Applied Mathematical Sciences, Vol. 4, 2010, no.

65, 3233 - 3238

A Note on the Ramsey Growth Model

with the von Bertalanffy Population Law
Luca Guerrini

Department of Mathematics for Economic and Social Sciences

University of Bologna, Italy
[email protected]

Abstract
In this paper, we consider the Ramsey growth model with CIES
utility function, Cobb-Douglas technology, and von Bertalanffy popu-
lation growth law. We show the model to have a unique non-trivial
steady-state equilibrium, which happens to be a saddle point with a
two dimensional stable manifold. Furthermore, we prove the optimal
path to be non-monotonic over time.

Mathematics Subject Classification: 91B62

Keywords: Ramsey model; von Bertalanffy population

1 Introduction
The Ramsey growth model is a neoclassical model of economic growth based
primarily on the work of the economist and mathematician Frank P. Ramsey
[18], that explains the fundamentals of consumption and capital accumula-
tion in a dynamic real equilibrium setting. It develops the standard Solow
[19] growth model by taking into account an endogenous determination of the
level of savings. Recently, Accinelli and Brida [1], Bucci and Guerrini [5], Fer-
rara and Guerrini [6 − 9], Germanà and Guerrini [10], and Guerrini [12 − 17],
have explored the implications of studying neoclassical growth models where
the change over time of the labor force is governed by the logistic law or
by a bounded population growth rate. Similarly, this framework investigates
the Ramsey model based on the von Bertalanffy [3] model of population dy-
namics. More concretely, within the neoclassical Ramsey growth model, we
assume a neoclassical technology, a constant intertemporal elasticity of sub-
stitution (CIES) utility function, and, following Accinelli and Brida [2], a von
Bertalanffy population growth law. With this setup, we demonstrate that the
3234 L. Guerrini

model has a unique non-trivial steady state equilibrium, which is a saddle

point with a two dimensional stable manifold. Now, many models of growth,
including Ramsey model, have the property that the transitional dynamics are
determined by a one dimensional stable manifold. As a consequence, all the
variables converge to their respective steady states at the same constant speed,
which is equal to the magnitude of the unique stable eigenvalue. By contrast,
in the present model, the stable transitional path may be a two dimensional
locus, thereby introducing important flexibility to the convergence and transi-
tion characteristics. Moreover, in contrast to the standard Ramsey model, we
prove the transitional dynamics to be non-monotonic over time.

2 The model
We start by considering the Ramsey model [18] of a representative consumer
whose objective is to choose per capita consumption c so as to maximize
 ∞ 1− 1
c σ − 1 −ρt
e dt,
0 1 − σ1
. .
subject to the budget constraint k = k α − (δ + L/L)k − c. The parameter
ρ > 0 is the rate of time preference, σ represents the constant intertemporal
elasticity of substitution, k is per capita capital stock, δ > 0 is capital depre-
ciation, α ∈ (0, 1), and L denotes population. Time argument is suppressed
to ease the burden of notations. Following Accinelli and Brida [2], L evolves
.
according to the von Bertalanffy equation [3], L = r(L∞ − L), where L∞ is
a theoretical maximum asymptote size of the labor force, and r is a constant
that determines the speed at which the labor force reaches the asymptote.
Today’s population is given by L0 . In order to solve this optimization problem
we use the Pontryagin maximum principle. The current-value Hamiltonian of
the problem is given by
1    
c1− σ − 1 α rL∞
H(k, c, L, λ) = +λ k − δ+ −r k−c ,
1 − σ1 L
where λ is the costate variable associated to the budget constraint. The con-
ditions for a path to be optimal are
Hc = 0 ⇒ c−1/σ = λ, (1)
 
. . rL∞
α−1
λ = ρλ − Hk ⇒ λ = −λ αk −ρ−δ− +r , (2)
L
together with the budget constraint, the von Bertalanffy equation, the bound-
ary condition k0 > 0, and the transversality condition lim e−ρt λk = 0. Differ-
t→∞
entiating Eq. (1) with respect to time, and using formula (1), we can rid Eq.
A note on the Ramsey growth model 3235

.
(2) of the λ, λ expressions. After rearrangement, we arrive at
 
. rL∞
α
k = k − δ+ − r k − c, (3)
L
 
. α−1 rL∞
c = σc αk −ρ−δ− +r , (4)
L
.
L = r(L∞ − L). (5)

These equations, together with the initial condition k0 , and the transversality
condition lim e−ρt c−1/σ k = 0, constitute the dynamic system which drives the
t→∞
economy over time.

