ECON0028

Economics of Growth
LECTURE 2: CAPITAL ACCUMULATION AND THE SOLOW
GROWTH MODEL

Luk sz R chel, October 2024

1
a
a
 Overview
• Objective: understand and appreciate the concept of capital and its role in
economic growth
• We will do that through the lens of the Solow growth model
• Theory that is centred on capital accumulation
• A useful framework to explain di erences in per-worker income

2
ff
 Plan for today

• What is capital?
• An aside on modelling in (macro)economics
• Solow model
• setup
• graphical solution and comparative statics
• analytical solution for the steady state
• convergence
• introducing sustained growth

3
 What is capital?

• Machines in factories
• Buildings in which we work
• Infrastructure (roads, bridges, broadband cables)
• Vehicles
• Computers
• …

4
5
6
GDP per worker and capital per worker

7
 Nature of capital

1. Capital is productive: enables workers to produce more output

2. Capital has to be produced/accumulated through investment: requires sacri ice
of consumption today
3. Capital is rival in its use: one factory can use a given machine at any given time
4. In a market economy, capital earns a return
5. Capital depreciates with usage and due to passage of time

f
Before we get to our irst
model…

f
General Comment: modelling in (macro)economics
• Objective is not to build one big model we use to address all issues
• descriptive realism is not the objective
• instead make modelling choices that are dependent on the issue
• whether a model is “good” is context dependent
• Approach to modelling in macro(economics) well summarised by following
statements:
• “All models are false; some are useful” https://en.wikipedia.org/wiki/
All_models_are_wrong
• “If you want a model of the real world, look out the window” (kidding, but only
half kidding)
• Point in case: geographically accurate London tube map

10
11
Crucial or critical assumptions

12
 13
Source: Rodrik (2015) “Economics Rules”
The Basic Solow Model

14
 Assumptions of the simplest version of the Solow model

• Competitive markets (implying w = MPL and r = MPK — you should know why)
• Assume population constant over time - we will relax this
• Assume technology constant over time - we will also relax this
• In a few slides we will allow for a constant growth rates of population and
technology (both will be exogenous)
• Why is this useful? Because we can study how changes in population or technology growth
rates a ect the economy (through the lens of the model).

• In the baseline model: all the action comes from K (the key endogenous variable)
• depreciation
• investment
• Important that you understand the concepts of exogenous and endogenous variables!
• E.g. it must be clear that whether a particular variable 15
is exogenous or endogenous depends on the speci ic model.
ff
f
 Built around two equations

• Production function
• Capital accumulation equation

16
 Cobb-Douglas Production Function: a review
• Recall that an often used aggregate production function is of the form:
α 1−α
Y = AK L
for α ∈ (0,1). A is the technology term; K is capital; L is labour.
• Output per worker is:
Y α −α α
y = = AK L = Ak ≡ f(k)
L
• We also have:
α−1 α−2
MPK ≡ fk = αAk > 0 fkk = α(α − 1)k <0

MPK ⋅ K
capital share of income = =α
Y
why? what is the labour share? 17
 Capital accumulation equation
• Capital changes over time b/c of new investment and b/c existing stock depreciates:
ΔK = investment − depreciation

• Investment is how much of output every period we put towards capital (instead of
consuming it right away). In the Solow model:
investment = s ⋅ Y
where s is a parameter (a constant fraction of income is saved and thus put towards
investment).

• Note: Weil calls this parameter γ. Obviously it doesn’t make any di erence what you call
it.
• Depreciation: assume that a constant fraction of capital depreciates every period:
depreciation = δ ⋅ K
18

ff
 Recall the continuous time notation
• The previous slide and Weil’s book use ΔK ≡ Kt − Kt−1
• This is the discrete time notation
· dK
• In continuous time the capital accumulation equation becomes ΔK = K ≡
dt
= sY − δK
· ·
K k
• With constant L we have
K k
= (check this — take logs, then di erentiate K/L). So the capital accumulation
equation becomes
·
k = sy − δk

