9 Growth and Trade With Capital and Knowledge

9 Growth and Trade with Capital and Knowledge
Adam Smith held that the large gains in the productivity of labor have their
origins in the large part of the skill, dexterity, and judgment, which are
consequences of the division of labor. The opportunities and incentives to
which workers respond in their investment in human capital are not seri-
ously considered by classical economists such as Smith, Ricardo, and
Marx. The omission was perhaps not so much misleading as it might ap-
pear, given the role of innovation and education in economic development
when the classical economists were constructing their theories. Neverthe-
less, to understand contemporary world economies, it is essential to exam-
ine possible effects of trade upon personal income distribution in a global-
izing world economy. It has been argued that productivity differences
explain much of the variation in incomes across countries, and technology
plays a key role in determining productivity.1 The pattern of worldwide
technical change is determined largely by international technology diffu-
sion because a few rich countries account for most of the world’s creation
of new technology. As globalization is deepening, it is important to pro-
vide analytical frameworks for analyzing global economic interactions. For
instance, it is important to examine how a developing economy like India
or China may affect different economies as its technology is improved or
population is increased; or how trade patterns may be affected as technolo-
gies are further improved or propensities to save are reduced in developed
economies like the US or Japan.
One of the first seminal attempts to render technical progress endoge-
nous in growth models was initiated by Arrow in 1962. He emphasized
one aspect of knowledge accumulation - learning by doing. In 1965 Uzawa
introduced a sector specifying in creating knowledge into growth theory.
The knowledge sector utilizes labor and the existing stock of knowledge to
produce new knowledge, which enhances productivity of the production
sector. Another approach is taken by, for instance, Kennedy in 1964,
Weizsäcker in 1966 and Samuelson in 1965, who took account of the as-
1 Different channels of inequalities are modeled by, for instance, Krugman and
Venables (1995), Manasse and Turrini (2001), Nakajima (2003), and Agénor
(2004).
324 9 Growth and Trade with Capital and Knowledge
sumption of “inducement through the factor prices”. In 1981 Schultz em-

phasized the incentive effects of policy on investment in human capital.2
There are many other studies on endogenous technical progresses.3 But on
the whole theoretical economists had been relatively silent on the topic
from the end of the 70s until the publication of Romer’s paper in 1986
(Romer, 1986). Since then there has been an even increasing number of
publications on the literature. In Romer’s approach, knowledge is taken as
an input in the production function and competitive equilibrium is ren-
dered consistent with increasing aggregate returns owing to externalities. It
is assumed that knowledge displays increasing marginal productivity but
new knowledge is produced by investment in research technology, which
exhibits diminishing returns. Various other issues related to innovation,
diffusion of technology and behavior of economic agents under various in-
stitutions have been discussed in the literature. There are also many other
models emphasizing different aspects, such as education, trade, R&D poli-
cies, entrepreneurship, division of labor, learning through trading, brain
drain, economic geography, of dynamic interactions among economic
structure, development and knowledge.4 These studies attempted to for-
malize trade patterns with endogenous technological change and monopo-
listic competition. They often link trade theory with increasing-returns
growth theory. Within such frameworks the dynamic interdependence be-
tween trade patterns, R&D efforts, and various economic policies are con-
nected. With the development of models with endogenous long-run
growth, economists now have formal techniques with which they explore
the relationship between trade policy and long-run growth either with
knowledge or with capital, but in most of them not with both capital and
knowledge within the same framework. One of problems with the new
growth theory is that its analytical framework is not effective in dealing
with capital accumulation and innovation in a single consistent framework.
It has been observed that world R&D activity and world production of
capital equipment are highly concentrated in a small number of countries.
As shown by Eaton and Kortum (2001), the countries that are most R&D
intensive are also the ones most specialized in making equipment. Al-
though each country may not spend much on R&D, the benefits may
2 See Arrow (1962), Uzawa (1965), Kennedy (1964), Weisäcker (1966),

Samuelson (1965), and Schultz (1981).
3 Works, for instance, by Sato and Tsutsui (1984) and Nelson and Winter (1982),
are examples along this line.

4 See, for example, Dollar (1986), Chari and Hopenhayn (1991), Krugman
(1991), Rauch (1991a), Stokey (1991), Nardini (2001), Martin and Ottaviano (2001),
Brecher et al. (2002), and Nocco (2005).
9.1 A National Growth Model 325
spread around the world through imitation and exports of capital goods
that embody new technology. It is argued that a country’s productivity de-
pends on its access to capital goods from around the world and its willing-
ness and ability to utilize them. Eaton and Kortum develop a model of
trade in capital goods to take account of this view. The theoretical frame-
work is a combination of the neoclassical growth model of technological
change embodied in new capital goods and a model of Ricardian trade. A
main purpose of the model is to link productivity to imports of capital
goods. The model also tries to discuss impact of trade barriers measured in
costs arising from factors such as marketing overseas, negotiating a foreign
purchase, transporting goods to foreign location, tariffs, non-tariff barriers,
distributing goods in foreign markets, installation in foreign production fa-
cilities, training foreign workers to use the equipment, and providing parts,
maintenance, and customer service from abroad.
This chapter examines interactions between growth, trade, knowledge
utilization, and creativity within a compact analytical framework. We con-
sider knowledge as an international public good in the sense that all coun-
tries access knowledge and the utilization of knowledge by one country
does not affect that by others. Section 9.1 introduces a growth model with
endogenous human capital accumulation for a national economy. Section 9.2
proposes a multi-country model with capital accumulation and knowledge
creation. This section assumes that knowledge creation is through learning by
doing and research. This section simulates the model to see how the system
moves over time and how the motion of the system is affected when some
parameters are changed. This section is organized as follows. Section 9.2.1
defines the multi-country model with capital accumulation and knowledge
creation. Section 9.2.2 examines the case when all the countries have the
same preference. We show that the motion of the global economy can be ex-
pressed by a two-dimensional differential equations system and we can ex-
plicitly determine the dynamic properties of the global economy. Section
9.2.3 shows that the dynamics of the world economy with J countries can be
described by ( J + 1) -dimensional differential equations. As mathematical
analysis of the system is too complicated, we demonstrate some of the dy-
namic properties by simulation when the world economy consists of three
countries. Sections 9.2.4 – 9.2.7 examine respectively effects of changes in
each country’s knowledge utilization efficiency and creativity, research pol-
icy, the propensity to save, and the population. The analytical results in Sect.
9.2.3 are proved in Appendix A.9.1.
9.1 A National Growth Model
This section introduces a growth model with endogenous human capital

accumulation for a national economy.5 As the international trade model
with capital and knowledge in the next section is complicated, we are first
concerned with a national economy to introduce the basic concepts for
economic dynamics with capital and knowledge. The model considers Ar-
row’s learning by doing and Uzawa’s education as two main sources of
human capital accumulation. Another important issue we will address in
this section is path-dependent economic development. An economy’s long-
term prosperity may depend on initial conditions is nowadays a familiar
idea in the growth literature. Many models capture different aspects of this
kind of phenomena in formal models.6 This section shows that when dif-
ferent sources of learning exhibit increasing and decreasing returns to
scale, then the system has multiple equilibrium points and its evolution is
characterized of being path-dependent.
9.1.1 The OSG Model with Endogenous Human Capital
The economy has one production sector and one education sector. The lat-
ter is called the university. We assume a homogenous and fixed national
labor force, N . The labor force is distributed between economic activities,
teaching and studying. We select commodity to serve as numeraire, with
all the other prices being measured relative to its price. We assume that
wage rates are identical among all professions. We introduce
F (t )  output level of the production sector at time t;

K (t )  level of capital stocks of the economy;
H (t )  level of human capital of the population;
N i (t ) and K i (t )  labor force and capital stocks employed by the produc-
tion sector;
5 The model is proposed by Zhang (2005a). See also Zhang (2006a: Sect. 3.2).
It should be noted that in Zhang (2007d) the time for education is an endogenous
variable.
6 There are a large number of the literature on economic growth with bifurcations
and chaos (for instance, Day, 1984; Hommes, 1991, 1998; Zhang, 1990, 1991; Azari-
adis, 1993; Boldrin, et al. 2001; Matsuyama, 1991, 2001; Shone, 2002). Zhang
(2005b, 2006b) introduces contemporary theories of differential and difference equa-
tions and their applications to economics.
N v (t ) and N e  number of teachers and number of students;

K e (t )  capital stocks employed by the university;
w(t ) and r (t )  wage rate and rate of interest.
We first model production and consumption. We assume that production

is to combine qualified labor force, H m (t )N i (t ), and physical capital,
K i (t ). Most aspects of our model are the same as the OSG model. The
production process is described by
F (t ) = AK iα (t )(H m (t )N i (t )) , A, α , β > 0 , α + β = 1.
β
The marginal conditions are given by

r (t ) = (1 − τ )αAki− β (t ), w(t ) = (1 − τ )βAH m (t )kiα (t ), (9.1.1)
where ki ≡ K i / H m N i and τ is the tax rate on the product level.

We denote per capita wealth by k (t ), where k (t ) ≡ K (t ) / N . Per capita
current income is given by
y (t ) = r (t )k (t ) + w(t ).
The per capita disposable income is given by
yˆ (t ) = y (t ) + k (t ).
The budget constraint is given by:
c(t ) + s (t ) = yˆ (t ) .
The utility function is given
U (t ) = c ξ (t )s λ (t ), ξ , λ > 0 , ξ + λ = 1.
The optimal solution is given by
c(t ) = ξyˆ (t ), s (t ) = λyˆ (t ). (9.1.2)
The change in the household’s wealth is equal to the savings minus the
wealth sold at time t , i.e.
k&(t ) = s ( yˆ (t ) ) − k (t ) = λyˆ (t ) − k (t ). (9.1.3)
We now study the behavior of the university. We assume that there are
two sources of improving human capital, through education and learning
by producing. Arrow first introduced learning by doing into growth theory;
Uzawa took account of trade offs between investment in education and

capital accumulation.7 We propose that human capital dynamics is given
by
υ K α e (H m N v ) v (H m N e ) e (9.1.4)
β β
υF
H& = e e + i π − δhH ,
N NH
where δ h (> 0) is the depreciation rate of human capital, υ e , υi , α e , β v ,
and β e are non-negative parameters. The above equation is a synthesis
and generalization of Arrow’s and Uzawa’s ideas about human capital ac-
cumulation. The term, υ e K eα e (H m N v ) v (H m N e ) e , describes the contribu-
β β
tion to human capital improvement through education. Human capital

tends to increase with an increase in the number of students. The term di-
vided by N measures the contribution per capita. We take account of
learning by doing effects in human capital accumulation by the term
υi F / H π . This term implies that contribution of the production sector to
human capital improvement is positively related to its production scale,
F , and is dependent on the level of human capital. The term, H π i , takes
account of returns to scale effects in human capital accumulation. The case
of π > (<) 0 implies that as human capital is increased it is more difficult
(easier) to further improve the level of human capital.
We assume that the students and teachers are paid by government’s tax
income at the same wage rate as the wage rate of workers. We assume that
the economy has a fixed ratio of the population who is getting education in
the university. The number of students is given by N e = ne N . Assume that
the total tax income is used for paying the students, teachers and the capi-
tal stocks employed by the university. The government spends w(t )N e
amount of money on students. Obviously, this assumption is strict. A way
to relax this assumption without increasing analytical complex of the
model is to assume that each individual spends a given amount of time on
education. The time distribution is given by Ti + Te = T , where T is the
7 Arrow (1962) and Uzawa (1965). Learning by doing has been introduced into
growth theory in different ways. For instance, Chari and Hopenhayn (1991), Par-
ente (1994), and Stokey (1988) study learning by doing as a force for sustained
growth; Brezis et al. (1993), Krusell and Rios-Rull (1996), and Jovanovic and
Nyarko (1996) show that learning by doing can give rise to the overtaking; Karp
and Lee (2001) examined learning by doing and the choice of technology; and
Liso et al. (2001) examine the implications of learning by doing for division of la-
bor.
fixed available time, Ti is the work time and Te is the time as a student.
The student gets free education but does not receive any wage. It can be
seen that the conclusions will not be different in the two cases and the
model is still analytically tractable. The budget for paying teachers and the
capital stocks of the university is given by
w(t )N v (t ) + r (t )K e (t ) = τF (t ) − w(t )N e . (9.1.5)
The university distributes its total resource, τF (t ) − w(t )N e , to the teach-

ers, N v (t ), and the capital stocks, K e (t ), in such a way that the output of
the university, υ e K eα e (H m N v ) v (H m N e ) e , will be maximized. The univer-
β β
sity’s problem is formed as follows
Max υ e K eα e (H m N v ) v (H m N e ) e ,
β β
subject to Eq. (9.1.5). The solution is given by

