Neoclassical Growth Model

Vibhor Verma1

1
Delhi Technological University
What is neoclassical?

The original neoclassical economists (Marshall, Jevons,

Walras, Wicksell, Pareto, Clark, Edgeworth, and Fisher) of the
late 19th century, used marginal utility and productivity to
analyze pricing in competitive markets. They explored the
possibility of a price system that ensures supply and demand
equality across all markets, emphasizing rational, maximizing
approaches and concepts like marginal productivity of wages
and perfect competition.
The description of theories, which while not necessarily
denying Keynesian strictures, ignore what are called Keynesian
’difficulties’ by assuming the existence of a government which
persistently, continually and successfully manipulates the
policy instruments at its disposal so as to maitain the
full-employment level of aggregate demand (Samuelson’s
neoclassical synthesis).
The original ’classical’ economists of the first half of the
nineteenth century were much more concerned with the long
run forces that governed the macroeconomy than with the
behaviour of individual markets - which was the attention of
the nineteenth century classicals. ’Neoclassical’ then could
mean the subordination of short-term problems to long-run
trends in modern neoclassical economic theory.
Assumptions:
Solow’s paper is devoted to a model of long-run economic growth
which accepts all the Harrod-Domar assumptions except that of
fixed proportions.
A single all-purpose commodity, whose rate of production is
designated Y (t), is produced and either consumed or
invested. No Keynesian distinction is drawn between those
who save and those who invest - saving is investment - no
separate investement function needs to be included in the
model. In terms of a ’corn’ (Y (t)) parable, any ’corn’ not
eaten is saved (S(t) = I (t)) and automatically becomes part
of the stock of ’corn’ capital (K (t)).
A simple proportional saving function is assumed.

S = sY

where 0 < s < 1.

The labour force grows at an exogenous constant proportional
rate n, i.e.,
L̇
=n
L
The technical possibilities of the economy are represented by a
continuous, constant returns to scale, aggregate production
function:

Y = F (K , L)
y = f (k)

where y = Y /L and k = K /L. Each input is essential in the

sense that

F (K , 0) = 0
F (0, L) = 0

The assumption of a continuous aggregate production

function is fundamentally different to the fixed-coefficients
form used by Harrod and Domar.
We will, moreover, assume that the aggregate production
function satisfies the following conditions:

f ′ (k) > 0
f ′′ (k) < 0
lim f ′ (k) = 0
k→∞
lim f ′ (k) = ∞
k→0

The last two conditions are referred to as Inada conditions.

The mechanism whereby most Keynesian difficulties have
been shunted aside is the necessary identitiy of ex-ante saving
and ex-ante investment with the resulting ’absence of an
investment function and the consequent failure to assign a
major role to entrepreneurial expectations about the future’.
The factor markets work perfectly - the wage rate and the
rate of profit adjust smoothly and instantaneously to changing
circumstances.
The Model

The Fundamental Equation of the neoclassical growth model

Y ≡C +I
Y C I
=⇒ (t) ≡ (t) + (t)
L L L
C I
=⇒ f (k(t)) = (t) + (t)
L L
(1 − s)Y I
=⇒ f (k(t)) = (t) + (t)
L L
I
=⇒ f (k(t)) = (1 − s)f (k(t)) + (t)
L
I
=⇒ (t) = sf (k(t))
L
Note that
K
k=
L
=⇒ lnk = lnK − lnL
1 dk 1 dK 1 dL
=⇒ = −
k dt K dt L dt
k̇ K̇ L̇
=⇒ = −
k K L
k̇ K̇
=⇒ = −n
k K
 Using

K̇ = I − δK

we get 2

Y (t) K (t)
=⇒ k̇ = s −δ − nk
L(t) L(t)
=⇒ k̇ = sf (k) − δk − nk
=⇒ k̇ = sf (k) − (n + δ)k

In steady state or on the balanced growth path, k̇ = 0, and

hence,

K̇
=n
K
sf (k ∗ ) = (n + δ)k ∗

where k ∗ is the steady-state per capita capital.

