same instantaneous discount function. Therefore. u = u∗ holds at the steady state (see (32) and (33)). ED1 − ED1 > 0 holds and Foreign becomes a net exporter of diﬀerentiated products (i. In that steady state production in both countries is incompletely specialized (Proposition 1).

Hopefully this analysis provides a useful paradigm for considering how the labor endowment of developing countries (e.g... capital accumulation makes it a major exporter of diﬀerentiated products. Uzawa’s endogenous time preferences). (ii) if the number of households is equal. Propositions 2 and 3 highlight the dominance of the developing country in the world economy: although its capital-labor ratio is lower than that of the developed country. 16 . only intra-industry trade occurs in the steady state irrespective of diﬀerences in the number of households (Proposition 2).g. the country with higher labor eﬃciency becomes the net exporter of the labor-intensive good (Proposition 3). Although our result depends critically on several restrictive assumptions (e. it establishes a link between the workhorse model of monopolistic competition and the size of labor endowment.equal amount of labor (l = l∗ ). China) works as a determinant of world trade patterns.

741–751.’ Journal of International Economics 11. E. Kobe University. B. Shimomura (2005) ‘An Uzawa-OnikiUzawa Dynamic Two-Country Model of International Trade. Krugman (1985) Market Structure and Foreign Trade. 17 . G. The MIT Press: Cambridge MA. Norman (1980) Theory of International Trade. A. The MIT Press: Cambridge MA. Stiglitz (1977) ‘Monopolistic Competition and Optimum Product Diversity. E. R. [7] Helpman. and E.References [1] Atkeson. Helpman (1991) Innovation and Growth in the Global Economy.. W (1988) ‘Time Preference in Dynamic Trade Models: An Empirical Critique. [3] Chen..’ Canadian Journal of Economics 23. Federal Reserve Bank of Minneapolis. A.. and J. Kehoe (2000) ‘Paths of Development for Early. and P. 305– 340. Economies of Scale and Monopolistic Competition: A ChamberlinHeckscher-Ohlin Approach. Cambridge University Press. Z. 297–308.’ Research Department Staﬀ Report 256. K. [4] Dixit.’ manuscript. [8] Helpman. and V. and P. (1992) ‘Long-run Equilibria in a Dynamic Heckscher-Ohlin Model..’ Economic Development and Cultural Change 36. [6] Grossman.. [2] Chen.and LateBloomers in a Dynamic Heckscher-Ohlin Model. [9] Hong. A. 923–943.’ American Economic Review 67.-L. K. and K. K. (1981) ‘International Trade in the Presence of Product Diﬀerentiation. [5] Dixit. Nishimura. E.

Shimomura (2007) ‘A New Dynamic Trade Model with Increasing Returns and Monopolistic Competition. and K. Long. Uzawa (1965) ‘Patterns of Trade and Investment in a Dynamic Model of International Trade. [16] Shimomura. [15] Oniki.’ in G.. 54–77. E. Shimomura (2006) ‘Indeterminacy in a Dynamic Two-Country Model.. Wang..-L. (1993) ‘Economic Development and Time Preference Schedule: The Case of Japan and East Asian NICs. K.’ Review of Economic Studies 32. Boldrin. 15–38. K. Chen and P. 103–112. Feichtinger. and Economic Policies: Essays in Honor of Murray C. T. (1993) ‘Durable Consumption Goods and the Pattern of International Trade. [14] Ogawa.[10] Kikuchi. Shimomura (2002) ‘Trade and Indeterminacy in a Dynamic General Equilibrium Model. ed. 244–259.’ Review of Development Economics 11. [13] Nishimura. eds. K. Welfare. (1992) ‘A Two-Sector Dynamic General Equilibrium Model of Distribution. V. B.’ in H. K. and H.’ Journal of Economic Theory 105..’ Economic Theory 29. [12] Nishimura. Dynamic Economic Models and Optimal Control. 18 . and K. Kemp. K. [18] Shimomura. (2004) ‘Indeterminacy in a Dynamic General Equilibrium Model of International Trade. 175–195. H. 307–324.’ Journal of Political Economy 99.. and K.C.’ in M. 105-123. 232–241. [11] Lawrence. [17] Shimomura. Herberg and N... North-Holland. Trade.’ Journal of Development Economics 42. eds. Michigan University Press. (1991) ‘Poverty and the Rate of Time Preference: Evidence from Panel Data. K.