3 Steady state equilibrium and local analysis

We now focus on the steady state, which is defined as a situation in which
the growth rates of consumption, capital and population are zero. An asterisk
below a variable will denote its stationary value. The analysis is confined to
interior steady states only, i.e. we will exclude the economically meaningless
solutions such as k∗ = 0, c∗ = 0 or L∗ = 0.
Lemma 3.1. There exists a unique non-trivial steady state (k∗ , c∗ , L∗ ), where
  1−α
1
α (1 − α)δ + ρ
k∗ = , c∗ = k∗α − δk∗ = k∗ , L∗ = L∞ .
δ+ρ α

Proof. The statement follows equating Eqs. (3) − (5) to zero, and then solving
the resulting system.
Proposition 3.2. The steady state equilibrium (k∗ , c∗ , L∗ ) is a saddle point.
Proof. Linearizing the dynamical system (3)−(5) around (k∗ , c∗ , L∗ ) yields the
approximated dynamic system
⎡ . ⎤ ⎡ ⎤ ⎡ ⎤
k k − k∗ ρ −1 bk∗
⎢ . ⎥
⎣ c. ⎦ = J ⎣ c − c∗ ⎦ , where J = ⎣ M 0 σbc∗ ⎦ ,
∗ ∗
(6)
L L − L∗ 0 0 −r

and M = σα(α−1)k∗α−2 c∗ < 0. We recall that an equilibrium point of a system

of differential equations is called hyperbolic if the Jacobian matrix calculated at
that point has no zero or purely imaginary eigenvalues (no eigenvalue has real
part equal to zero). There exists a general result in the theory of differential
equations, known as the Hartman-Grobman theorem (see, e.g., Guckenheimer
and Holmes [11]), which guarantees that, if the equilibrium point is hyperbolic,
3236 L. Guerrini

in a neighbourhood of the equilibrium point the qualitative properties of the

non-linear system (3) − (5) are preserved by the linearization (6). In our case,
it is immediately seen that one root of J ∗ is λ1 = −r < 0, and the other
two roots are solutions of the equation λ2 − ρλ + M = 0. Thus, we get
λ2 = (ρ − ρ2 + 4M)/2 < 0, λ3 = (ρ + ρ2 + 4M )/2 > 0. Since the matrix
J ∗ has one real positive (unstable) and two real negative (stable) roots, we can
conclude that the steady state is (locally) a saddle point (see, e.g., Blume and
Simon [4]).

Since there are two negative eigenvalues, the stable manifold is a plane going
through the steady state. Consequently, the economy’s speed of convergence,
namely how long it takes for the economy to adjust to the steady state, at
any point of time is a weighted average of the two stable eigenvalues. Clearly,
over time, the weight of the smaller (more negative) eigenvalue declines, so
that the larger of the two stable eigenvalues describes the asymptotic speed of
convergence. Next, because of the transversality condition, the optimal path is
restricted to the stable manifold spanned by the eigenvectors associated with
the negative eigenvalues. The solutions along the stable arm of the saddle-path
are then given by

k − k∗ = A1 eλ1 t + A2 eλ2 t , c − c∗ = A1 v21 eλ1 t + A2 v22 eλ2 t , L − L∞ = A1 v31 eλ1 t ,

where A1 , A2 , A3 are arbitrary constants, to be determined using the initial

conditions, and (1, v2i , v3i ), i = 1, 2, is the normalized eigenvector associated
with the stable eigenvalue λi . We find that v22 = ρ−λ2 > 0, v21 = [σc∗ (λ1 −ρ)−
λ2 k∗ (λ2 − ρ)]/(k∗ λ1 − σc∗ ) > 0, and v31 = (λ2 − λ1 )(ρ − λ2 − λ1 )/b(λ1 k∗ − σc∗ ).
Furthermore, setting t = 0 we obtain that v31 A1 = L0 − L∞ . Since L0 < L∞ ,
we get that v31 A1 < 0. Thus, v31 and A1 are non-zero with opposite signs. As
well, v31 = 0 and sign(v31 ) = sign(λ1 − λ2 ) yield that λ1 = λ2 .

Proposition 3.3. The equilibrium paths of capital and consumption are

non-monotonic over time.