α e (τF − wN e ) β v (τF − wN e ) (9.1.6)
Ke = , Nv = .
(α e + β v )r (α e + β v )w
Labor force and capital stocks are fully employed
Ni + Ne + Nv = N , Ki + Ke = K. (9.1.7)
We have thus built the model. We now examine properties of the dy-
namic system.
9.1.2 The Dynamics and Multiple Equilibrium Points
From Eq. (9.1.7), we get N i + N v = N − N e . From Eqs. (9.1.1) and

(9.1.6), we obtain
α e ki N i H m (τ − (1 − τ )βN e / N i )
Ke = ,
ατ 0
β v (τN i − (1 − τ )βN e ) (9.1.8)

Nv = ,
βτ 0
where we use
F = N i H m f , τ 0 ≡ (α e + β v )(1 − τ ).
From Eqs. (9.1.8) and N i + N v = N − N e , we solve

N − N e + β v N e / (α e + β v )
Ni = τ 0β ,
τ 0 β + τβ v
β v (τN i − (1 − τ )βN e ) (9.1.9)

Nv = .
τ 0β
Hence, given the government policy on education (measured by τ and
N e ), and the university’s production character (measured by α e and β v )
we uniquely determine the labor distribution, which is invariant in time. If
τ , N e , α e , and/or β v are shifted, then the labor distribution will be
changed. The invariance of the labor distribution for the given policy and
university production character is due to the assumed Cobb-Douglas func-
tional forms. As shown in the appendix, the invariance is not generally
held.
According to the definition of ki (t ), from Eqs. (9.1.8), we have
α e (τ − (1 − τ )βN e / N i )
K e = aK i , a ≡ .
ατ 0
From the above equations and K i + K e = Nk , we solve
Nk (t ) aNk (t ) (9.1.10)
K i (t ) = , K e (t ) = .
1+ a 1+ a
As ki = K i / H m N i , we obtain
k (t ) (9.1.11)
ki (t ) = ni ,
H m (t )
where ni ≡ N / N i (1 + a ). We see that ki (t ) is a function of k (t ) and H (t ).

Summarizing the above discussions, we have the following lemma.
Lemma 9.1.1
For any given levels of wealth and human capital, k (t ) and H (t ), all the
other variables in the system are uniquely determined at any point of time.
The values of the variables are given as functions of k (t ) and H (t ) by the
following procedure: N i and N v by Eqs. (9.1.9) → K i (t ) and K e (t ) by
Eqs. (9.1.10) → ki (t ) by Eq. (9.1.11) → f = Akiα and F = N i H m f →
r (t ) and w(t ) → y = rk + w and yˆ = y + k → c and s by Eqs. (9.1.2)

→ U = cξ s λ .
We now show how to determine by k (t ) and H (t ) at any point of time.

From Eqs. (9.1.3) and (9.1.4), and Lemma 9.1.1, it is not difficult to see
that the dynamics of k (t ) and H (t ) are explicitly given by the following
two differential equations
k&(t ) = λ* k α H βm − ξk ,
H& = υ e*k α e H β v m + β e m + υi* H m −π −αm k α − δ h H , (9.1.12)
where
λ* ≡ λA(1 − τ )(αni− β + βniα ),
αe
 aN  υ e N vβ v N eβ e υ Anα N
υ ≡ 
*
e
 , υi* ≡ i i i .
1 + a  N N
Lemma 9.1.1 guarantees that if we know values of k (t ) and H (t ), then

we can solve all the other variables as functions of k (t ) and H (t ). Hence,
to examine the dynamic properties of the system, it is sufficient to examine
dynamic properties of Eqs. (9.1.12).
A steady state of the system is given by
λ* k α H βm = ξk ,
υ e*k α e H β v m+ β e m + υi* H m−π −αm k α = δ h H . (9.1.13)
From the first equation in Eqs. (9.1.13), we solve k = (λ* / ξ ) H m .

1/ β
Substituting this equation into the second equation in Eqs. (9.1.13) yields
Φ( H ) ≡ Φ e ( H ) + Φ i ( H ) − δ h = 0 , (9.1.14)
where
αe / β α /β
 λ*   λ* 
Φ e (H ) ≡   υ H , Φ i (H ) ≡  
* xe
υ i* H xi ,
ξ  ξ 
e
xe ≡ (α e + β v + β e )m − 1, xi ≡ m − π − 1.
We see that the number of economic equilibrium points is equal to the

number of solutions of the equation, Φ(H ) = 0 for 0 < H < ∞ . As shown
in Fig. 9.1.1, the equation may have none, one equilibrium point, or two
equilibrium points. As shown in Fig. 9.1.1a, if xe < 0 and xi < 0 , the
equation monotonically decreases in H and it passes the horizontal axis
only once. Figure 9.1.1b depicts the case of xe > 0 and xi > 0 , the func-
tion monotonically increases in H and it passes the horizontal axis only
once. Figure 9.1.1c depicts the case of xe > 0 and xi < 0 ( xe < 0 and
xi > 0 ).
a) both exhibit decreasing returns b) both exhibit increasing returns
Fig. 9.1.1. The two sectors exhibit different returns to scale effects
We can prove the conditions in Fig. 9.1.1. The following proposition

shows that the properties of the dynamic system are determined by the two
returns to scale parameters, xe and xi .
Proposition 9.1.18
(1) If xe < 0 and xi < 0, the system has a unique stable equilibrium; (2) If
xe > 0 and xi > 0 , the system has a unique unstable equilibrium; and (3)
If xe > 0 and xi < 0 ( xe < 0 and xi > 0 ), the system has either no equi-
librium, one equilibrium or two equilibrium points. When the system has
two equilibrium points, the equilibrium with low (high) level of H is sta-
ble (unstable).
We only interpret the stability condition, xe < 0 and xi < 0. From the
definitions of xe and xi , we may interpret xe and xi respectively as
measurements of returns to scale of the university and the industrial sector
in the dynamic system. When xe (< (>) 0, we say that the university dis-
plays decreasing (increasing) returns to scale in the dynamic economy. We
conclude that if both the university and the production sector display de-
creasing returns, then the dynamic system has a unique stable equilibrium.
If the two sectors exhibit decreasing returns to scale, the system will ap-
proach to its equilibrium in the long term. In a traditional society like the
one constructed by Adam Smith where increases in human capital mainly
come from division of labor and traditional education, the economic sys-
tem tends to be dominated by stability. In a newly industrializing econ-
omy, education may exhibit increasing returns to scale and learning by do-
ing may not be effective in improving human capital. The economy may
have multiple equilibrium points. If the society fails to explore increasing
returns effects from education, it may not achieve rapid industrialization.
We now demonstrate dynamics of the nonlinear system with multiple equi-
librium points.
9.1.3 The Path-Dependent Motion of the System by Simulation
We stimulate the model with two equilibrium points. We specify the pa-
rameters as follows
α = 0.35, N = 1, A = 2, N e = 0.06, τ = 0.08, λ = 0.7,
α e = 0.7, β e = 0.7, β v = 0.7, ν e = 1.8, ν i = 0.02,
(9.1.15)
δ h = 0.08, π = 0.3, m = 0.8.
8 The proof of the proposition is referred to Zhang (2005a).

Under the above specifications, we have xe = 0.4 and xi = − 0.5. The

university exhibits increasing returns to scale and the production sector is
characterized of decreasing returns. The system has two equilibrium points
(k1 , H1 ) = (17.858, 2.226), (k 2 , H 2 ) = (10.401, 19.509).
The one with lower levels of human capital and per capita wealth is sta-
ble; the other is unstable, also as analytically proved. Figure 9.1.2 depicts
the vector field and the steady states of the dynamic system. As shown in
Fig. 9.1.2, an economy with low level of human capital, even if it was ini-
tially rich, tends to converge to the stable equilibrium with low standard of
living and low level human capital. An economy with high level of human
capital, even if it was initially poor, tends to experience sustained growth.
This nonlinear dynamic system has path-dependent features. Here, we can
see the significance of cultural value for education. Japanese, Korean, and
Chinese-dominated economies like Singapore and Taiwan could have sus-
tained economic growth irrespective of their initial poor conditions in the
1960s, mainly because of their cultural values on education, rather than
due to high saving rates.9 In the 1950s, no one could have foreseen rapid
economic development of East Asia, because few economists recognized
the significance of education in economic development and fewer knew
the validity of rationalism in classical Confucianism for modern econo-
mies.10 In the literature of economic growth and development published in
the 1960s and 1970s, capital accumulation is the main engine of economic
development. Economists failed to properly interpret economic evolution
of these regions because they did not properly examine the cultural values
of education in this region.
We simulate three paths of the economy with the initial values
(k0 , H 0 ) = (1.5, 10), (k0 , H 0 ) = (115, 12), (k0 , H 0 ) = (70, 23).
The paths with (k 0 , H 0 ) = (1.5, 10) and (k 0 , H 0 ) = (115, 12 ) converge to
the low levels of human capital and per capita wealth. It is interesting to
note that the path with (k 0 , H 0 ) = (115, 12 ) starts with high level of wealth.
But its level of human capital is not improved over time. As decreasing re-
turns dominate this path, its prosperity does not last long. The path with
(k0 , H 0 ) = (70, 23) will grow infinitely because the increasing return to
9 The modern economic developments of Confucian regions, Japan, Korea,
Singapore, Hong Kong, and Mainland China are systematically examined by
Zhang (1998a, 2002a, 2003b, 2006c, 2007b).
10 Among well-known East Asian thinkers, perhaps only Fukuzawa Yukichi
(1835-1901) is exceptional.
scale dominates the economic evolution. Indeed, this kind of infinite

growth will not happen in reality as our model neglects many other signifi-
cant factors such as endogenous population change, negative externalities
such as pollution, and limitations of natural resources, which are neglected
in this model.
H
25
20
15
10
20 40 60 80 100 120
Fig. 9.1.2. Path dependent economic evolution
As far as qualitative features of economic development are concerned,

