Question 1

The Solow model: Consider an economy with an aggregate production function

Yt = BKtα N 1−α , where 0 < α < 1, Yt and Kt are output and the capital stock
at time t, N is the number of workers. Assume N and B are constants and
this is a closed economy. Let s be the saving rate and d be the depreciation
rate of capital.

a. Use the Solow model to show that the long run growth rate of this econ-
omy is zero and derive the steady-state level of capital per worker. Explain
the economic meaning of the equations and the graph that you used in
your answer.

Answer:
Solow model assumes that economy saves a constant fraction of output and,
as it is a closed economy, this equals investment:

It = sYt = sBKtα N 1−α

The capital accumulation equation states that

Kt+1 = (1 − d)Kt + It

With a constant number of workers, the per worker capital accumulation equa-
tion is
kt+1 − kt = sBktα − dkt
From this equation, we argue with help of Figure 1 why a unique1 steady state
level of capital per worker k ∗ exists and why the economy must converge to it.
Imposing the steady state on the equation itself, we obtain
( )1/1−α
∗ sB
k =
d

1
Strictly speaking another steady state exists where k ∗ = 0 - but it is not stable (the economy
never converges to it. It is also not very interesting for our analysis, as it implies no activity in the
economy - so we ignore it.

1
 Figure 1: Solow Convergence
dk

sf (k)

klow k∗ khigh k

b. Suppose the economy is in the steady state of part (a). There is a sudden
arrival of migrant workers of number M once and for all. Each immigrant
carries with them q units of capital per person. Suppose the migrant
workers settle down in the economy and adopt the same saving rate as
the local workers. What will happen to the growth rate and level of
output per worker in the short run and in the long run? Explain with the
equations and graph you used in part (a).

Answer:

We need to study three cases: q = k ∗ , q > k ∗ , q < k ∗ and study movement

along the saving curve for these three different cases.
If q = k ∗ , capital per worker does not change, so steady state stays the same,
hence output per worker and growth rate does not change in the short or long
run.
If q > k ∗ , capital per worker and output per worker increases to kq>k∗ in
Figure 2) (jump off the steady state) in the short run but starts decreasing
(because the saving curve is below the depreciation curve) until the economy
returns to the initial steady state. So, output per worker increases in the short
run but does not change in the long run, and growth rate of output per worker
is negative in the short run and 0 in the long run.
If q < k ∗ , capital per worker and output per worker decreases to kq<k∗ in
Figure 2) (jump off the steady state) in the short run, but starts increasing
(because the saving curve is above the depreciation curve) until the economy
returns to the initial steady state. So, output per worker decreases in the short

2
 run but does not change in the long run, and growth rate of output per worker
is positive in the short run and 0 in the long run.

Figure 2: Convergence following inflow of migrant workers

dk

sf (k)

kq<k∗ kq=k∗ kq>k∗ k

Question 2
Output Y is produced by firms according to the production function Y =
K α (BN )1−α , where K is the capital stock, N is the workforce (equal to the
population), B is the level of labour-augmenting technology, and α is a pa-
rameter that lies between 0 and 1. The population N is constant over time
(n = 0). Capital depreciates at rate d, so the evolution of the capital stock is
K ′ − K = I − dK, where K ′ is next year’s capital stock and I is the level of
investment. Investment is equal to saving sY , which is a constant fraction s
of income.

Now assume there is a ‘learning-by-doing’ effect: new ideas (higher B) are

generated as a by-product of the production process, so an expansion of the
capital stock leads to the discovery of new ideas. Ideas are assumed to be a
public good (non-rival and non-excludable), hence suppose that the level of
technology B available to all firms depends on the economy-wide capital stock
K according to B = λK, where λ is a positive constant.
(a) Taking account of the learning-by-doing effect, derive the relationship between
y and k. Sketch the saving line sy and the depreciation line dk in a diagram.

3
 Answer:
Substituting the equation B = λK giving the learning-by-doing relationship
between knowledge B and aggregate capital K/ into the production function
Y = K α (BN )1−α :
Y = K α ((λK)N )1−α = K α λ1−α K 1−α N 1−α = λ1−α N 1−α K
Dividing both sides by N leads to y = λ1−α N 1−α k, which is linear in capital per
person k. This implies the saving line sy is now a straight line. The learning-
by-doing effect raises the aggregate return on capital to the point where the
marginal product of capital is no longer diminishing.
y

sy
dk

Since both the saving line and depreciation line are straight lines, they generally
do not intersect at any positive level of capital per person. This means the
economy does not have a steady state, and capital per person and income per
person can grow perpetually (endogenous long-run growth) if the saving line
is steeper than the depreciation line.
(b) Explain the effects of increasing the saving rate s. Will the higher saving rate
necessarily result in greater long-run consumption per person c = (1 − s)y than
would otherwise have been obtained?

Answer:
An increase in the saving rate shifts the saving line upwards. Since there is no
steady state, this means that the growth rate of capital per person and output
per person is permanently higher. As output per person y will diverge over
time ever further from the path it would otherwise have followed, eventually
c = (1 − s)y must become larger than it would otherwise have been because
there is only a one-off change to 1 − s.

4
 y

s2 y
s1 y
dk

It could be argued that even if there is a learning-by-doing effect, knowledge

might not increase exactly in proportion to the capital stock. Assume instead
that B = λK β , where β is a constant that lies between 0 and 1.

(c) How would this alternative assumption affect the results found in parts (a) and
(b)?

Answer:
Substituting the learning-by-doing equation B = λK β into the production
function:
(( ) )1−α
Y = K α λK β N
= K α λ1−α K (1−α)β N 1−α
= λ1−α N 1−α K α+β−αβ
= λ1−α N 1−α K 1−(1−α)(1−β)

Dividing both sides by N :

λ1−α N 1−α K 1−(1−α)(1−β)

y=
N (1−α)(1−β) N 1−(1−α)(1−β)
( )1−(1−α)(1−β)
1−α (1−α)(1−(1−β)) K
=λ N
N
= λ1−α N (1−α)β k 1−(1−α)(1−β)

This means sy is proportional to k 1−(1−α)(1−β) . Since α and β are both between

0 and 1, 1 − (1 − α)(1 − β) also lies strictly between 0 and 1, which means

5
that k 1−(1−α)(1−β) is an increasing and concave function of k. The saving line is
not a straight line unlike in part (a). Therefore, unlike part (a), it follows that
the economy would have a steady state for capital per person, which means
there is no endogenous long-run growth. Unlike part (b), an increase in the
saving rate does not necessarily raise long-run consumption per person because
the usual logic of the Golden rule applies. However, since k 1−(1−α)(1−β) is less
concave than k α , there is a greater range of saving rates for which higher saving
increases consumption in the long run.