Wolfe. J. Value. Chapter 7.’ Quarterly Journal of Economics 112.. H.The Development Process of Rapidly Growing Economies: From Theory to Empirics.’ in J. and Optimal Asset Holdings. J. N. Capital and Growth: Papers in Honour of Sir John Hicks. the Consumption Function. 485–504. E. Cheltenham. UK. Edward Elgar Publishing Inc. ed. (1997) ‘Growth and Interdependence. (1970) ‘Factor Price Equalization in a Dynamic Economy. [21] Ventura. 153– 167. 19 . [20] Uzawa.’ Journal of Political Economy 78. 57–84. [19] Stiglitz. (1968) ‘Time Preferences. 456–488. University of Edinburgh Press: Edinburgh.

As we shall show later.1 Existence Let us consider the existence of the steady state. l = l∗ ). one can obtain 0 = e N 1/(1−σ) ψ − ξ(r k + w l). which implies that as long as the international diﬀerences in those economic fundamentals are not very large. uniqueness and stability of the steady state with incomplete specialization in the present two-country dynamic general equilibrium model. and initial factor endowments are common between Home and Foreign (m = m∗ = 1. uniqueness and stability are guaranteed. technologies. the existence. and N becomes 0 = r(ξp)k + w(ξp)l − e[N 1/(1−σ) p]ψ[u(r(ξp))]. Uniqueness and Stability of the Steady State with Incomplete Specialization in Both Countries Here. it is clear from (30)-(33) that k = k ∗ holds at the steady state. the system of equations which describes the steady-state k. 20 . Since we assume l = l∗ .6 Appendix: Existence. k= eψ − wl r (50) holds. we shall prove the existence. the determinant of the Jacobian at a symmetrical steady state is not zero. ¯ From (47). Therefore. We shall focus on the symmetric case where preferences. 6. (47) (48) (49) N y = 2ξ[r (ξp)k + w (ξp)l]. ¯ N y = 2e [N 1/(1−σ) p]N 1/(1−σ) ψ[u(r(ξp))]. p. Combining (48) − (49).

p¯ y Substituting (51) into (52). we obtain the ﬁrst relationship between N and p:14 ˆ N = θw − 1. where θe = [pe N 1/(1−σ) /e]. 0 = θe eψ − [θr (eψ − wl) + θw wl]. one can obtain pN y = 2eψ ¯ or N = pe N 1/(1−σ) e (52) (51) 2θe eψ. respectively. one can obtain N= 2θe Θw(ξp)l. θw < 0. one can obtain 0 = e N 1/(1−σ) ψ − ξ r (eψ − wl) + w l . θr = (ξpr /r). p ˆ (54) Since the diﬀerentiated products are capital-intensive. which is greater than 1. multiplying p to (49). r Multiplying p and rewriting this in terms of elasticity. 21 . Rearranging this. 15 It can be easily shown that the right-hand side of (53) goes to ∞ (0) when p goes to 0 (∞).Substituting (50) into this. we obtain: eψ = Θwl. (54) implies that N is decreasing in p: we can depict (53) as Curve AA in Figure 1. Next. that is.15 14 Note that θe (θr and θw ) is constant when the preference (the production technologies) takes the Cobb-Douglas form. where Θ ≡ [(θr − θw )/(θr − θe )]. p¯ y (53) In terms of proportional change. and θw = (ξpw /w).

. let us turn to the other condition. 16 Let us deﬁne p as the solution of u(r(ξp)) = 0 ⇐⇒ r(ξp) = ρ(0). which implies that the right-hand side of (55) goes to 0 when p goes to p. (56) implies that N is increasing in p: we can depict (55) as Curve BB in Figure 1. 17 It is apparent from (50) and (51) that the steady-state capital stock is positive.16 Based on the foregoing argument. then we can obtain: N 1/(1−σ) p = β Rearranging this. ˆ N 1 = (σ − 1) 1 − p ˆ θe θw − θr θψ θρ . From (51). On the other hand. Since θψ and θρ are positive.Now. limp→p β(Θw(ξp)l/ψ[u(r(ξp))]) = ∞. the right-hand side of (55) goes to ∞ when p goes to ∞. Then. one can conclude as follows. (55) In terms of proportional change.17 LEMMA A1: There uniquely exists a steady state in which production is incompletely specialized. 22 . e[N 1/(1−σ) p] = Let e−1 ≡ β. (56) where θψ ≡ [uψ (u)/ψ(u)] and θρ = [uρ (u)/ρ(u)]. ψ[u(r(ξp))] Θw(ξp)l ψ[u(r(ξp))] . one can obtain N= p β(Θw(ξp)l/ψ[u(r(ξp))]) σ−1 Θw(ξp)l . we obtain the second relationship between N and p.