Fig. 9.1.2 provides some insights into difference in the economic devel-
opment in Mainland China and Taiwan during the period of 1950 to 1980.
The two areas started the economic development with similar economic
conditions but different average educational levels. Before the economic
reform in 1978 started in Mainland China, the living standards and educa-
tional achievements in the two Chinese societies had been enlarging. It is
only in recent years that Mainland China has begun to explore the oppor-
tunities of economic development. Structurally speaking, Mainland
China’s political economic system had devaluated modern (Western) edu-
cation so that no sector in the society could have explored potential bene-
fits of increasing returns offered by the Western civilization. Both cultural
values and political systems matter in our “neoclassical” model – I call this
model as neoclassical in the sense that except the utility function, all the
assumptions accepted in this study were developed and accepted in the lit-
erature of neoclassical economic growth theory developed in the 1950s

and 1960s.
9.1.4 Comparative Dynamic Analysis11
We first examine effects of tax on the dynamic system. In our system, the
tax income is totally spent on education. We may thus interpret increases
in the tax rate as the promotion policy taken by the government. We may
simulate the model again. Here, we are interested in the path-dependent
case. We still specify the parameter values as in (9.1.15) except the tax
rate, τ . We reduce the resource for education. Let us consider the case that
the expenditure on education is reduced from 8 percent of the GDP to 7
percent, that is
τ : 0.08 ⇒ 0.07.
Figure 9.1.3 shows the simulation results – the points with larger sizes is
the new steady states and the other two points with smaller sizes are the
old steady states. The two steady states shift as follows:
(k1 , H1 ) : (17.858, 2.226) ⇒ (14.758, 1.718),
(k 2 , H 2 ) : (101.401, 19.509) ⇒ (230.783, 53.417 ).
We see that the new stable steady state has lower levels of human capi-
tal and per capita wealth; but the new unstable steady states have much
higher levels of k and H . It seems promising with the new education pol-
icy because the new unstable steady state of higher k and H is much bet-
ter than the old unstable one. Nevertheless, the economy with the discour-
aging policy has more chances to the traditional trap than to the economic
miracle. For instance, if we start from the following three points as in the
previous example in Fig. 9.1.2:
(k0 , H 0 ) = (1.5, 10), (k0 , H 0 ) = (115, 12), (k0 , H 0 ) = (70, 23).
As demonstrated in Fig. 9.1.3, all the paths with these initial condi-
tions end up in the poverty trap. But in Fig. 9.1.2, the path with
(k 0 , H 0 ) = (70, 23) exhibits the economic miracle. This example shows
that the discouraging policy deprives the society from development oppor-
tunity. The “chance” for development is loss due to the new policy. As
shown in Fig. 9.1.3, for the economy to experience sustained growth, the
11 Here, we are only concerned with simulation results. We refer to Zhang
(2005a) for the analytical results on comparative statics analysis.

economy must have a much higher initial level of human capital than in
the case of τ = 0.08 . Hence, if the society reduces its investment in educa-
tion, it will have much less opportunities to experience sustained economic
growth, even though heavy investment in education will not guarantee sus-
tainable development of the nonlinear system in certainty.
H
70
60
50
40
30
20
10
50 100 150 200 250 k
Fig. 9.1.3. The path dependent-development as the education is discouraged
We now examine effects of the propensity to save. We still specify the

parameters values as in (9.1.15), except the propensity to save λ. We in-
crease the propensity to save. Let us consider the case that the propensity
to save is increased from 0.7 to 0.73 , that is, λ : 0.7 ⇒ 0.73 . Figure
9.1.4 shows the simulation results – the points with larger sizes is the new
steady states and the other two points with smaller sizes are the old steady
states. The two steady states shift as follows:
(k1 , H1 ) : (17.858, 2.226) ⇒ (14.758, 1.718),
(k 2 , H 2 ) : (101.401, 19.509) ⇒ (230.783, 53.417 ).
We see that the new stable steady state has higher levels of human
capital and per capita wealth; but the new unstable steady state has lower
levels of k and H . Figure 9.1.4 depicts the impact of change in λ on the
dynamics of the system.
H
25
20
15
10
k
20 40 60 80 100 120
Fig. 9.1.4. An increase in the propensity to save
9.2 Trade and Growth with Learning-by-Doing and

Research
It is well known that dynamic-optimization models with capital accumula-

tion are associated with analytical difficulties. To avoid these difficulties,
this study applies an alternative approach to consumer behavior. It will be
demonstrated that the multi-country trade model with capital accumulation
and knowledge creation becomes analytically tractable with the new ap-
proach to consumer behavior. The model in this section is a further devel-
opment of the two models by Zhang. Zhang (1992) proposed a multi-
country model with capital accumulation and knowledge creation. The
study used the traditional approach to household behavior as in the Solow
one-sector growth model, assuming a constant fraction using for saving.
The knowledge creation is only through Arrow’s learning by doing. This
study models the behavior of households in an alternative way and as-
sumes that knowledge creation is through learning by doing and research.
Although Zhang (1993c) introduced research into growth model, the
model was limited to a two-country economy and the study was only con-
cerned with equilibrium. This section synthesizes the main ideas in the
previous two models, though it extends the previous studies in some as-
pects. This study models behavior of consumers different from the previ-
9.2 Trade and Growth with Learning-by-Doing and Research 339
ous studies. Moreover, the previous studies were only concerned with ex-
amining equilibrium and comparative statics analysis. As no simulation
was provided in the previous studies, it is almost impossible to see how the
multi-country system moves over time. This section simulates the model to
see how the system moves over time and how the motion of the system is
affected when some parameters are changed. This section is organized as
follows. Section 9.2.1 defines the multi-country model with capital accu-
mulation and knowledge creation. Section 9.2.2 examines the case when
all the countries have the same preference. We show that the motion of the
global economy can be expressed by a two-dimensional differential equa-
tions system and we can explicitly determine the dynamic properties of the
global economy. Section 9.2.3 shows that the dynamics of the world econ-
omy with J countries can be described by ( J + 1) -dimensional differential
equations. As mathematical analysis of the system is too complicated, we
demonstrate some of the dynamic properties by simulation when the world
economy consists of three countries. Sections 9.2.4 – 9.2.7 examine re-
spectively effects of changes in each country’s knowledge utilization effi-
ciency and creativity, research policy, the propensity to save, and the
population. The analytical results in Sect. 9.2.3 are proved in Appendix
A.9.1.
9.2.1 The Multi-Country Trade Model with Capital and

Knowledge
Each country has one production sector and one university. The university
is financially supported by the government through taxing the production
sector. Knowledge growth is through learning by doing by the production
sector and R&D activities by the university. In describing the production
sector, we follow the neoclassical trade framework. It is assumed that the
countries produce a homogenous commodity.12 Most aspects of production
sectors in our model are similar to the neo-classical one-sector growth
model.13 There is only one (durable) good in the global economy under
consideration. Households own assets of the economy and distribute their
incomes to consume and save. Production sectors or firms use capital and
labor. Exchanges take place in perfectly competitive markets. Production
sectors sell their product to households or to other sectors and households
sell their labor and assets to production sectors. Factor markets work well;
12 This follows the Oniki-Uzawa trade model and its various extensions with one
capital goods.
13 Burmeister and Dobell (1970).
factors are inelastically supplied and the available factors are fully utilized
at every moment. Saving is undertaken only by households, which implies
that all earnings of firms are distributed in the form of payments to factors
of production. We omit the possibility of hoarding of output in the form of
non-productive inventories held by households. All savings volunteered by
households are absorbed by firms. We require savings and investment to
be equal at any point of time. The system consists of multiple countries,
indexed by j = 1, ..., J . Each country has a fixed labor force, N j ,
( j = 1, ..., J ). Let prices be measured in terms of the commodity and the
price of the commodity be unity. We denote wage and interest rates by
w j (t ) and r j (t ) , respectively, in the j th country. In the free trade system,
the interest rate is identical throughout the world economy, i.e.,
r (t ) = rj (t ).
For convenience, we term the people working in the production sector
as workers and the people working in the university as scientists. The
population is classified into workers and scientists. We introduce
K (t )  the capital stocks of the world economy;

F j (t )  the output level of the production sector by country j ;
K j (t ) and K j (t )  the capital stocks employed and the wealth owned by
country j ;
i , r  subscript indexes denoting the production sector and the univer-
sity, respectively;
N qj (t ) and K qj (t )  the labor force and capital stocks employed by sec-
tor q , q = i , r , in country j ;
kˆ j (t ) , c j (t ) and s j (t )  the wealth owned by, the consumption levels of
and the total savings made by per person in country j ; and
w j (t ) — the wage rate in country j .
Behavior of producers
First, we describe behavior of the production sections. We assume that
there are three factors, physical capital, labor, and knowledge at each point
of time t . The production functions are given by
F j (t ) = A j Z (t )K ijα (t )N ijβ (t ),
mj j j
A j > 0 , α j + β j = 1, α j , β j > 0 , j = 1 , L , J ,
in which Z (t ) (> 0) is the world knowledge stock at time t . Here, we call

m j country j ' s knowledge utilization efficiency parameter. If we interpret
mj /β
Z N j as country j ' s human capital or qualified labor force, we see that
the production function is a neoclassical one and homogeneous of degree
one with the inputs. As cultures, political systems and educational and
training systems vary between countries, m j are different.
Markets are competitive; thus labor and capital earn their marginal prod-
ucts, and firms earn zero profits. The rate of interest, r (t ), and wage rates,
w j (t ), are determined by markets. Hence, for any individual firm r (t ) and
w j (t ) are given at each point of time. The production sector chooses the
two variables, K j (t ) and N ij (t ), to maximize its profit. The marginal con-
ditions are given by
r + δ kj = τˆ j A jα j Z j k j
m −β j α
, w j = τˆ j A j β j Z j k j j ,
m
(9.2.1)
where δ kj are the depreciation rate of physical capital in country j and
K ij (t )
k j (t ) ≡ , τˆ j ≡ 1 − τ j ,
N ij (t )
in which τ j is country j' s tax rate on its production sector.
Behavior of consumers
Each worker may get income from wealth ownership and wages. Con-
sumers make decisions on consumption levels of goods as well as on how
much to save. Let kˆ (t ) stand for the per capita wealth in country j . Each
j
consumer of country j obtains income
y j (t ) = r (t )kˆ j (t ) + w j (t ), j = 1, L, J , (9.2.2)
from the interest payment rkˆ j and the wage payment w j . The disposable
income is equal to
yˆ j (t ) = y j (t ) + kˆ j (t ). (9.2.3)
The disposable income is used for saving and consumption.