2 k̇ K̇ K̇ I −δK
k
= K
− n =⇒ k̇ = L
− nk = L
− nk = s YL − δ KL − nk
Phase Diagram
Note that

k̇ = sf (k) − (n + δ)k
d k̇
=⇒ = sf ′ (k) − (n + δ)
dk
Since f ”(k) < 0, and

lim f ′ (k) = ∞
k→0
lim f ′ (k) = 0
k→∞

The slope of actual investment, sf ′ (k) starts from a very high

value relative to n + δ, slope of the break-even investment;
decreases to be equal to n + δ and then goes below it. Therefore,
k̇ increases (at a decreasing rate), reaches a maximum and then
decreases. At k̇ = 0, k = k ∗ . Note that the phase diagram is the
graphical representation of the fundamental equation of the Solow
growth model, i.e., k̇(k) = sf (k) − (n + δ)k.
Transitional Dynamics

The long run growth rates in the Solow model are determined by
exogenous factors - in the steady state, the per capita quantities k,
y and c do not grow and the aggregate variables K , Y and C grow
at the rate of population growth n.3

k̇ = sf (k) − (n + δ)k
k̇ sf (k)
=⇒ γk ≡ = − (n + δ)
k k

3
since K , L are growing at the rate n in steady state, due to constant
returns to scale, Y is also growing at the same rate, n, in steady state.
Note that
d f (k) f ′ (k)k − f (k)
( )= 2
dk k  k 
f (k) ′
=− − f (k) /k
k

The term inside the bracket is marginal product of labour divided

by k and hence is positive.4 Therefore, the derivative of f (k)/k
with respect to k is negative. Hence, sf (k)/k is downward sloping.

4
Y = Lf (k) =⇒ ∂Y
∂L
= f ( KL ) + L.f ′ ( KL )( −K
L2
) = f (k) − kf ′ (k)
Since
sf (k) ∞
lim =” ”
k→∞ k ∞
sf (k) 0
lim =” ”
k→0 k 0

using the l’Hôpital’s rule5 and the Inada conditions,

sf (k)
lim = lim sf ′ (k) = 0
k→∞ k k→∞
sf (k)
lim = lim sf ′ (k) = ∞
k→0 k k→0

Therefore, the curve sf (k)/k falls monotonically from infinity to

zero. Meanwhile, (n + δ) is a horizontal line. Since sf (k)/k is
monotonic, it intersects the horizontal line (n + δ) once and only
once for k > 0, which is at the steady state per capita capital k ∗ .

f (x) f (x) f ′ (x)

5
If limx→a g (x)
= ”∞
∞
” or ” 00 ”, then limx→a g (x)
= limx→a g ′ (x)
Note that for k < k ∗ , k̇
k > 0, therefore k increases until it settles
at k ∗ . For k > k ∗ , k̇
k < 0, therefore k decreases until it settles at
k ∗. At k ∗ , k̇k = 0. Therefore, in the steady state, k, y and c do
not change.
The reason for declining rates of growth is diminishing returns to
capital. When k is relatively low, the average product of capital,
f (k)/k is relatively high, and hence, the gross investment per unit
of capital sf (k)/k is relatively high. Since, depreciation happens at
the constant rate n + δ, the difference between sf (k)/k and n + δ,
that is, k̇k , is relatively high.
Analogously, if the economy starts at k(0) > k ∗ , the growth rate
of k is negative, and k falls over time until it reaches k ∗ .
The growth rate of output per capita is given by

ẏ k̇ kf ′ (k) k̇
= f ′ (k) =
y f (k) f (k) k
′
The expression kff (k)
(k)
is called the capital share, i.e., the share of
rental income in capital in the total income. We denote it by αK .
Thus, the behaviour of ẏ /y mimics that of k̇/k. More specifically,

ẏ kf ′ (k) k̇
=
y f (k) k
kf ′ (k) [sf (k) − (n + δ)k]
=
f (k) k
′
= sf (k) − (n + δ)αK
 (n + δ)f ′ (k)
 
∂(ẏ /y ) f ”(k).k k̇
= − [1 − αK ]
∂k f (k) k f (k)

If k̇k ≥ 0, ∂(∂k
ẏ /y )
< 0, i.e., if k ≤ k ∗ , ẏ /y necessarily falls as k rises
(and as y rises).
If k̇/k < 0(k > k ∗ ), the sign of ∂(ẏ /y )/∂k is ambiguous for a
general form of the production function, f (k). However, if the
economy is close to its steady state, the magnitude of k̇/k will be
small, and ∂(ẏ /y )/∂k < 0 will surely hold even if k > k ∗ .
Convergence

 
∂(k̇/k) ′ f (k)
= s f (k) − /k < 0
∂k k

Smaller values of marginal of k are associated with larger k̇/k.