Proof. Monotonicity of capital (resp. consumption) path occurs provided the

. .
equation k = 0 (resp. c = 0) has no solution. This means that there is no
t > 0 such that A1 λ1 eλ1 t + A2 λ2 eλ2 t = 0 (resp. A1 v21 λ1 eλ1 t + A2 v22 λ2 eλ2 t = 0),
namely
    
1 A2 λ2 1 A2 λ2 v22
t= ln − resp. ln − .
λ1 − λ2 A1 λ1 λ1 − λ2 A1 λ1 v21

Let k0 > k∗ (similarly for k0 < k∗ ). We need to show that λ1 − λ2 > 0 and
(−A2 /A1 ) (λ2 /λ1 ) > 1. Now, A1 + A2 = k0 − k∗ > 0. If A1 < 0, then we must
A note on the Ramsey growth model 3237

have A2 > 0, so that −A2 /A1 > 1. As well, it is v31 > 0, yielding λ1 > λ2 , i.e.
λ2 /λ1 > 1. If A2 < 0, then A1 > 0 and v31 < 0. In this case, we get λ1 −λ2 < 0,
0 < (−A2 /A1 ) (λ2 /λ1 ) < 1, so that there exists t > 0. An analogous discussion
holds for c.

4 Conclusions
In this paper, we have considered a modified version of the Ramsey growth
model, obtained by assuming a von Bertalanffy law formulation for the evolu-
tion of population. Within this setup, the model is shown to be described by
a three dimensional dynamical system with a unique non-trivial steady state
equilibrium (a saddle). In contrast to standard Ramsey model, the equilib-
rium paths of capital and consumption are proved to be non-monotonic over
time. Moreover, the stable manifold is two dimensional, so that as the system
approaches the stationary state, capital and consumption converge to their
steady state values with a speed of convergence which is determined by the
eigenvalue with the smaller absolute value.

References
[1] E. Accinelli and J.G. Brida, The Ramsey model with logistic population
growth, Economics Bulletin, 3 (2007), 1-8.

[2] E. Accinelli and J.G. Brida, The dynamics of the Ramsey economic growth
model with the von Bertalanffy population growth law, Applied Mathe-
matical Sciences, 1 (2007), 109-118.

[3] L. von Bertalanffy, A quantitative theory of organic growth (Inquiries on

growth laws. II), Human Biology, 10 (1938), 181-213.

[4] L.E. Blume and C.P. Simon, Mathematics for Economists, Norton, New
York, 1994.

[5] A. Bucci and L. Guerrini, Transitional dynamics in the Solow-Swan

growth model with AK technology and logistic population change, B.E.
Journal of Macroeconomics (Topics), 9 (2009), 1-16.

[6] M. Ferrara and L. Guerrini, On the dynamics of a three sector growth

model, International Review of Economics, 55 (2008), 275-283.

[7] M. Ferrara and L. Guerrini, The Ramsey model with logistic population
growth and Benthamite felicity function revisited, WSEAS Transactions
on Mathematics, 8 (2009), 41-50.
3238 L. Guerrini

[8] M. Ferrara and L. Guerrini, More on the Green Solow model with logistic
population change, WSEAS Transactions on Mathematics, 8 (2009), 97-
106.

[9] M. Ferrara and L. Guerrini, A note on the Uzawa-Lucas model with un-
skilled labor, Applied Sciences, 12 (2010), 90-95.

[10] C. Germanà, and L. Guerrini, On the closed-form solution of the improved

labor augmented Solow-Swan model, Applied Sciences, 7 (2005), 101-106.

[11] J. Guckenheimer and P. Holmes, Nonlinear Oscillations, Dynamical Sys-

tems, and Bifurcations of Vector Fields, Springer-Verlag, 1990.

[12] L. Guerrini, The Solow-Swan model with a bounded population growth

rate, Journal of Mathematical Economics, 42 (2006), 14-21.

[13] L. Guerrini, Logistic population change and the Mankiw-Romer-Weil

model, Applied Sciences, 12 (2010), 96-101.

[14] L. Guerrini, The Ramsey model with a bounded population growth rate,
Journal of Macroeconomics, 32 (2010), 872-878.

[15] L. Guerrini, A closed-form solution to the Ramsey model with logistic

population growth, Economic Modelling, 27 (2010), 1178-1182.

[16] L. Guerrini, The Ramsey model with AK technology and a bounded pop-
ulation growth rate, forthcoming Journal of Macroeconomics, (2010).

[17] L. Guerrini, Transitional dynamics in the Ramsey model with AK tech-

nology and logistic population change, Economics Letters, 109 (2010),
17-19.

[18] F.P. Ramsey, A mathematical theory of savings, Economic Journal, 38

(1928), 543-559.

[19] R.M. Solow, A contribution to the theory of economic growth, Quarterly

Journal of Economics, 70 (1956), 65-94.

Received: June, 2010