At each point of time, a consumer distributes the total available budget

between savings, s j (t ), and consumption of goods, c j (t ). The budget con-
straint is given by
c j (t ) + s j (t ) = yˆ j (t ) = rkˆ j (t ) + w j (t ) + kˆ j (t ). (9.2.4)
At each point of time, consumers have two variables to decide. A con-

sumer decides how much to consume and to save. Equation (9.2.4) means
that consumption and savings exhaust the consumers’ disposable personal
income.
We assume that utility levels that the consumers obtain are dependent
on the consumption level of commodity, c j (t ), and the savings, s j (t ). The
utility level of the consumer in country j , U j (t ), is specified as follows
U j (t ) = c j 0 j (t )s j 0 j (t ), ξ 0 j , λ0 j > 0 , (9.2.5)
ξ λ
where ξ 0 j and λ j are respectively household j ’s propensities to consume

and to hold wealth. Here, for simplicity, we specify the utility function
with the Cobb-Douglas from. It would provide more insights if we take
some other forms of utility functions.
Maximizing U j subject to the budget constraints (9.2.4) yields
c j (t ) = ξ j yˆ j (t ), s j (t ) = λ j yˆ j (t ), (9.2.6)
in which
ξ0 j λ0 j
ξj ≡ , λj ≡ .
ξ 0 j + λ0 j ξ 0 j + λ0 j
According to the definitions of s j (t ), the wealth accumulation of the

representative household in country j is given by
& (9.2.7)
kˆ j (t ) = s j (t ) − kˆ j (t ).
Knowledge creation and behavior of the university

Like capital, a refined classification of knowledge and technologies tend
to lead new conceptions and modeling strategies. Some major new knowl-
edge and inventions that had far reaching and prolonged implications, such
as Newton’s mechanics, Einstein’s theory of relativity, steam engine, elec-
tricity, and computer. Small improvements and non-lasting improvements
take place everywhere, serendipitously and intentionally. Innovations may
also happen in a drastic, discontinuous fashion or in a slow, continuous

manner. The introduction of the first steam engine rapidly triggered a se-
quence of innovations. The same is true about electricity and computer.
Bresnahan and Trajtenberg (1995) argued that technologies have a treelike
structure, with a few prime movers located at the top and all other tech-
nologies radiating out from them. They characterize general purpose tech-
nologies by pervasiveness (which means that such a technology can be
used in many downstream sectors), technological dynamism (which means
that it can support continuous innovational efforts and learning), and inno-
vational complementarities (which exist because productivity of R&D in
downstream sectors increases as a consequence of innovation in the gen-
eral purpose technology, and vice versa). This study uses knowledge in a
highly aggregated sense. We assume a conventional production function of
knowledge in which labor, capital, and technology are combined to create
new knowledge in a deterministic way. This is an approximate description
of the idea that devoting more resources to research yields more rapidly
new knowledge. There does not appear to have certain evidence for sup-
porting any form of how increases in the stock of knowledge affect the
creation of new knowledge. We do not require that the creation function
for knowledge have constant returns to scale in capital and labor. It is pos-
sible that doubling the number of computers and scientists increases three
times of the knowledge creation than before – the university’s knowledge
creation exhibits increasing returns to scale in scientist and capital. It is
also possible for the university to have decreasing returns to scale. We thus
should allow three possibilities - increasing, constant, decreasing returns to
scale in scientists and capital – in the university’s knowledge creation.
We consider that research is carried out only by the universities.14 We
propose the following equation for knowledge growth
J  τ F (t )  (9.2.8)
Z& (t ) = ∑  ε ij + τ rj Z rj (t )K rj rj (t )N rj rj (t ) − δ z Z (t ),
ij j ε α β
j =1  Z (t ) 
in which δ z (≥ 0) is the depreciation rate of knowledge, and ε qj , τ qj , α rj
and β rj are parameters. We require τ qj , α rj , and β rj to be non-negative.
14 In some studies it is assumed that the research sector consists of two sub-
sectors: a private research sector and a government research sector, for instance,
Park (1998). In Park’s model, the government may create knowledge useful for
defense, space, and environment and the private sector for industrial, agricultural,
and consumption goods. Some overlapping knowledge, like mathematical and sci-
entific knowledge, may be tailored for research as particular activities.
To interpret Eq. (9.2.8), first let us consider a special case that knowl-
edge accumulation is through learning by doing. The parameters τ ij and
ε ij
δ z are non-negative. We interpret τ ij F / Z as the contribution to knowl-
edge accumulation through learning by doing by country j ' s production
sector. To see how learning by doing occurs, assume that knowledge is a
function of country j ' s total industrial output during some period
 t 
Z (t ) = a1  F j (θ ) dθ  a2 + a3 ,
∫
 0 
in which a1 ,a2 and a3 are positive parameters. The above equation implies
that the knowledge accumulation through learning by doing exhibits de-
creasing (increasing) returns to scale in the case of a2 < (>) 1 . We inter-
pret a1 and a3 as the measurements of the efficiency of learning by doing
by the production sector. Taking the derivatives of the equation yields
τ ij F j
Z& = ε ij
Z
ε α β
in which τ ij ≡ a1a2 and ε ij ≡ 1 − a2 . The term, τ rj Z rj K rj rj N rj rj , is the con-
tribution to knowledge growth by country j ' s university. It means that
knowledge production of the university is positively related to the capital
stocks, K rj , employed by the university and the number of scientists, N rj .
To interpret the parameter, ε rj , we notice that on the one hand, as the
knowledge stock is increased, the university may more effectively utilize
traditional knowledge to discover new theorems, but on the other hand, a
large stock of knowledge may make discovery of new knowledge difficult.
This implies that ε rj may be either positive or negative. It is reasonable to
assume that the more equipments, books, and buildings, and scientists the
university employs, the more productive it becomes. That is, α rj and β rj ,
are positive.
The universities are financially supported by the governments. In this
model, the governments collect taxes to support the universities. As tax in-
come are used only for supporting the utilities, we have
(r (t ) + δ )K (t ) + w (t )N (t ) = τ F (t ),
kj rj j rj j j j = 1, ... , J . (9.2.9)
The university pays the interest, (r + δ kj )K rj , for the equipments it uses

and the scientists’ wage, w j N rj , it employs; it obtains the research fund,
τ j F j , from the government. We now design the distribution policy to de-
termine the number of scientists and the amount of equipments.
We determine K rj (t ) and N rj (t ) by assuming that country j ' s univer-
sity utilizes its financial resource, τ j F j (t ), in such a way that its output –
contribution to knowledge growth – is maximized. The behavior of the
university is thus formulated by
Max τ rj Z
ε rj
(t )K rjα (t )N rjβ (t ),
rj rj
s.t.: (r (t ) + δ kj )K rj (t ) + w j (t )N rj (t ) = τ j F j (t ).
Country j ' s university allocates the financial resource as follows
α jτ j F j (t ) β jτ j F j (t ) (9.2.10)
K rj (t ) = , N rj (t ) = ,
r (t ) + δ kj w j (t )
where
α rj β rj
αj ≡ , βj ≡ .
α rj + β jr α rj + β rj
If the other conditions remain the same, an increase in the tax rate or
output enables the university to utilize more equipments and to employ
more people. An increase in factor price will reduce the employment level
of the factor.
Full employment and the demand and supply balance

The assumption that the labor force and capital are always fully em-
ployed in each country is represented by
N ij (t ) + N rj (t ) = N j , K ij (t ) + K rj (t ) = K j (t ). (9.2.11)
The total capital stocks employed by the world is equal to the wealth
owned by the world. That is
J J
(9.2.12)
K (t ) = ∑ K (t ) j = ∑ kˆ (t )N
j j .
j =1 j =1
The world production is equal to the world consumption and world net
savings. That is
J
C (t ) + S (t ) − K (t ) + ∑δ kj K j (t ) = F (t ),
j =1
where
J J J
C (t ) ≡ ∑ c j (t )N j , S (t ) ≡ ∑ s j (t )N j , F (t ) ≡ ∑ F j (t ).
j =1 j =1 j =1
We have thus built the model with trade, economic growth, capital ac-
cumulation, knowledge creation and utilization in the world economy in
which the domestic markets of each country are perfectly competitive, in-
ternational product and capital markets are freely mobile and labor is in-
ternationally immobile.
9.2.2 The Dynamics when the World Has the Same Preference
This section examines a special case when the households in the world
have the identical preference and the depreciation rates are the same
among the economies. We are interested in this case because we can ex-
plicitly determine dynamic properties of the system as shown below. We
require
ξ = ξ j , λ = λ j , δ k = δ kj , α = α j , j = 1, ...., J .
We now show that all the variables in the dynamic system can be ex-
pressed as functions of k1 (t ) and Z (t ) at any point. First, from Eqs. (9.2.1)
we obtain
m
k j = M j Z j k1 , j = 1, ... , J , (9.2.13)
in which
1/ β
 A jτ j  m j − m1
M j ≡   , mj ≡ .
 A1τ 1  β
Country j' s capital intensity of the production function can be ex-
pressed as a unique function of the knowledge and country 1' s capital in-
tensity of the production function. The ratio between any two countries’
capital intensities is related to the two countries’ tax rates and the level of
the knowledge. We determine the rate of interest and the wage rates as
functions of k1 (t ) and Z (t ) as follows
r = τˆ1 A1αZ m1 k1− β − δ k , w j = τˆ j A j β α jα Z j j k1α , j = 1, L, J . (9.2.14)

αm + m
From Eqs. (9.2.1) and (9.2.10), we have

K rj α jτ j N rj β jτ j (9.2.15)
= , = .
K ij τ jα j N ij τ j β j
From Eqs. (9.2.11) and (9.2.15), we solve the capital and labor distribu-
tion between the production sector and the university in country j as fol-
lows
K qj (t ) = aqj K j (t ), N qj = bqj N j , q = i , r , j = 1, ... , J , (9.2.16)
where
α jτ j τˆ jα j
arj ≡ , aij ≡ ,
α jτ j + τˆ jα α jτ j + τˆ jα
β jτ j τˆ j β
brj ≡ , bij ≡ .
β jτ j + τˆ j β β jτ j + τˆ j β
The labor distribution is constant as it is determined by the tax rate and
capital distribution is proportional to the total capital stocks employed by
the country. By k j = K ij / N ij and Eqs. (9.2.16) and (9.2.13), we have
m
K j = M j Z j k1 , j = 1, ..., J , (9.2.17)
where M j = N ij M j / aij . Adding all the equations in (9.2.17) yields
K = k1Λ 0 (Z ), (9.2.18)
∑
J
where we use K = j =1
K j and
J
Λ 0 (Z ) ≡ ∑ M j Z
mj
.
j =1
m
From F j = A j Z j K ijα N ijβ and Eqs. (9.2.16) and (9.2.17), we have
F j = aijα A j N ijβ M αj Z
m j +αm j
k1α . (9.2.19)
Substituting Eqs. (9.2.16), (9.2.17) and (9.2.19) into Eq. (9.2.8), we

have
Z& = Λ (k1 , Z ) ≡
∑ {τ }
J
m j + αm j − ε ij α α β α rj m j + ε rj α
ij aijα Aj N ijβ M αj Z k1α + τ rj arj rj M j rj N rj rj Z k1 rj − δ z Z . (9.2.20)
j =1
We see that the motion of Z can be described as a unique function of

k1 and Z .
From Eqs. (9.2.2) and (9.2.3), we have yˆ j = (1 + r )kˆ j + w j . Substituting
s j = λyˆ j and the above equations into Eqs. (9.2.7), we have
& (9.2.21)
kˆ j = λw j − (1 − λ − λr )kˆ j .
Multiplying the equation for k̂ j by N j and then adding the J resulted

equations, we have
( ) (9.2.22)
J
K& = λβ k1α ∑τˆ j A jα jα Z j j − λ − τ 1 A1αλ Z m1 k1− β K ,
αm + m
j =1
∑ kˆ j N j and λ ≡ 1 − λ + λδ k .
J
where we use Eqs. (9.2.14) and K = j =1
Taking derivatives of Eq. (9.2.18) with respect to t yields
Kk&1  J m −1 
(9.2.23)
K& = +  k1 ∑ m1 M j Z j  Z& .
k1  j =1 
Substituting Eqs. (9.2.22), (9.2.18) and (9.2.22) into Eq. (9.2.23) yields
λβ k1α
( )
J
∑τˆ A α
αm j + m j
k&1 = j j
α
j Z − λ − τˆ1 A1αλZ m1 k1− β k1
Λ0 j =1
 J m −1  Λ
(9.2.24)
−  k1 ∑ m1 M j Z j  .
 j =1  Λ0
Summarizing the above results, we obtain the following lemma.
Lemma 9.2.1
Assume that all the households in the world have the same preference. The
motion of the two variables, k1 (t ) and Z (t ), are given by two differential
Eqs., (9.2.20) and (9.2.24). For any given k1 (t ) and Z (t ), r (t ) and
w j (t ), j = 1, L, J , by (9.2.14). The variables, kˆ j (t ), are solved by Eqs.