• Substituting in the production function we have condensed the model to one equation only:
·
k = sf(k) − δk
• This equation tells us how capital per worker changes over time.
• Mathematically this is a (non-linear) di erential equation (why is it non-linear?)
• We will analyse this equation graphically, by plotting sf(k) and δk
19
ff
ff
·
k = sf(k) − δk

20
 A slightly more general
Solow Model
(with population growth)

21
 Adding population growth

• Suppose that population of workers grows at a constant rate n. Easy to

accommodate in the model.
• But irst, recall 2 alternative ways of writing this:
Lt − Lt−1 ·
L
=n =n
Lt−1 L

• An aside: what is a reasonable value for n for the world as a whole? For the US /
Europe?

22
f
Population growth rate

23
 Adding population growth
K
• Consider the capital labor ratio k ≡ . What is its growth rate?
L
• Take logs and di erentiate:
log k = log K − log L
· ·
K k
= +n 🧠
K k
• Our main equation is:
·
K = sY − δK divide by K:
·
K sY
= − δ use equation 🧠:
K K
·
k sY
+n= − δ multiply by k:
k K
·
k = sf(k) − (δ + n)k 24
ff
Solow diagram with population growth

25
 Comparative statics
• Comparative statics tells us what the model implies about the behaviour of endogenous variables
when we change an exogenous variable / parameter
• although you could also study exogenous changes to endogenous variables: e.g. an earthquake wipes out half of k

• For example: what happens when there is an increase in the saving rate?

26
 Comparative statics II
• What else could we analyze? [make sure to study these at home! 😇]

• Changes in the depreciation rate δ

• Changes in the population growth rate n

• Changes in the capital share α

• Changes in technology A

• Exogenous shifts to the endogenous variable k

• Important: what real world phenomena could these shifts represent?

• Note that we can study once and for all changes, but also temporary shifts
• Example: suppose the govt raises saving through some policy (e.g. tax-free saving
programme), then loses power and the new government reverses this policy
• s is higher for 5 years, but then goes back down. Let’s trace out the e ects on the board.
27

ff
Steady state and the
transition towards it

28
 Steady state

• The level of capital per worker towards which the economy converges: k*
• Weil calls it k ss
- obviously it doesn’t matter what you call it.

• k* is the steady state level of capital per worker. We also have y*, c*
• Steady state: if you start there, you stay there
• Interesting questions about convergence towards steady state:
1. Does the economy converge to k* from anywhere (i.e. any k0)?
2. Does it converge monotonically?
3. How fast does it converge?
• Let’s irst solve for the steady state, and then answer questions 1-3.

29
f
 Solving for the steady state
• Solving ≡ expressing the endogenous variables in terms of exogenous variables and
parameters
·
• Recall the key equation (in per capita terms) is k = sf(k) − (δ + n)k.
·
• Steady state k: k is zero: 0 = sf(k*) − (δ + n)k*.

• This is just a simple non-linear equation. Assuming Cobb-Douglas, solve for k*:
α
0 = sA(k*) − (δ + n)k*
1

(n + δ)
sA 1−α
k* =

• Now can calculate st st output per worker and consumption per worker:
α

(n + δ)
α sA 1−α
y* = A(k*) = A c* = (1 − s) ⋅ y*

• It is vitally important you can solve for the steady state and interpret the results! 30
 Why are some countries rich and others poor?
Let’s use the solution to le rn…
α

(n + δ)
sA 1−α
y* = A

Higher steady state (“long-run”) income per worker if:

• higher A
• higher s
• lower n
• lower δ
• higher α

why? [note that all the answers are ultimately about k*]
Moreover, the model allows us to provide quantitative answers to interesting questions.
Example: assume α = 1/3 and 2 countries identical except sA = 2sB. What is the di erence in income per capita in st st?
yA α
= 2 1 − α = 2 ≈ 1.4
yB 31
a
ff
Testing the predictions

32
Important point: growth in the steady state
How f st does income per c pit grow in the ste dy st te?

• Answer based on the model so far: 0%

• by the way, what is GDP growth?

