Innovation, Economic Growth and
development
Merit course – 2006
Stylized facts of economic growth (Kuznets)
Two visions of economic growth
Evolutionary growth theory

Simon Kuznets: Modern Economic
Growth
High rate of growth of GDP per capita
– Relative to previous periods
– Relative to non-developed countries
100,000
10,000
1,000
100
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
World US UK

Simon Kuznets: Modern Economic
Growth
High rate of productivity growth (not just
growth of GDP)

Simon Kuznets: Modern Economic
Growth
High rate of structural change
– agriculture -> industry -> services
– Small enterprises -> large enterprises (managerial
firm)
Japan United States
United Kingdom
100 100
100
90 90
90
80 80
80
70 70
70
60 60
60
50 50
50
40 40 40
30 30
30
20 20 20
10 10
10
0 0 0
1840 1860 1880 1900 1920 1940 1960 1980 2000 1840 1860 1880 1900 1920 1940 1960 1980 2000 1840 1860 1880 1900 1920 1940 1960 1980 2000

Kuznets – MEG (continued)
Change in structure of society (secularization,
urbanization)
Developed countries reach out to the rest of
the world (“globalization”)
Inequality between countries

Economic growth & economic theory
Classical economists (18th and 19th century):
Smith, Ricardo, Marx
Schumpeter: the role of innovation
Postwar: neo-classicals, post-Keynesians
Modern: evolutionary and endogenous
growth

Modern theory
Evolutionary theory emerges as an attempt to
endogenize technology in economic theory
Endogenous growth theory is the (later) neo-
classical attempt to do the same

Technology in evolutionary theory
Strong uncertainty (vs. risk)
Who copes with uncertainty: homo
economicus or evolution?
The metaphor of the Blind Watchmaker in
economics

The Blind Watchmaker (Dawkins)
Uses (random) trial and error
Does not optimize, but adapt
May realize completely different development
paths, if “the tape were played twice”

Two world views
Economic growth as an equilibrium process:
smooth growth patterns
Economic growth as a dis-equilibrium
process: Transformation (in the underlying
structure) and changes of rhythm

Deterministic and reversible time
In a deterministic system, if you know the
laws of nature, and the initial state, you can
predict the future perfectly
– For example, Newtonian mechanics
Even with “weak randomness”, you may have
a system that is essentially deterministic
In economic growth theory, the steady state
plays an important role
– Key variables in the economy grow at a fixed and
constant rate

Historical time
Mixture of chance and necessity
– Random factors can change the course of history
Transformation of structures and institutions
Irreversible and path dependent
Evolutionary economics portrays economic
growth as a process in historical time, but
much of mainstream economic theory is
based on a reversible time

Some questions on economic growth
GDP pc in Japan in 1900 was 1180, in
Argentina in 1900: 2756; in 2000 it was
21069 and 8544 (respectively); how can we
explain this?
How did regional growth patterns in Italy
diverge so much?
Can we explain such questions with a
deterministic, a-historical approach?

Technology and economic growth:
evolution and history
Two models:
– Conlisk model: evolution and growth
– Silverberg/Verspagen: structural transformations
and growth

The Conlisk model
y( t ) = [ y i ( t )] is
an infinitely long, ordered vector
of plant productivities
Labour L(t) populates plants (1 unit for each
plant), grows exogenously at rate n, and is
allocated efficiently over plants, hence
Int [ L ( t )]
∑ y (t )
Y (t ) = i
i =1
New plants m arrive due to exogenous saving:
m ( t ) = Int [ sL ( t )]

Growth in the Conlisk model
y( t + 1) = Rank[ x ( t ), (1 − δ ) y( t )]
x(t) is the vector of m(t) new plants
log x i ( t ) = µ ( t ) + σε i ( t ), ε i ( t ) IID( 0,1)

Specification of novelty (evolutionary
mechanism)
z(t +1) = Rankx(t ),z(t ))
(
Knowledge stock (all
inventions ever made):
k( t )
1
∑ log z i ( t )
µ( t ) =
Innovations are “mutations” of
k( t ) i =1
best-practice knowledge:
k(t ) = Int(βL(t ))

Results for evolutionary specification
growth converges to a fixed rate g
∂g ∂g ∂g
> 0, > 0, < 0.
∂σ ∂m ∂k
Greater variation in productivities in new plants
increases the likelihood of large innovations
Larger number of new plants every year increases the
number of opportunities to increase best-practice
If k grows larger, the list grows longer and it is harder to
improve on it

Conclusion
Randomness plays an essential role
The growth rate is a “random walk”: random shocks
have a permanent effect on the growth path

Evolutionary models and historical
transformations
Can an evolutionary model explain a
phenomenon like the emergence of modern
economic growth?
Evolution, self-organization and complexity
theory: emergent properties
– Micro-level interactions lead to ordered patterns at
the macro level
A model by Silverberg & Verspagen

Silverberg Verspagen model
Competing technologies, diffusing according
to differential profit rates
Firms, each innovating
Innovations drawn from Poisson distribution,
R&D determines arrival rate, fixed innovation
step
Profits re-invested, in R&D or capital
expansion: trade-off for the firm
R&D strategies

R&D strategies and firm interaction
R&D strategy is fraction R&D/profits
Change strategy as a result of
– Random mutation (fixed probability)
– Relatively bad performance (compared to other
firms)
Imitation of R&D strategy of other (successful) firms
Mutation of R&D strategy
All formulated in terms of probabilities and
thus governed by randomness

Experimental setup
Start all firms with 0 R&D strategies
Allow firms to “discover” R&D
Observe what happens if and when they
discover R&D

Model results – a typical run

Summary of the results
Sudden change of system characteristics is
like the transformation to modern economic
growth (Industrial Revolution)
The exact timing of the “transformation to
modern economic growth” is hard to predict
The probability of the transformation
happening (within a fixed time window)
depends on model parameters, such as
technological opportunities