1 Lecture Notes: The Solow Model

Questions:

1. Why large differences in growth rates?

2. Why persistent differences in productivity?
3. What drives overall world growth?

1.1 The Solow model: theory

• Technology: Firms produce a generi good Y with the production function
1−α
Y = F (K, AL) = K α (AL)

– L is the number of workers

– A is the number of ”efficiency units” per worker
– AL is the aggregate number of efficiency units. Assume implicitly
that all efficieny units are perfect substitutes)
– Note that F satisfies constant return to scale and the Inada conditions
– The generic good can be used for consumption and investment
– Assume constant growth in L and A:

Lt+1 = (1 + n) Lt
At+1 = (1 + g) At

– Law of motion for capital:

Kt+1 = (1 − δ) Kt + It

– Closed economy

• Preferences: Assume a labor supply of one unit per person and assume a
constant savings rate:
St = sYt

• Markets: Competitive markets for labor, capital, and the consumption/investment

good.

• Competitive equilibrium:

1
 1. Firm optimization: prices are marginal productivities,

max {F (K, AL) + (1 − δ) K − wL − (1 + r) K}

K,L
⇒
0 = F1 (K, AL) + 1 − δ − (1 + r)
0 = F2 (K, AL) − w
⇒
1−α Y
r+δ = αK α−1 (AL) =α
K
1−α Y
w = αK α (AL) /L = α
L

2. Individual optimization: CHEAT by assuming l = 1 and C = (1 − s) Y .

3. Market clearing
(a) Market for labor:
(b) Market for capital: Closed economy implies savings equal invest-
ments
St = It
(c) Goods market clearing requires

Yt = Ct + It

• Normalize capital and output per efficency unit

Yt
yt =
At Lt
Kt
kt =
At Lt
Note that
1−α α
K α (At Lt )

Kt
yt = t = = ktα ≡ f (kt )
At Lt At Lt

• Dynamics of the economy: Note: Kt+1 is a stock, while It and capital

growth ∆Kt+1 = Kt+1 − Kt are flows

∆Kt+1 = Kt+1 − Kt = (1 − δ) Kt + It − Kt
= sYt − δKt

2
 In efficiency units:

∆kt+1 = kt+1 − kt
Kt+1 Kt
= −
At+1 Lt+1 At Lt
Kt+1 Kt At+1 Lt+1
= −
At+1 Lt+1 At+1 Lt+1 At Lt
Kt+1 Kt
= − (1 + g) (1 + n)
At+1 Lt+1 At+1 Lt+1
Kt+1 − Kt − Kt [(1 + g) (1 + n) − 1]
=
At+1 Lt+1
sYt − δKt − Kt [(1 + g) (1 + n) − 1]
=
At+1 Lt+1
sYt − Kt [g + n + δ + gn]
=
At+1 Lt+1

At+1 Lt+1 sYt − Kt [g + n + δ + gn]

∆kt+1 =
At Lt At Lt
∆kt+1 (1 + g) (1 + n) ≈ sf (kt ) − [g + n + δ] kt
| {z } | {z }
gross inv.
replacement needed
to keep capital constant

• Definition: a steady state (with technical progress and population growth)

is an equilibrium path in which

∆kt+1 = 0.

• Solve for steady state k ∗ :

0 = sf (k ∗ ) − [g + n + δ + gn] k ∗
⇒
α
f (k ∗ ) (k ∗ ) g + n + δ + gn
= =
k∗ k∗ s
 1
 1−α
s
k∗ =
g + n + δ + gn

• Consider effect on k ∗ from changes in savings rate s and in population

growth n. GRAPH (y, k graph and time-series)

3
 • Golden rule: solve for the maximum steady-state consumption c∗∗ :
C∗ = Y ∗ − I∗
I∗ = (g + n + δ + gn) ALk ∗
Y ∗ − I∗ Y ∗ − (g + n + δ + gn) ALk ∗
c∗ = =
AL AL
= f (k ∗ ) − (g + n + δ + gn) k ∗
≈ f (k ∗ ) − (g + n + δ) k ∗

∂c∗
= f 0 (k ∗∗ ) − (g + n + δ) = 0
∂k ∗
⇒
0 ∗∗
f (k ) = g+n+δ
Recall that, from firm’s optimization,
rt + δ = f 0 (kt )
so requirement for golden rule is
r∗∗ = n + g,
i.e., all return to capital is reinvested (zero dividends consumed by capi-
talists).

1.2 Empirical performance of the Solow model

• Want to evaluate the Solow model empirically. Focus on dynamics (i.e.,
speed of growth).
• ”Neoclassical hypothesis”: all differences in growth and GDP per capita
are due to differences in capital
• Problem: difficult to measure capital. Solution: derive implied growth in
Yt without need to measure Kt :
1−α
Y = K α (AL)
⇒
1 − 1−α
K = Y α (AL) α

Kt+1 = sYt + (1 − δ) Kt
⇒
1 1
− 1−α − 1−α
Yt+1
α
(At+1 Lt+1 ) α = sYt + (1 − δ) Yt α (At Lt ) α
  α1     α1
Yt+1 − 1−α Yt Yt − 1−α
(At+1 ) α Lt+1 = s Lt + (1 − δ) (At ) α Lt
Lt+1 Lt Lt
 − 1−α  1     α1
At+1 α
Lt+1 Yt+1 α 1−α Yt Yt
= (At ) α
s + (1 − δ)
At Lt Lt+1 Lt Lt

4
 which boils down to
− 1−α 1 1−α 1
(1 + n) (1 + g) α
(ỹt+1 ) α = (At ) α
sỹt + (1 − δ) (ỹt ) α (1)

where ỹ is GDP per capita:

Yt
ỹt =
Lt

1. Assume that all countries have the same δ = 5% and g = 2.5% (same
as the US, 1960-2000). Note: 1 − α is labor’s share of output: from
firm’s optimization we have
Y
w = (1 − α)
L
⇒
wL
1−α = ,
Y
which is 2/3 for the US ⇒ α = 1/3.
US
2. Assume that the US is in steady state in 1960. Measure Y1960 and
US US US
L1960 . Pin down K1960 : and A1960
j j
3. Measure ỹ1960 = Y1960 /Lj1960 for a country j.
4. Assume that Aj1960 = AU S
1960 . Use equation (1) to project future values
j
of Yt (note: for the US it is simply
t−1960
ỹtU S = (1 + g) US
· ỹ1960

5. Observation: Solow model implies too fast convergence. So a large

amount of the differences in output across countries must be driven
by differences in Ajt .

1.3 Conclusion
1. Empirical standpoint: Solow model fails to explain in a satisfactory way
the great disparities in output levels and growth rates

2. Theoretical standpoint: Solow model (or simple extensions of it) cannot

explain the growth in At , which is the main drive of growth

Need ”new growth theory” to explain why Ajt does or does not grow.