• There is no sustainable growth in income per capita in the Solow model

• Interpretation?
• Capital accumulation alone cannot drive long-run growth in the standard of
living.
• Why?
• Because capital runs into diminishing returns.

33
a
a
a
a
a
 Transition dynamics
Wh t if we st rt below or bove k*?

• Normalize A = 1. Note that

·
k α−1
= sk − (δ + n)
k
This equation characterizes the growth
rate of capital per worker.
• Again, we can analyze it graphically.
• Clearly, capital (and hence the
economy!) grows/declines faster the
further away we are from the steady
state.
• What is the intuition? 34
a
a
a
Answering the 3 questions on convergence

1. Does the economy converge to k* from anywhere (i.e. any k0)?

2. Does it converge monotonically?
3. How fast does it converge?

35
Answering the 3 questions on convergence

1. Does the economy converge to k* from anywhere (i.e. any k0)?

2. Does it converge monotonically?
3. How fast does it converge?

A1: from any k0 >0

A2: yes
A3: it depends on how close / far away from the steady state it is

36
 Conditional convergence

• The previous slides imply that there should be convergence among countries which have the
same steady state.
• Countries that are further away from the steady state should grow faster than the ones that
are close to the steady state.
• Three interesting predictions of the model:
• if two countries have the same parameters but di erent levels of income, the country with
lower income grows …………
• if two countries have the same level of income but di erent rates of investment, then the
country with a higher rate of investment grows …………
• a country that raises its level of investment will experience an …………….. in its rate of per
capita income growth.

37
ff
ff
 The Solow Model with
Exogenous Technology
Growth

38
 Solow model with technology growth
·
α 1−α A Y 1−α α
• Suppose Y = K (AL) and now = g. What happens to y = = A k ?
A L
• First, we can take logs and di erentiate:
·y · ·
A k
• = (1 − α) + α 🎃
y A k
• This suggests that output per worker is growing as long as g
· > 0. But at what
k
rate will it grow? We need to igure out in this model.
k
• If we treat AL as “e ective workforce”, we can proceed pretty much as before…

• Our strategy for next 3 slides: show that K/AL is constant in the steady state.
·
From that deduce the growth rate of k. Using equation 🎃, calculate y/y.
39
ff
ff
f
 Analysis is identical to before
K
• Consider the capital to e ective labor ratio k̃ ≡
AL
. What is its growth rate?

• Take logs and di erentiate:

log k̃ = log K − log L − log A

· ·
K k̃
= +n+g 🦾
K k̃

• We can again re-express our main equation:

·
K = sY − δK divide by K:
·
K sY
= − δ use equation 🦾:
K K
·
k̃ sY
+n+g= − δ multiply by k:
k̃ K
· α
k̃ = sf(k̃) − (δ + n + g)k̃ check that f(k̃) = k̃ so:
·
k̃ = sk̃α − (δ + n + g)k̃ 40
ff
ff
Solow diagram with technological progress
KEY: the model converges to const nt k̃*
· α
k̃ = sk̃ − (δ + n + g)k̃

41
a
a
 Balanced growth path
·
k̃
• So in the long-run = 0. Using this and equations 🦾 and 🧠 we ind that the growth rate of
k̃
capital per worker is simply
· Recall:
k ·
=g ·
K k̃
k = +n+g 🦾
K k̃
• Therefore the growth rate of per capita GDP is (from equation 🎃): ·
k K
·
·y · · = −n 🧠
A k k K
= (1 − α) + α = g
y A k
• Thus, output, capital and consumption per worker all grow at a constant rate g.
• This is called a balanced growth path: it’s basically like a growing steady state.
• Thus, we have a model of a growing economy, even in the long run! 👻
• But: exogenous technological change drives all of the long-run growth in this model.
42

f
 Main takeaways

• We have built and solved our irst model!

• Key feature: the capital accumulation equation.
• Useful laboratory for studying:
• income di erences across countries, and
• reaction of the economy to shocks
• Main insight: capital accumulation cannot drive sustainable growth because of
diminishing returns.

43
ff
f
 Next week…

• More on the Solow model

• Use the production function to perform “growth accounting”
• Human capital

44