(9.2.21) as follows
kˆ j (t ) = e ∫
− (1 − λ − λr ) d τ
(
h j + λ ∫ w j (τ ) e ∫
(1 − λ − λr )d τ
dτ , )
j = 1, L , J , (9.2.25)
where h j are constants to be determined by initial conditions. For any
given positive values of Z (t ), k1 (t ) and kˆ j (t ) at any point of time, the
other variables are uniquely determined by the following procedure:
k j (t ), j = 2, L, J by (9.2.13) → N qj , q = i , r , j = 1, L, J by (9.2.16)
→ K j (t ) by (9.2.17) → K qj (t ) by (9.2.16) → yˆ j = (1 + r )kˆ j + w j by
(9.2.20) → c j (t ) and s j (t ) by (9.2.6) → F j = Z j K ijα N ijβ .
m
The dynamic properties of the world economy are determined by two

differential equations. Equilibrium is determined by
∑ {τ }
J
m j +αm j −ε ij α α β α rj m j + ε rj α
ij aijα A j N ijβ M αj Z k1α + τ rj arj rj M j rj N rj rj Z k1 rj
j =1
= δzZ ,
( ) (9.2.26)
J
λβ k1α ∑τˆ j A jα jα Z
αm j + m j
− λ − τˆ1 A1αλZ m1 k1− β k1Λ 0 = 0 .
j =1
By the second equation in Eqs. (9.2.26), we solve
λ
1/ β (9.2.27)
k1 = Ω10/ β   ,
λ 
where
β J
Ω 0 (Z ) ≡ ∑τˆ A α
αm j + m j
j j
α
j Z + τˆ1 A1αZ m1 .
Λ0 j =1
Substitute Eq. (9.2.27) into the first equation in Eqs. (9.2.26)

Ω(Z ) ≡
∑ {τ }− δ
J
m j +αm j −ε ij −1 α rj m j + ε rj −1 α /β
ij Z Ωα0 / β + τ rj Z Ω 0 rj z = 0, (9.2.28)
j =1
where
α /β
λ
τ ij ≡ τ ij aij A j N ij M j  
α β α
> 0,
λ 
α rj / β
α α β λ
τ rj ≡ τ rj arj rj M j rj N rj rj   > 0.
λ 
From Lemma 9.2.1 and the above discussions, we have the following
corollary.
Corollary 9.2.1
The number of equilibrium points is the same as the number of solutions of
Ω(Z ) = 0 , for Z > 0 . For any solution Z > 0 , all the other variables are
uniquely determined by the following procedure: k1 by (9.2.27) → r and
w , j = 1, L, J , by (9.2.14) → kˆ = λw / (1 − λ − λr ) → k , j =
j j j j
2, L, J by (9.2.13) → N qj , q = i , r , j = 1, L, J by (9.2.16) → K j by
(9.2.17) → K qj by (9.2.16) → yˆ j = (1 + r )kˆ j + w j by (9.2.20) → c j and
m
s j by (9.2.6) → F j = A j Z j K ijα N ijβ .
The number of equilibrium points is the same as the number of solutions

of Ω(Z ) = 0 , for Z > 0 . As the expression is tedious, it is difficult to ex-
plicitly judge under what conditions the equation has a unique or multiple
equilibrium points. To see that equation (9.2.28) may have either a unique
or multiple equilibrium points, we are concerned with a case that all the
countries have identical population, identical production function, equal
tax rate, and identical learning by doing and university’s knowledge crea-
tion functions. In this case, the world economy is the same as a single
economy. It is straightforward to show that in this case Eq. (9.2.28) be-
comes
Ω(Z ) = τ 0i Z xi + τ 0 r Z xr − δ z = 0 ,
in which we omit index j as all the countries are identical and

m αrm
xi ≡ − ε i − 1, xr ≡ + ε r − 1,
β β
τ 0i ≡ Jτˆi A(βα ατ + ατ ) > 0 , τ 0 r ≡ Jτˆr (βα ατˆ + ατˆ ) r

α /β α /β
> 0.
It can be shown that the dynamics in this case is the same as that of the
model in Sect.9.1. In this case, the dynamic properties of the model have
been examined. The properties are summarized in the following corollary.
Corollary 9.2.2
If xi < 0 and xr < 0 (or xi > 0 and xr < 0 ), the system has a unique sta-
ble (unstable) equilibrium point; and if xi < 0 and xr < 0 ( xi > 0 and
xr < 0 ), the system may have none, one, or two equilibrium points. When
the system has two equilibrium points, the one with higher value of Z is
stable and the other one is unstable.
By the definitions of xi and xr , we interpret xi and xr respectively as

measurements of returns to scale of the production sector and university in
the dynamic system. When x j < (>) 0 , we say that sector j displays de-
creasing (increasing) returns to scale in the dynamic economy. The above
proposition tells us that if the both sectors display decreasing (increasing)
returns, the dynamic system has a unique equilibrium; if one sector dis-
plays decreasing (increasing) returns and the other sector exhibits increas-
ing (decreasing), the system may have none, one, or two equilibrium
points.
9.2.3 The World Economic Dynamics
The previous section examined the dynamic properties when the world
population has an identical preference. It is straightforward to carry out
dynamic analysis as the world economy is actually controlled only by two-
dimensional differential equations. We will not further examine the behav-
ior of the system because we will simulate the model when the households
have different preferences. This section shows that in general case the dy-
namics of the world economy can be expressed by a ( J + 1) − dimensional
differential equations system.
Lemma 9.2.2
The dynamics of the world economy is governed by the following
(J + 1) − dimensional differential equations system with Z (t ), k1 (t ) and
kˆ (t ), j = 2, L , J , as the variables
j
( )
J
Z& = Λ (k1 , Z ) ≡ ∑ τ ij Z j ij φ j j + τ rj Z rj φ j rj − δ z Z ,
m −ε α ε α
j =1
( {} )
k&1 = Λ1 k1 , kˆ j , Z ≡
J J
 1
∑ n j Λ j + λ1 w1 − n0 Rψ − R ∑ n j k j − n0ψ Z Λ 
ˆ ,
 j =2 j =2  n0ψ k1
( )
kˆ j = Λ j k1 , k j , Z ≡ λ j w j − (1 − λ j − λ j r )kˆ j , j = 2 , ... , J ,
&
in which φ j , R , Λ j , ψ , ψ Z , ψ k1 , r and w j are unique functions of

Z (t ), k1 (t ) and kˆ j (t ) at any point of time, defined in Appendix, and
n0 , n j , and τ qj are parameters defined in the appendix. For any given
positive values of Z (t ), k1 (t ) and kˆ j (t ) at any point of time, the other vari-
ables are uniquely determined by the following procedure: kˆ1 (t ) by
(A.9.1.7) → k j (t ), j = 2, L, J by (A.9.1.1) → r (t ) → w j (t ), j =
1, L, J by (A.9.1.2) → N qj , q = i , r , j = 1, L, J by (A.9.1.4) → K j (t )
by (A.9.1.5) → K qj (t ) by (A.9.1.4) → yˆ j (t ) by (A.9.1.8) → c j (t ) and
s j (t ) by (9.2.6) → F j = A j Z j K ij j N ij j .
m α β
We have the dynamic equations for the world economy with any num-
ber of countries. The system is nonlinear and is of high dimension. It is
difficult to generally analyze behavior of the system. We now solve equi-
librium problem. For simplicity, we require δ k = δ kj , j = 1, ..., J . Equa-
tions (A.9.1.1) and (A.9.1.2) now become
k j = φ j (k1 , Z ) = τ kj Z j k1 1
m β /β j
,
w j = φ j (k1 , Z ) = τ wj Z (9.2.29)
m0 j α
k1 wj , j = 1, L, J ,
where
1/ β j
 τˆ A α  m j − m1 α
τ kj ≡  j j j  , mj ≡ , τ wj ≡ τˆ j A j β jτ kj j
 τˆ1 A1α1  βj
β1α j
m0 j ≡ m j + α j m j , α wj ≡ .
βj
By Eqs. (9.2.7), we have s j = kˆ j . By the definition of R and Eqs.

(9.2.1), we have
R(k1 , Z ) = λ1 (λu1 − τˆ1 A1α1 Z m1 k1− β1 ), (9.2.30)
in which λu1 ≡ 1 / λ1 − 1 + δ k . From the equations for k j in (9.2.29) and

Eqs. (A.9.1.5), we have
J
β /βj
τ kj N ij k1 1 Z
mj (9.2.31)
K =ψ = ∑ .
j =1 aij
From s j = kˆ j and Eqs. (9.2.6), we have yˆ j = kˆ j / λ j . Substitute

yˆ j = kˆ j / λ j into (A.9.1.8)
τ wj Z 0 j k1 wj
m α (9.2.32)
kˆ j = , j = 2 , ..., J ,
λuj − τˆ1α1 A1 Z m1 k1− β1
where we use Eqs. (9.2.29) and (9.2.1) and λuj ≡ 1 / λ j − 1 + δ k . By Eqs.

(A.9.1.12), at equilibrium we have
Ω k (k1 , Z ) ≡
α β /β
τ wj n j Z 0 j k1 wj J τ N k 1 jZ
m mj
J
(9.2.33)
∑ λ
j =1 uj − τˆ α A Z m1 − β1
k
− n0∑
j =1
kj ij 1
aij
= 0,
1 1 1 1
in which n1 = 1, we use Λ = Λ j = 0 , and Eqs. (9.2.29)-(9.2.32). Substi-

β /β j
tuting φ j = τ kj Z j k1 1
m
into Eq. (A.9.1.13) and setting the resulted equa-
tion at equilibrium, we have
Ω Z (k1 , Z ) ≡
∑ (τ τ )− δ
J
α ij β α j /βj α β α rj / β j (9.2.34)
+ τ rjτ kj rj Z rj k1 1
x x
ij kj Z ij k1 1 z = 0,
j =1
in which
xij ≡ m j − ε ij + α ij m j − 1, xrj ≡ ε rj + α rj m j − 1.
We see that two equations, Ω k (k1 , Z ) = 0 and Ω Z (k1 , Z ) = 0 , contain

two variables, k1 and Z . The two equations determine equilibrium values
of k1 and Z . By Eqs. (9.2.32), we determine k̂ j for j = 2 , ... , J . Follow-
ing the procedure in Lemma 9.2.2, we determine all the other variables at
equilibrium. We see that the main problem is to solve Ω k (k1 , Z ) = 0 and
Ω Z (k1 , Z ) = 0 , for k1 > 0 and Z > 0 .
As we cannot explicitly solve the equilibrium values of k1 and Z , we
simulate the model to illustrate properties of the dynamic system. We
specify the parameters as follows:
 N1   3   A1   1   m1   0.4   τ 1   0.05 
               
 N2  =  4,  A2  =  0.8  ,  m2  =  0.2  , τ 2  =  0.04  ,
 N  8  A   0.7   m   0.1  τ   0.02 
 3    3    3    3  
 α1   0.3   α1r   0.4   β1r   0.4   τ i1   0.02 

               
 α 2  =  0.32  ,  α 2 r  =  0.4  ,  β 2 r  =  0.3  , τ i 2  =  0.01  ,
 α   0.31   α   0.4   β   0.2  τ   0.01 
 3    3r     3r     i3   
 τ r1   0.08   ε i1   0.1   ε r1   0.4   ξ 01   0.2 

               
τ r 2  =  0.06  ,  ε i 2  =  0.2  ,  ε r 2  =  0.3  ,  ξ 02  =  0.2  ,
τ   0.03   ε   0.3   ε   0.2   ξ   0.3 
 r3     i3     r3     0.3   
 λ01   0.75  (9.2.35)