p. In the case where the diﬀerentiated products are more capital-intensive than the homogeneous good.N A B N B O p p A p Figure 1 6. ξp) = r(ξp)k + w(ξp)l. k > k1 (ξp). r(ξp)]}. 1 pf (k. 139). what remains to be argued concerning uniqueness is to exclude a steady state where at least one country is completely specialized. ¯ 0 < k < k2 (ξp). y ). f 2 (k. l. and f 1 (k. F (k. we can express the steady-state Home 18 For the derivation of the monopolistically competitive industry’s implicit production funcsee Helpman and Krugman (1985. 23 . l).18 ¯ ¯ ¯ r w Making use of the above GDP function. k2 (ξp) < k < k1 (ξp). l. y ) ≡ y n(k. l. let us consider the whole GDP function. tion. For this purpose. where ki (ξp) ≡ l{ci [w(ξp). r(ξp)]/ci [w(ξp). f 1.2 The Non-Existence of the Steady State with Complete Specialization in Home and/or Foreign Now. y ). it can be expressed as follows.

then both F (k. Let us consider the Jacobian matrix of the steady state. Thus. LEMMA A2: When the two countries are suﬃciently close in terms of factor endowment ratio.and Foreign budget constraints as 0 = 0 = F (k. ξp))]. no country can specialize in producing only one good in the steady state. we can conclude as follows. ξp) hold from properties of the GDP function and vice versa. F (k ∗ . ξp) > F (k ∗ . ξp) − e[N 1/(1−σ) p]ψ[u(Fk (k ∗ . ξp) − e[N 1/(1−σ) p]ψ[u(Fk (k. 6. ξp) and Fk (k. Therefore. ξp))]. ξp) ≤ Fk (k ∗ . 0 0 Zp¯ψ y Z yN ψ ¯ − 2ψ 2(σ−1)ψ Z yN ψ ¯ Zp¯ψ y − 2ψ 2(σ−1)ψ y ¯ −2ξ 2 (r k + w l) θe − σ−1 + 1 y ¯ 2ψe N 2/(1−σ) p¯ y 2(σ−1) p¯ y 2(σ−1) 0 24 . (57) (58) If k > k ∗ .3 Local Saddlepoint-Stability Let us assume that the two countries are identical. (57) and (58) together imply that there is no steady state such that k = k ∗ holds. 0 0 0 ρ 0 ρ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 −eψ 0 0 0 0 −eψ −ξr −ξr 0 0 0 0 0 0 0 0 0 0 ρ 0 −ρ 0 0 0 0 0 0 0 0 ρ 0 −ρ 0 0 −eψ 0 Zρ 0 −Zeψ 0 −Zeψ −δρ 0 −eψ 0 Zρ 0 −Zeψ 0 −Zeψ −δρ 0 0 yN ψ ¯ 2ψ 0 yN ψ ¯ 2ψ 0 0 −Zξr 0 −Zξr 0 0 .

First. and the corresponding eigenvalue as x. from Ξ(x) = 0. 0 O4 Let us make the following calculations to obtain the above determinant. the 7th row minus the third row multiplied by eψ /x. Finally. where I6 0 I≡ . When the ﬁrst column is subtracted from the second column. we see that Ξ(x) x(ρ−x) 0 Ξ(x) x(ρ−x) pZ y ψ ¯ 2(σ−1)ψ pZ y ψ ¯ 2(σ−1)ψ ¯N − Z y2ψψ + ¯N − Z y2ψψ + eψ Zξr x eψ Zξr x Ω(x) = (ρ − x) x 4 2 0 − ξr eψ ρ−x yN ψ ¯ 2ψ − ξr eψ ρ−x yN ψ ¯ 2ψ y+ ¯ − pξr y ¯ (ρ−x)(σ−1) θe σ−1 −2ξ 2 (r k + w l) 2ψe N 2/(1−σ) +1 y ¯ = (ρ − x) Ξ(x) × 0 −ξr eψ yN ψ ¯ 2ψ 0 Ξ(x) −ξr eψ yN ψ ¯ 2ψ (ρ−x)xpZ y ψ ¯ 2(σ−1)ψ (ρ−x)xpZ y ψ ¯ 2(σ−1)ψ ¯N (ρ − x)x − Z y2ψψ + ¯N (ρ − x)x − Z y2ψψ + eψ Zξr x eψ Zξr x (ρ − x)¯ + y − θe σ−1 pξr y ¯ σ−1 . Next. Then x is determined by the characteristic equation Ω(x) = |J − xI| = 0. Then. Furthermore. we add the 5th row multiplied by ρ /(ρ − x) to the 7th row. we obtain Ω(x) = (ρ − x)Ξ(x) 1 × −1 0 0 0 Ξ(x) −ξr eψ yN ψ ¯ 2ψ (ρ−x)xpZ y ψ ¯ 2(σ−1)ψ (ρ−x)xpZ y ψ ¯ 2(σ−1)ψ ¯N (ρ − x)x − Z y2ψψ + ¯N (ρ − x)x − Z y2ψψ + eψ Zξr x eψ Zξr x (ρ − x)¯ + y − θe σ−1 pξr y ¯ σ−1 . we have 25 . let us add both the ﬁrst row multiplied by ξr /(ρ − x) and the second row multiplied by ξr /(ρ − x) to the 9th row. we have x1 = ρ > 0.Denote the above matrix by J. −2(ρ − x)ξ 2 (r k + w l) 2ψe N 2/(1−σ) +1 y ¯ where Ξ(x) ≡ (Zeψ + δρ )x2 − ρ(Zeψ + δρ )x − Zeψ ρ ρ. and the 8th row minus the 4th row multiplied by eψ /x. −2(ρ − x)ξ 2 (r k + w l) 2ψe N 2/(1−σ) +1 y ¯ For Ω(x) = 0. and add the 6th row multiplied by ρ /(ρ − x) to the 8th row.