Consider a group of economies that have the same k ∗ but different
initial quantities of capital per person k(0). Less advanced
economies with lower values of k(0) and y (0) have higher rates of
growth of k and in the typical case, also higher growth rates of y
(i.e., for k ≤ k ∗ or for small k above k ∗ ).
Countries with lower starting values of capital-labour ratio have
higher per capita growth rates k̇/k and tend thereby, to catch up
or converge to those with higher capital-labour ratio. Poorer
countries tend to grow faster than rich countries. This is referred
to as absolute convergence.
Another type of convergence is conditional convergence where an
economy grows faster the further it is from its own steady state.
 
k̇ f (k)/k
= (n + δ) −1
k f (k ∗ )/k ∗

Just like before, assume k(0)poor < k(0)rich but with different
saving rates, spoor < srich . The rich economy is growing faster if
∗
k(0)rich is further away from krich ∗
than k(0)poor is from kpoor . Low
saving rate of the poor economy offsets its higher average product
of capital as a determinant of economic growth. Hence, the poor
economy may grow at a slower rate than the richer one.6

6
The proof of the equation above is left as an exercise.
Effects of increase in rate of saving

Consider a permanent increase in the rate of saving, from sOLD to

sNEW at time t0 with the respective steady state capital-labour
∗
ratios being kOLD ∗
< kNEW . At the old steady state, the permanent
increase in the rate of saving leads to a reduction in per capita
consumption. Consumption c starts to increase as the economy
moves towards the new steady state. Whether per capita
consumption c in the new steady state in more or less than that in
the old steady state is ambiguous.
Note that, per capita output y is not increasing in the old steady
state, so that ẏ /y = 0. When there is an increase in the saving
rate, k increases as the economy converges to a new steady state.7
∗
Therefore, ẏ /y > 0. But since k < kNEW , ẏ /y is decreasing as k
∗
increases (at a decreasing rate) towards kNEW ∗
and at kNEW is
∗
again 0. Hence, y = yNEW on the new balanced growth path.

7
Since ẏ
y
= f ′ (k) k̇k and in steady state k̇
k
=0
Effects of increase in rate of saving
Effects of increase in rate of saving
Effects of increase in rate of saving
The steady state level of per capita consumption is

c ∗ = (1 − s)f (k ∗ )

Since sf (k ∗ ) = (n + δ)k ∗ ,

c ∗ = f (k ∗ (s, n, δ)) − (n + δ)k ∗ (s, n, δ)

∂c ∗ ∂f (k ∗ ) ∂k ∗ ∂k ∗
= − (n + δ)
∂s ∂k ∗ ∂s ∂s
The per-capita consumption is maximized at the rate of
saving s where,

∂f (k ∗ (sGOLD ))
=n+δ
∂k ∗ (sGOLD )
f ′ (kGOLD ) = n + δ
The saving rate at which per capita consumption is maximized is
called the golden rule saving rate and is denoted by sGOLD .
Effects of increase in rate of saving
The Golden Rule and Dynamic Inefficiency
The figure above considers three possible rates of saving,
s1 < sGOLD < s2 . The corresponding steady state levels of capital
are k1∗ < kGOLD
∗ < k2∗ .
Consider the economy at s2 > sGOLD , so that c2∗ < cGOLD .
Suppose starting from the steady state at s2 , the saving rate is
permanently reduced to sGOLD . The per capita consumption c
initially increases by a discrete amount and then monotonically
falls in the transition towards its new steady state value cGOLD .
Since c2∗ < cGOLD , we conclude that c exceeds its previous value
c2∗ at all transitional dates, as well as in the new steady state.
Hence when s > sGOLD , the economy is oversaving in the sense
that per capita consumption at all points in time could be raised
by lowering the saving rate. An economy that oversaves is said to
be dynamically inefficient, because the path of consumption lies
below feasible alternative paths at all points in time.
Similarly, for s < sGOLD , the economy is undersaving and hence
dynamically inefficient.
Technical Progress

In our discussion, technical progress is assumed to be factor

augmenting. Technical progress shifts the production function
such that more output is produced even though the stock of
capital and the labour force may not have increased. It is as if
the factors of production have been augmented.

Y = F (A(t)K , B(t)L)

where A(t)K and B(t)L are referred to as effective capital

and effective labour respectively.
If Ȧ(t) > 0, B(t) = 1, then technical progress is said to be
purely capital-augmenting.
If A(t) = 1, Ḃ(t) > 0, then technical progress is said to be
purely labour-augmenting.
Assume that technical progress proceeds at a constant,
exogenous rate m.

Ȧ(t)
Y = F (A(t)K , L); =m
A(t)
implies purely capital-augmenting technical progress at the
constant proportional rate m.

Ḃ(t)
Y = F (K , B(t)L); =m
B(t)
implies purely labour-augmenting technical progress at the
constant proportional rate m.
and
Ȧ(t) Ḃ(t)
Y = F (A(t)K , B(t)L); = = m,
A(t) B(t)
implies an equally capital and labour augmenting technical
progress at the constant proportional rate m.
Dynamics of the Solow model with technical progress

We take technical progress as labour-augmenting(Harrod

neutral) proceeding at an exogenous saving rate of g .