   
 λ02  =  0.7  , δ k = 0.05 , δ Z = 0.04 .
 λ   0.65 
 03   
Country 1, 2 and 3' s populations are respectively 3 , 4 and 8. Country
3 has the largest population. Country 1, 2 and 3' s total productivities,
A j , are respectively 1, 0.8 and 0.7 . Country 1, 2 and 3' s utilization ef-
ficiency parameters, m j , are respectively 0.4 , 0.2 and 0.1. Country 1

utilizes knowledge mostly effectively; country 2 next and country 3 util-
izes knowledge lest effectively. We call the three countries respectively as
developed, industrializing, and underdeveloped economies (DE, IE, UE). The
DE has the highest tax rate for supporting research and the UE has the
lowest tax rate. We specify the values of the parameters, α j , in the Cobb-
Douglas productions approximately equal to 0.3. 15 The DE’s learning by do-
ing and university creativity parameters, τ i1 and τ r1 , are the highest among
the countries. The returns to scale parameters in learning by doing, ε ij , are
all positive, which implies that knowledge exhibits decreasing returns to scale
in learning by doing. The depreciation rates of physical capital and knowl-
edge are specified respectively at 0.05 and 0.04 . The DE’s propensity to
save is 0.75 and the UE’s propensity to save is 0.65. The value of the IE’s
propensity is between the two other countries. Similar to the previous sector,
we introduce country j ' s returns to scale parameters for the production
sector and the university respectively as follows:
mj α rj m j
xij* ≡ − ε i1 − 1, xrj* ≡ + ε rj − 1, j = 1, 2 , 3.
βj βj
It is straightforward to calculate that with the specified values of the pa-
rameters, we have xij* < 0 and xrj* < 0 for all j . As no sector in the global
economy exhibits increasing returns to scale, it is expected that the dy-
namic system has a unique equilibrium point and it is stable. We now
show that the dynamic system has a unique equilibrium point. Figure 9.2.1
plots the two equations, Ω k (k1 , Z ) = 0 and Ω Z (k1 , Z ) = 0 , for k1 > 0 and
Z > 0 . The solid lines represent Ω k (k1 , Z ) = 0 and the dashed line stands
for Ω Z (k1 , Z ) = 0 .
From Fig. 9.2.1, we see that the two equations have multiple solutions.
Nevertheless, it can be shown that only the following solution
k1 = 20.567 , Z = 20.610 .
is meaningful and all the other variables are not economically meaningful.
For instance, we also have a solution as k1 = 2.195 and Z = 7.726 . Nev-
15 The value is often used in empirical studies. For instance, Abel and Bernanke
(1998).
ertheless, this point is economically meaningless because at this point, we

have
Z
25
20
ΩZ = 0
15
10
Ωk = 0
k1
5 10 15 20 25 30
Fig. 9.2.1. Solutions of Eqs. (9.2.33) and (9.2.34)
kˆ1 = − 34.230 , kˆ2 = 22.546 , kˆ3 = 3.557 .
As yˆ1 = λ1kˆ1 < 0 , we see that the disposable income is negative, which
means negative consumption in country 1.
We evaluate the other variables at the unique equilibrium point,
k1 = 20.567 and Z = 20.610 , as in Table 9.2.1. The global output is 46.2
and the interest rate is about 6.5 percent. The shares of the global outputs
by the DE, ID and UD are respectively 52 , 23.5 and 24.5 percent. The
population shares of the three economies are respectively 20 , 26.7 and
53.3 percent. The per-worker output levels of the DE, ID and UD are re-
spectively 8.31, 2.78 and 1.43. The differences in labor productivity are
mainly due to the differences in knowledge utilization efficiency. The table
also gives the labor and capital distributions between the sectors in each
country and the capital distribution among the three countries. More than
half of the global capital stocks is employed by the DE. The DE uses more
capital stocks in research than the IE, even though its number of scientists
is less than the number in the IE. The wage rates in the DE, ID and UD are
respectively 5.53 , 1.82 and 0.97 .
Table 9.2.1. The equilibrium values of the global economy
Z K F r C
20.610 126.665 46.199 0.065 39.831
Country 1 Country 2 Country 3 National shares
F1 24.026 F2 10.848 F3 11.326 F1 / F 0.520
F1 / N i1 8.310 F2 / N i 2 2.783 F3 / N i 3 1.430 F2 / F 0.235
K1 64.681 K2 30.791 K3 31.193 F3 / F 0.245
K i1 59.465 Ki2 28.939 Ki3 29.881 K1 / K 0.511
K r1 5.216 Kr2 2.132 K r3 1.311 K2 / K 0.243
N i1 2.891 Ni2 3.898 N i3 7.922 K3 / K 0.246
N r1 0.109 Nr2 0.149 N r3 0.078 Kˆ / Kˆ
1
0.649
K̂1 82.265 K̂ 2 24.888 K̂ 3 19.511 Kˆ 2 / Kˆ 0.196
C1 21.937 C2 8.889 C3 9.001 Kˆ / Kˆ
3
0.154
w1 5.530 w2 1.817 w3 0.967 C1 / C 0.551
k̂1 27.422 k̂ 2 6.222 k̂3 2.439 C2 / C 0.223
ŷ1 34.734 ŷ 2 8.444 ŷ3 3.565 C3 / C 0.226
c1 7.313 c2 2.222 c3 1.126
The trade balances of the three countries are given by

( )
E j (t ) = Kˆ j (t ) − K j (t ) r (t ), j = 1, 2 , 3.
When E j (t ) is positive (negative), we say that country j is in trade

surplus (deficit). When E j (t ) is zero, country j ' s trade is in balance. We
calculate the trade balances at equilibrium as follows
E1 = 1.146 , E2 = − 0.385 , E3 = − 0.761.
The DE is in trade surplus and the other two economies in trade deficit.
So far we have been concerned with equilibrium. Although we did not
prove the stability of the equilibrium point, we expect that the equilibrium
is stable. We start with different initial states not far away from the equi-
librium point and find that the system approaches to the equilibrium point.
This implies that the system is locally stable. In Fig. 9.2.2, we plot the mo-
tion of the system with the following initial conditions
k1 (0 ) = 17 , kˆ2 (0) = 8 , kˆ3 (0 ) = 3 , Z (0) = 17 .

The system approaches to its equilibrium in the long term.
120 22
100 K 20 F1 0.071
80 18
0.069 10 20 30 40 50t
16
60 F 14 0.068
40
Z 12 F3 0.067 r
10 20 30 40 50
t t 0.066
10 20 30 40 F 50
(a) Z (t ), K (t ) and F (t ) (b) F j (t ) (c) r (t )
2
20
C1 80
18 70 1
0.75 E1
16 60 K̂1 0.5
50 0.25
14
12 40 K̂ 2 -0.25 10 E2 30
20 40 50t
30 -0.5
10
C t t -0.75 E3
10 20C 30 340 50 10 20
2 K̂303 40 50
(d) C j (t ) (e) K̂ j (f) E j (t )
7
5 w1 25
4
6 c1
20
k̂1 5
3 15 4
2 w2 10 3 c2
k̂2 2
c3
w3 t 10 20 30 40 50
t t
10 20 30 40 50 10 20 30 40 50
k̂3
(g) w j (t ) (h) k̂ j (i) c j (t )
Fig. 9.2.2. The motion of some variables
9.2.4 Knowledge Utilization Efficiency
We simulated the motion of the dynamic system. It is important to ask

questions such as how a developing economy like India or China may af-
fect the global economy as its technology is improved or population is
enlarged; or how the global trade patterns may be affected as technologies
are further improved or propensities to save are increased in developed
economies like the US or Japan. The rest of this section examines effects
of changes in some parameters on dynamic processes of the global eco-
nomic system.
First, we examine the case that all the parameters, except country 1' s
knowledge utilization efficiency, m1 , are the same as in (9.2.35). We in-
crease the knowledge efficiency parameter, m1 , from 0.4 to 0.45. The
simulation results are demonstrated in Fig. 9.2.3. In the plots, a variable
∆x j (t ) stands for the change rate of the variable, x j (t ), in percentage due
to changes in the parameter value from m10 ( = 0.4 in this case) to m1
( = 0.45 ). That is
x j (t ; m1 ) − x j (t ; m10 ) (9.2.36)
∆x j (t ) ≡ ×100 ,
x j (t ; m10 )
where x j (t ; m1 ) stands for the value of the variable x j with the parameter
value m1 at time t and x j (t ; m10 ) stands for the value of the variable x j
with the parameter value m10 at time t . We will use the symbol ∆ with
the same meaning when we analyze other parameters.
As the DE improves its knowledge utilization efficiency, the knowledge
and capital of the global economy are increased; the output level of the
global economy falls initially and then rises. The DE’s output level rises;
the other two countries’ output levels fall initially and then rise. As the rate
of interest rises initially and knowledge rises but not much initially, we see
that the costs of production are high for the IE and UE and their productiv-
ities are not much improved, the two economies’ output levels fall initially.
As time passes, the world accumulates more knowledge and the rate of in-
terest falls, the IE’s and UE’s output levels are increased. We see that in
the long term the DE’s trade balance is improved and the other two
economies’ trade balances slightly deteriorate. In the long term the wage
rates and the levels of per capita consumptions and wealth in the three
economies are all improved. Hence, we conclude that as UE improves its
knowledge utilization efficiency, all the consumers in the globe benefit in
the long term.
We now examine effects of the underdeveloped economy’s knowledge
efficiency upon the global economy. We allow
m3 : 0.1 ⇒ 0.2 .
The effects of the UE’s improvement in knowledge utilization are pro-
vided in Fig. 9.2.3. As in the case when the DE improves its knowledge
utilization efficiency, the knowledge and capital of the global economy are
increased; different from the case when the DE improves its knowledge
utilization efficiency, output level of the global economy rises all the time.
The UE’s output level rises; the other two countries’ output levels are af-
fected slightly. The rate of interest rises over the time. The DE and IE’s
trade balances are improved and the UE’s trade balance deteriorates. In the
long term the wage rate, the per-capita wealth and consumption level are
increased; the the wage rate, the per-capita wealth and consumption level
of the DE and IE are effected slightly.
It should be noted that as the UE improves its knowledge utilization ef-
ficiency, the economic variables of the UE are improved, but some vari-
ables of the other economies might not be improved. For instance, Figure
9.2.5 illustrates the case that the wage rate and the output level in the IE
are actually reduced. This implies that, for instance, if India and China
more effectively apply knowledge, economies like Taiwan and Korea
might be hurt, even though the US and Japan may benefit
25
20
30 ∆F1 20
15
∆K 15
10
∆Z 20
5
t 10 ∆F2
10
5
∆r
10 20 30 40 50
-5 t t
-10 ∆F 10 20 30 40 ∆F503 10 20 30 40 50
(a) ∆Z (t ), ∆K (t ) and ∆F (t ) (b) ∆F j (t ) (c) ∆r (t )
30 ∆C1 30 0.4 ∆E2