the term on x3 becomes positive if σ is suﬃciently large. r k + w l. Therefore. Next. 26 . since the term. which we assume. 2ψ ψ σ−1 −(Zeψ + δρ )2ψe N 2/(1−σ) − where only the second term is negative. if 19 Indeed. when the 1st row is added to the 2nd row. y According to A(x). it can be shown that the sum of the second term and the third one becomes positive if σ > [(θr r k + θw w l)/(r k + w l)]. since all of the ﬁrst and second derivatives of ψ and ρ are positive. is positive due to the convexity of the GDP function with respect to p. we obtain Ω(x) = (ρ − x)Ξ(x) 1 × 0 0 Ξ(x) (ρ−x)xpZ y ψ ¯ 2(σ−1)ψ (ρ−x)xpZ y ψ ¯ (σ−1)ψ ¯N (ρ − x)x − Z y2ψψ + ¯N (ρ − x)x − Z yψ ψ + eψ Zξr x 2eψ Zξr x 0 −ξr eψ 0 yN ψ ¯ 2ψ (ρ − x)¯ + y − θe σ−1 pξr y ¯ σ−1 −2(ρ − x)ξ 2 (r k + w l) 2ψe N 2/(1−σ) +1 y ¯ = (ρ − x)Ξ(x)¯ y Ξ(x) × −ξr eψ yN ψ ¯ 2ψ (ρ−x)xpZψ (σ−1)ψ ¯ (ρ − x) − xZ yN ψ + 2eψ Zξr ψ ρ−x+ − θe σ−1 pξr σ−1 −2(ρ − x)ξ 2 (r k + w l) 2ψe N 2/(1−σ) +1 ≡ (ρ − x)Ξ(x)¯A(x). the term on x3 becomes pZψ yN ψ ¯ · 2ξ 2 (r k + w l) 2ψ (σ − 1)ψ y N ψ y N Zψ ¯ ¯ θe + · + (Zeψ + δρ )2ξ 2 (r k + w l) +1 .x2 > 0 > x3 such that Ω(x2 ) = Ω(x3 ) = 0.19 Then.

since the sum of the second term and the third one in the square brackets becomes ξr eψ ρ pξr θe pe N 1/(1−σ) +1 − p r σ−1 e ξr eψ ρ = (θr − θe ) p > 0. the steady state with both countries being incompletely specialized is locally saddlepoint-stable. A(x) = 0 has one negative root x4 . A(0) = 2eψ ρ −Zρ × −ξr eψ yN ψ ¯ 2ψ 0 ρ+ − pξr σ−1 Zξr −ρξ 2 (r k + w l) ψe N 2/(1−σ) + + (ξr ) eψ 2 2 θe σ−1 θe σ−1 +1 = 2eψ ρZ θr 2/(1−σ) −ρ ρ 1 + σ−1 ψe N × θr −ξr ρ 1 + σ−1 e N 1/(1−σ) ψ +1 θe σ−1 . k and k ∗ . x3 and x4 . Since there are two state variables. there are two negative characteristic roots. LEMMA A3: When the two countries are suﬃciently close.. it follows that the steady state is a saddle point. 1+ θr σ−1 27 . i.A(0) > 0 holds. +1 ρ ρξ (r k + w l) which is positive.e. Therefore.