Y (t) = F (K (t), A(t)L(t))

Ȧ(t) = gA(t)

Due to constant returns to scale,

Y (t) K (t)
= F( , 1)
A(t)L(t) A(t)L(t)
Y (t) K (t)
=⇒ y = f (k); y = ,k =
A(t)L(t) A(t)L(t)

where A(t)L(t) is effective labour, y is output per unit of

effective labour and k is capital per unit of effective labour.
The dynamics of k

K (t)
k(t) =
A(t)L(t)
K̇ (t)A(t)L(t) K (t)
k̇(t) = − (A(t)L̇(t) + Ȧ(t)L(t))
(A(t)L(t))2 (A(t)L(t))2
K̇ (t) ˙
K (t) L(t) Ȧ(t) K (t)
= − −
A(t)L(t) A(t)L(t) L(t) A(t) A(t)L(t)
sY (t) − δK (t) K (t) K (t)
= −n −g
A(t)L(t) A(t)L(t) A(t)L(t)
= sf (k(t)) − (n + δ + g )k(t)
In steady state, k̇/k = 0. At the steady state k = k ∗ ,

sf (k ∗ ) = (n + g + δ)k ∗
ẏ
=0
y
K̇ Ẏ
= =n+g
K Y
d
dt [Y (t)/L(t)]
=g
Y (t)/L(t)

Proofs of these results are left as an exercise.

Transitional Dynamics

k̇ = sf (k) − (n + δ + g )k
k̇ sf (k)
=⇒ γk ≡ = − (n + δ + g )
k k
Note that for conditional convergence,
 
k̇ f (k)/k
= (n + g + δ) − 1
k f (k ∗ )/k ∗
K
where k = AL .
Effects of increase in rate of saving
Effects of increase in rate of saving
Effects of increase in rate of saving
A change in the rate of saving has level effects but no growth
effects, i.e., per capital output grows at the same rate as before
from a higher level in the new steady state. Note that
consumption per unit of effective labour in steady state is
maximized at the rate of saving sGOLD where,

∂f (k ∗ (sGOLD ))
=n+g +δ
∂k ∗ (sGOLD )
f ′ (kGOLD ) = n + g + δ
The long run effects of increase in rate of saving on steady state
output per unit of effective labour, y ∗ = f (k ∗ ), are given by

∂y ∗ ∂k ∗ (s, n, g , δ)
= f ′ (k ∗ )
∂s ∂s
Since sf (k ∗ (s, g , n, δ)) = (n + g + δ)k ∗ (s, n, g , δ)

∂k ∗ ∂k ∗
sf ′ (k ∗ ) + f (k ∗ ) = (n + g + δ)
∂s ∂s
∂k ∗ f (k ∗ )
=⇒ =
∂s (n + g + δ) − sf ′ (k ∗ )
∂y ∗ f ′ (k ∗ )f (k ∗ )
=⇒ =
∂s (n + g + δ) − sf ′ (k ∗ )
s ∂y ∗ s f ′ (k ∗ )f (k ∗ )
=⇒ ∗ =
y ∂s f (k ∗ ) (n + g + δ) − sf ′ (k ∗ )
 (n+g +δ)k ∗
By substituting s = f (k ∗ ) in steady state,

s ∂y ∗ k ∗ f ′ (k ∗ )/f (k ∗ )
=
y ∗ ∂s [1 − k ∗ f ′ (k ∗ )/f (k ∗ )]
αK (k ∗ )
=
[1 − αK (k ∗ )]

where αK (k ∗ ) = k ∗ f ′ (k ∗ )/f (k ∗ ) is the capital share.8

8
If markets are competitive and there are no externalities, capital earns its
marginal product. Since Y = ALf (k), ∂Y ∂K
= ALf ′ (k) AL
1
= f ′ (k). Thus, if
capital earns its marginal product, the share of total income earned by capital
(per unit of effective labour) on the balanced growth path is k ∗ f ′ (k ∗ )/f (k ∗ ) or
αK (k ∗ ). Intuitively, a small value of αK (k ∗ ) makes the impact of saving on
output low for two reasons. First, it implies that the actual investment curve,
sf (k) bends fairly sharply. As a result, an upward shift of the curve moves its
intersection with the break-even investment line relatively little. Secondly, a low
value of αK (k ∗ ) means that the impact of change in k ∗ on y ∗ is small.
Bibliography

Barro, R.J., Sala-i-Martin, X Economic Growth, 2nd Edition

Jones, H.W., An Introduction To Modern Theories of
Economic Growth
Mankiw, G., Macroeconomics, 9th Edition
Romer, D., Advanced Macroeconomics, 4th Edition