20 ∆K̂1 0.2
20
10 ∆K̂ 2 ∆E3 10 20 30 40 50
10 ∆C 2 -0.2
50 t -0.6
-0.4
∆C3 10 20 ∆K̂ 340
30 ∆E1
10 20 30 40 50t -10
-0.8
-10 -20
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
30
30 ∆w1 20 ∆k̂1
30
∆c1
20
20 10
∆k̂ 10 ∆c2
10 ∆w2 10 20 30 ∆k̂
2
40 350 t
t
-10 10 20 30 ∆c403 50 t
10 20 30
∆w350
40 -20 -10
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.3. The developed economy improves its knowledge utilization efficiency
50 5
12
10
∆F 40 ∆F3 4 ∆r
8 ∆K 30 3
6
20 2
4 ∆Z 10
2 ∆F1 1
10 20 30 40 50 t 10 20 30 ∆F 50 t
40 10 20 30 40 50t
2
50 50
∆C 3 ∆K̂ 3
1
40 40
0.5 ∆E1
30 30 ∆E2
20 20 10 20 30 40 50 t
10 ∆C1 10 ∆K̂1 -0.5 ∆E3
t -1
10 20
∆C 2
30 40 50 10 20
∆K̂ 240
30 50
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
50
∆w3 50
∆k̂3 50 ∆c3
40 40 40
30 30 30
20 20 ∆k̂1 20
10 10 10 ∆c1
∆w1 t
10 20 30 40∆w50t 10 20 ∆k̂ 40
30 50 t 10 20 30∆c240 50
2 2
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.4. The underdeveloped economy raises knowledge utilization efficiency
0.1
10 20 30 40 50
-0.1
-0.2
-0.3
-0.4
-0.5 ∆F2 = ∆w2
-0.6
Fig. 9.2.5. The negative effects on F2 and w2
We now examine effects of changes in creativity parameters in learning

by doing and research, τ ij and τ rj . We increase the DE’s learning by do-
ing efficiency as follows: τ i1 : 0.2 ⇒ 0.25. The effects are plotted in Fig.
9.2.6. The knowledge, global wealth and output levels are increased. The
rate of interest rises initially and falls later on. The total output and con-
sumption levels, total wealth, per capita consumption levels, and per capita
wealth levels of the three economies are all increased. The trade balance of
the DE improves and the other two economies deteriorate. It can be seen
that the effects of change in other τ ij or any τ rj are similar.
20
10 ∆F1
15 ∆Z 8
0.5
∆F2 0.25
t
10 ∆K 6
-0.25 50 100 150 200
5 ∆F
4
∆F3 -0.5
2 -0.75
-1 ∆r
50 100 150 200
t 50 100 150 t
200
10 10
8 ∆C1 8 ∆K̂1 0.06
6 ∆C2 6 ∆K̂ 2 0.04 ∆E1

0.02
4 4
∆K̂ 3 50 100 150 200
t
2
∆C3 2 -0.02 ∆E2
50 100 150 200 t 50 100 150 200t ∆E3
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
10 10
10
8
∆w1 8 8 ∆c1
∆k̂1 ∆k̂ 2 ∆c2
6
∆w2 6 6
4 4 4
∆c3
2 ∆w3 2
∆k̂3 2
50 100 150 200

t 50 100 150
t
200 50 100 150 200
t
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.6. The developed economy improves its creativity in research
9.2.5 The National Research Policies and the Global Economy
We now study how changes in the research policies affect the global econ-
omy. The DE increases its tax rate as follows: τ 1 : 0.05 ⇒ 0.07 . The DE’s
tax rate is increased from 5 percent to 7 percent. As the DE strengthens
its research policy, the knowledge, global wealth and output level are in-
creased. The rate of interest is reduced. The three economies’ output levels
are increased. The wage rate, total consumption and wealth levels, per-
capita wealth and consumption levels in each economy are increased. The
IE’s trade balance improves and the other two countries’ trade balances de-
teriorate. It should be remarked that the desirable results for the DE to in-
crease its tax for supporting research don’t hold if the tax rate is too high.
For instance, if we increase the tax rate as follows: τ 1 : 0.05 ⇒ 0.5 , then
country 1' s consumption, output level and wealth will fall. Evidently, if
few workers are engaged in economic production, all the knowledge cre-
ated by the DE will only increase the output levels of the other two coun-
tries and the DE itself does not economically benefit.
14 5 ∆F1 1 ∆r
12
10 ∆Z 4
t
8 3
2
∆F2 -1
50 100 150 200
6
4
2
∆K 1
∆F3 t
-2
50 100 150 200 -3
∆F 100
50 150 200
t -1
5 ∆C1 5 ∆K̂1
4 4 ∆K̂ 2 0.08
3 ∆C 2 3
0.06
0.04 ∆E2
2 2 0.02 ∆E3
1 ∆C3 1 ∆K̂ 3 50 100 150 200t
-0.02
∆E1
200 t 200 t
-0.04
50 100 150 50 100 150
-1 -1
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
5 ∆w1 5 ∆k̂1 5 ∆c1

4 4 4
3 ∆w2 3 ∆k̂ 2 3 ∆c 2
2
2 2
1
∆w3 t 1 ∆k̂3 1 ∆c3
-1 50 100 150 200
50 100 150 200 t 50 100 150 200 t
-2 -1 -1
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.7. The developed economy increases its tax rate
We now allow the UE to increase its tax rate as follows:

τ 3 : 0.02 ⇒ 0.04 . As the UE strengthens its research policy, the knowl-
edge, global wealth and output level are increased. The rate of interest is
increased. The DE and IE’s output levels are increased; but the UE’s out-
put level is reduced. The wage rate, total consumption and wealth levels,
per-capita wealth and consumption levels in the DE and the IE are in-
creased; the wage rate, total consumption and wealth levels, per-capita
wealth and consumption levels in the UE are reduced. The DE and IE’s
trade balances improve and the UE’s trade balance deteriorates. This im-
plies that the UD will harm itself if it strengthens the research policy.
2 0.7
3 ∆Z ∆F1 0.6
1 0.5
2
∆K ∆F2 t
0.4
0.3
∆r
1 50 100 150 200
0.2
∆F -1 ∆F3 0.1
50 100 150 200 t -2 50 100 150
t
200
∆C1
2 2
∆K̂1 0.04
1 1
0.02 ∆E1
∆C 2 ∆ K̂
50 100 150 200 t 50 100 150 2 200 t 50 100∆E
150 200t
-1 -1 2
-0.02
-2
∆C3 -2
∆K̂ 3 -0.04 ∆E3
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
2 2
∆k̂1 2 ∆c1
1 ∆w1 1 1
∆w
150 2 200 t ∆ k̂ t ∆c2 t
50 100 50 100 150 2 200 50 100 150 200
-1 -1 -1
-2
∆w3 -2
∆k̂3 -2 ∆c3
-3
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.8. The underdeveloped economy increases its tax rate
9.2.6 Preference Change
We now allow the DE to increase its propensity to save as follows:

λ01 : 0.75 ⇒ 0.78 .
The results are plotted in Fig. 9.2.9. As the DE increases its propensity to
save, the knowledge, global wealth and output level are increased. The rate
of interest is reduced. The DE and IE’s output levels are increased; but the
UE’s output level is reduced. The wage rate, total consumption and wealth
levels, per-capita wealth and consumption levels in the DE and the IE are
increased; the wage rate, total consumption and wealth levels, per-capita
wealth and consumption levels in the UE are reduced. The DE trade bal-
ance improves and the IE and UE’s trade balance deteriorate. As the DE
increases the propensity to save, the UD loses.
4 t
5
∆K 3 ∆F1 ∆F2 -0.5 50 100 150 200
4
3
∆Z 2 -1
-1.5
1 -2
2
∆F t -2.5
1
-1
50 100 150 200 -3
-3.5 ∆r
50 100 150 200 t
∆F3 -2
4 8
3 ∆C1 ∆C2 6 ∆K̂1 0.1 ∆E1
2
4 0.05
1
t ∆K̂ 2
50 100 150 200
2
50 100 150
t
200
-1
-2 50 100 150 200
t -0.05 ∆E2
-3 ∆C3 -2 ∆E3
∆K̂ 3
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
4 4
3 ∆w1 8
∆k̂1 3 ∆c1
2 ∆w2 6
2
1 4
∆k̂ 2 1 ∆c2
50 100 150 200
t 2
50 100 150 200
t
-1 -1
-2 200 t -2
-3 ∆w3 -2
50 100 150
∆ kˆ3 -3 ∆c3
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.9. The developed economy increases its propensity to save
We now allow the UE to increase its propensity to save as follows:

λ03 : 0.65 ⇒ 0.7 . The results are plotted in Fig. 9.2.10. The knowledge,
global wealth and output level are increased. The rate of interest is re-
duced. The DE and IE’s output levels are increased; but the UE’s output
level is reduced. The wage rate, total consumption and wealth levels, per-
capita wealth and consumption levels in the DE and the IE are increased;
the wage rate, total consumption level, per-capita consumption level in the
UE are reduced. The UE trade balance improves and the IE and DE’s trade
balance deteriorate.
The relationship between population change and economics is a chal-
lenging area. Although this study assumes the population fixed, it is impor-
tant to examine effects of changes in the population sizes. As different
countries have different levels of knowledge utilization efficiency and
creativity, increases in the population sizes may have different effects upon
the global economy. For instance, it is important to examine implications
of possibly negative population growth in developed economies and rapid

population growth in underdeveloped economies.16
4 ∆F1 200t
∆Z 2 -0.2 50 100 150
3
∆K 1 ∆F2 -0.4
2 -0.6
50 100 150
t
200 -0.8
1
∆F -1 -1
∆r
t ∆F3 -1.2
50 100 150 200 -2
2 ∆C1 6
5 ∆K̂ 3 0.06
50 100
∆C2 t
150 200 4
0.04 ∆E3
0.02
-2 3
∆C3 2 ∆K̂ 2 50 100 150 200t
-4 1 ∆K̂1 -0.02 ∆E2
50 100 150
t
200
-0.04 ∆E1
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
2 ∆w1 6
∆k̂3 2 ∆c1
1 ∆w2 5
4
∆c2 t
50 100 150 200
200 t 3
∆c3
-1
50 100 150
2 ∆k̂1 -2
1 ∆k̂ 2 -4
-2
-3
∆w3 t
50 100 150 200
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.10. The underdeveloped economy increases its propensity to save
9.2.7 Population Change and the Global Economy
First, we are concerned with the effects of an increase in the DE’s popu-
lation as follows: N1 : 3 ⇒ 3.5 . The effects are plotted in Fig. 9.2.12. The
knowledge, global wealth and output levels are increased. The rate of in-
terest falls. The total output and consumption levels, total wealth, per cap-
ita consumption levels, and per capita wealth levels of the three economies
16 It has been observed that the effect of population growth varies with the level
of economic development and can be positive for some developed economies.
Theoretical models with human capital predict situation-dependent interactions
between population and economic growth (see, Ehrlich and Lui, 1997; Galor and
Weil, 1999; and Boucekkine et al. 2002).
are all increased in the long term. The trade balance of the DE improves
and the other two economies deteriorate.
25
t
20
∆K
30
25 ∆F1 -0.5 50 100 150 200
15 ∆F 20 -1
15 -1.5
10
10 ∆F2 -2
-2.5 ∆r
5
∆Z 5 ∆F3 t -3
50 100 150 200
t 50 100 150 200
30
25
30
25 ∆K̂1 0.15 ∆E1
20 ∆C1 20
0.1
15 15 0.05
10 ∆C2 10 ∆K̂ 2 t
5 5 -0.05
50 100
∆150
E2 200
50∆C3100 150 200

t 50∆ K̂ 3100 150 200
t -0.1 ∆E3
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
15
12.5 ∆w1 12.5
∆k̂1 12.5 ∆c1
10 10
10 ∆k̂ 2
7.5 7.5 7.5
∆c2
5 ∆w2 5
2.5 ∆k̂3
5
2.5
2.5
∆w3 t t ∆c3 t
-2.5 50 100 150 200 -2.5 50 100 150 200 -2.5 50 100 150 200
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.11. The developed economy increases its population
We now examine the case when the DE’s population changes as fol-
lows: N 3 : 8 ⇒ 9 . The effects are plotted in Fig. 9.2.12. The knowledge,
global wealth and output levels are increased. The rate of interest rises.
The total output, total consumption levels, and total wealth of each econ-
omy are all increased in the long term. The trade balance of the DE and IE
improve and the trade balance of the UE deteriorates.
In the DE and IE, the wage rates, per capita consumption levels and per
capita wealth are all increased; in the UE, the wage rate, per capita con-
sumption level and per capita wealth are all reduced.
5 10 2
4 ∆Z ∆F 8 ∆F3
3
1.5 ∆r
∆K 6
1
2 4
∆F1 0.5
1 2
t 50 100∆F2150 200
t 50 100 150
t
200
50 100 150 200
12 12
10 10 ∆K̂ 3 0.1 ∆E1
8 ∆C3 8 0.05 ∆E2
200t
6 6
4 ∆C1 4 ∆K̂1 -0.05
50 100 150
2 2 -0.1 ∆E3
∆C ∆K̂ 2200 t
50 100 2
150 200 t 50 100 150
(d) ∆C j (t ) (e) ∆K̂ j (f) ∆E j (t )
2 ∆w1
3
2 ∆k̂1
3
2
∆c1
1
∆w2 1
∆k̂ 2
1 ∆c2
200 t t 200t
50 100 150
-1 50 100 150 200 50 100 150
-1 -1
-2
-3 ∆w3 -2
∆k̂3 -2 ∆c3
-3 -3
(g) ∆w j (t ) (h) ∆k̂ j (i) ∆c j (t )
Fig. 9.2.12. The underdeveloped economy increases its population
9.3 Conclusions
This chapter proposed a multi-country growth model with capital accumu-

lation and knowledge creation. Different from the growth models with the
Ramsey approach, the alternative utility function determines saving and
consumption without leading to a higher dimensional dynamic system like
by the traditional approach. The dynamics of J -country world economy is
controlled by a ( J + 1) -dimensional differential equations system. We also
simulated the motion of the model and demonstrated effects of changes in
the parameters. It is well known that one-sector growth model has been
generalized and extended in many directions. It is not difficult to general-
ize our model along these lines. It is straightforward to develop the model
in discrete time. We may analyze behavior of the model with other forms
of production or utility functions. There are multiple production sectors
and households are not homogenous. In the contemporary literature, pri-
vate research and endogenous population have been emphasized.
Appendix 369
Appendix
A.9.1 Proving Lemma 9.2.2
First, from Eqs. (9.2.1) we obtain

1/ β j
 τˆ j A jα j Z j
m

k j = φ j (k1 , Z ) ≡   , j = 1, ..., J ,
 τˆ1 A1α1 Z m1 k1− β1 + δ j  (A.9.1.1)
 
where δ j ≡ δ k1 − δ kj . It should be noted that φ1 = k1 . From Eqs. (9.2.1)
and (A.9.1.1), we determine the wage rates as functions of k1 (t ) and Z (t )
as follows
w j = φ j (k1 , Z ) ≡ τˆ j A j β j Z j φ j j (k1 , Z ), j = 1, L, J . (A.9.1.2)
m α

K rj α jτ j N rj β jτ j (A.9.1.3)
= , = .
K ij τˆ jα j N ij τˆ j β j
From Eqs. (9.2.11) and (A.9.1.3), we solve the capital and labor distri-
bution between the production sector and the university in country j as
follows
K qj (t ) = aqj K j (t ), N qj = bqj N j , q = i , r , j = 1, ... , J , (A.9.1.4)
where
α jτ j τˆ jα j β jτ j
arj ≡ , aij ≡ , brj ≡
α jτ j + τˆ jα j α jτ j + τˆ jα j β jτ j + τˆ j β j
τˆ j β j
bij ≡ .
β jτ j + τˆ j β j
We conclude that the labor distribution is constant as it is determined by
the tax rate and capital distribution is proportional to the total capital
stocks employed by the country.
By k j = K ij / N ij and Eqs. (A.9.1.4), we have
N ij k j (A.9.1.5)
Kj = , j = 1, ..., J .
aij
As k j are functions of k1 (t ) and Z (t ), we see that K j (t ) are also func-

tions of k1 (t ) and Z (t ). From Eqs. (A.9.1.4), we also solve K rj (t ) as func-
tions of k1 (t ) and Z (t ). We see that the capital distribution among the
countries and between sectors in each country are uniquely determined as
functions of k1 (t ) and Z (t ). By K = ∑ j =1 K j , we see that K is also
J
uniquely determined as a function of k1 and Z . We denote this function as

follows: K = ψ (k1 , Z ).
m α β
Substituting F j = Z j K ij j N ij j into Eq. (9.2.8), we have
Z& = Λ (k1 , Z ) ≡
∑ (τ )
J
β m j − ε ij α β ε α (A.9.1.6)
ij A j N ij j Z K ij j + τ rj N rj rj Z rj K rj rj − δ z Z .
j =1
We see that the motion of Z can be described as a unique function of

k1 and Z .
From Eqs. (9.2.12), we solve
J
(A.9.1.7)
kˆ1 = n0ψ (k1 , Z ) − ∑ n j kˆ j ,
j =2
in which
1 Nj
n0 ≡ , nj ≡ , j = 2 , ..., J .
N1 N1
{ } ( )
Introduce kˆ(t ) ≡ kˆ2 (t ), L, kˆJ (t ) . We see that country 1' s per capita
wealth, kˆ1 (t ), can be expressed as a unique function of the knowledge,
country 1 ’s capital intensity of production function and the other coun-
{ }
tries’ per capita wealth, kˆ(t ) , at any point of time.
yˆ j = (1 + r )kˆ j + w j . (A.9.1.8)
Appendix 371
Substituting s j = λ j yˆ j and the above equations into Eqs. (9.2.7), we

have
&
( )
kˆ1 = Λ1 k1 , kˆ1 , Z ≡ λ1 w1 − R (k1 , Z )kˆ1 ,
(A.9.1.9)
( )
kˆ j = Λ j k1 , kˆ j , Z ≡ λ j w j − (1 − λ j − λ j r )kˆ j , j = 2 , ..., J ,
& (A.9.1.10)
in which R(k1 , Z ) ≡ 1 − λ1 − λ1r . Equations (A.9.1.10) are the differential

equations for k j (t ) in Lemma 9.2.2, j = 2 , ... , J . Taking derivatives of Eq.
(A.9.1.7) with respect to t yields
& J
& (A.9.1.11)
k1 = n0ψ k1 k&1 + n0ψ Z Z& − ∑ n j k j ,
j =2
where ψ k1 and ψ Z are the partial derivatives of ψ (k1 , Z ) with respect to

k1 and Z . Equaling the right-hand sizes of Eqs. (A.9.1.9) and (A.9.1.11),
we get
J
&
n0ψ k1 k&1 + n0ψ Z Z& − ∑ n j kˆ j = λ1 w1 − Rkˆ1 .
j =2
Substitute Eq. (A.9.1.7) into the above equation

( {} )
k&1 = Λ1 k1 , kˆ j , Z ≡
J J
 1 (A.9.1.12)
∑ n j Λ j + λ1 w1 − n0 Rψ + R ∑ n j k j − n0ψ Z Λ 
ˆ ,
 j =2 j =2  n0ψ k1
where we use Eqs. (A.9.1.10) and (A.9.1.6). This is the differential equa-
tion for k1 (t ) in Lemma 9.2.2. Substitute Eqs. (A.9.1.4), (A.9.1.5),
(A.9.1.1) and (A.9.1.12) into Eq. (A.9.1.6), we have
( ) (A.9.1.13)
J
Z& = Λ(k1 , Z ) = ∑ τ ij Z j ij φ j j + τ rj Z rj φ j rj − δ z Z ,
m −ε α ε α
j =1
where
α −α β α
τ ij ≡ τ ij A j N ij , τ rj ≡ τ rj arj rj aij rj N rj rj N ij rj .
This is the differential equation for Z (t ) in Lemma 9.2.2. In summary,

we have proved Lemma 9.2.2.

9 Growth and Trade With Capital and Knowledge

Uploaded by

Copyright:

Available Formats

9 Growth and Trade With Capital and Knowledge

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

9 Growth and Trade With Capital and Knowledge

Uploaded by

Copyright:

Available Formats

9 Growth and Trade with Capital and Knowledge

sumption of “inducement through the factor prices”. In 1981 Schultz em-

2 See Arrow (1962), Uzawa (1965), Kennedy (1964), Weisäcker (1966),

are examples along this line.

9.1 A National Growth Model

This section introduces a growth model with endogenous human capital

9.1.1 The OSG Model with Endogenous Human Capital

F (t )  output level of the production sector at time t;

N v (t ) and N e  number of teachers and number of students;

We first model production and consumption. We assume that production

The marginal conditions are given by

where ki ≡ K i / H m N i and τ is the tax rate on the product level.

k&(t ) = s ( yˆ (t ) ) − k (t ) = λyˆ (t ) − k (t ). (9.1.3)

Uzawa took account of trade offs between investment in education and

tion to human capital improvement through education. Human capital

The university distributes its total resource, τF (t ) − w(t )N e , to the teach-

sity’s problem is formed as follows

subject to Eq. (9.1.5). The solution is given by

9.1.2 The Dynamics and Multiple Equilibrium Points

From Eq. (9.1.7), we get N i + N v = N − N e . From Eqs. (9.1.1) and

β v (τN i − (1 − τ )βN e ) (9.1.8)

From Eqs. (9.1.8) and N i + N v = N − N e , we solve

β v (τN i − (1 − τ )βN e ) (9.1.9)

where ni ≡ N / N i (1 + a ). We see that ki (t ) is a function of k (t ) and H (t ).

r (t ) and w(t ) → y = rk + w and yˆ = y + k → c and s by Eqs. (9.1.2)

We now show how to determine by k (t ) and H (t ) at any point of time.

H& = υ e*k α e H β v m + β e m + υi* H m −π −αm k α − δ h H , (9.1.12)

Lemma 9.1.1 guarantees that if we know values of k (t ) and H (t ), then

υ e*k α e H β v m+ β e m + υi* H m−π −αm k α = δ h H . (9.1.13)

From the first equation in Eqs. (9.1.13), we solve k = (λ* / ξ ) H m .

We see that the number of economic equilibrium points is equal to the

a) both exhibit decreasing returns b) both exhibit increasing returns

We can prove the conditions in Fig. 9.1.1. The following proposition

9.1.3 The Path-Dependent Motion of the System by Simulation

8 The proof of the proposition is referred to Zhang (2005a).

Under the above specifications, we have xe = 0.4 and xi = − 0.5. The

scale dominates the economic evolution. Indeed, this kind of infinite

Fig. 9.1.2. Path dependent economic evolution

As far as qualitative features of economic development are concerned,

erature of neoclassical economic growth theory developed in the 1950s

9.1.4 Comparative Dynamic Analysis11

11 Here, we are only concerned with simulation results. We refer to Zhang

(2005a) for the analytical results on comparative statics analysis.

50 100 150 200 250 k

Fig. 9.1.3. The path dependent-development as the education is discouraged

We now examine effects of the propensity to save. We still specify the

Fig. 9.1.4. An increase in the propensity to save

9.2 Trade and Growth with Learning-by-Doing and

It is well known that dynamic-optimization models with capital accumula-

9.2.1 The Multi-Country Trade Model with Capital and

K (t )  the capital stocks of the world economy;

in which Z (t ) (> 0) is the world knowledge stock at time t . Here, we call

where δ kj are the depreciation rate of physical capital in country j and

in which τ j is country j' s tax rate on its production sector.

consumer of country j obtains income

The disposable income is used for saving and consumption.

At each point of time, a consumer distributes the total available budget

At each point of time, consumers have two variables to decide. A con-

H& = υ ek α e H β v m + β e m + υi H m −π −αm k α − δ h H , (9.1.12)

υ ek α e H β v m+ β e m + υi H m−π −αm k α = δ h H . (9.1.13)