geography and growth: the new economic geography (neg)...
TRANSCRIPT
Geography and Growth:
The New Economic Geography
(NEG) Perspective
Harry Garretsen
DIMETIC Summer School, 30-6-2009, Pecs
Nice timing for this lecture…..
• Right after Ron Boschma’s lecture:
is there life after EEG’s take on growth?
• Increasingly, NEG takes a dynamicapproach → deals with geography and growth
• NEG “begins” in 1991 and time to takestock (13 October 2008…………….)
Outline• New Economic Geography (NEG): key insights
(“trade and geography”, Krugman’s Nobelprize!)
• NEG work after Krugman (1991)
• Relationship with other theories of geographyand/or growth (very quickly)
• Stylized facts about economic growth
• NEG applied to economic growth
• Summing up/discussion…………..
NEG key insights I
• NEG’s core model: Krugman (1991, JPE)
• NEG originates in international trade theory, notin economic geography or growth theory
• We proceed in 3 steps: Krugman (1979, 1980,
1991)
• International trade theory in 1979: old (=18th
century) theory (Ricardo) at odds with facts
• Theory: inter-industry trade; facts: intra-industry
trade (it’s not “cloth for wine” anymore)
Manufacturing intra-industry trade; 1988-2000, selected countries
Manufacturing intra-industry trade (% of total manufacturing); 1988-2000
25
50
75
1991 1994 1997 2000year
% in
tra
-in
du
str
y tra
de
Hungary
Japan
South Korea
Mexico
USA
Germany
Australia
Intra-industry trade; China and Chinese Taipei, 1976-2005
Intra-industry trade; Grubel-Lloyd index (3-digit level, weighted average)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
1975 1980 1985 1990 1995 2000 2005year
GL
-in
de
x
China
Chinese Taipei
NEG Key Insights II:
• Krugman (1979): introduce internalincreasing returns to scale
• Internal IRS: model of imperfect competition (Dixit and Stiglitz, 1977)
• Rationale for intra-industry trade, but norole for geography (or growth) yet
iixl βα +=
Average costs under increasing returns to scale
Average
costs
Output
NEG Key Insights III
• Krugman 1980: add transport costs to IRS
-assume two countries A and B: S>S* (market size A larger than B);
-and assume transport costs T>0;
if α >TS*, then locate firm in larger market
• “Home market effect”: geography matters
• But: why should S>S* to begin with????
NEG Key Insights IV
• Krugman (1991): 1st NEG model:
add factor (labour) mobility to T and IRS
• Also external IRS (pecuniary or marketsize externality)
• The 1 million $ Q: where will footloosefirms&workers locate?
• Answer: it depends……………
NEG Key Insights V
• ……..it depends on relative strength of agglomeration and spreading forces
• Agglomeration forces: home market effect, price index effect
• Spreading forces: competition effect
• Relative strength: key model parameters, notably T
[Where’s the novelty of Krugman 1991?]
The relative real wage in region 1 (case of T=1.7)
0,97
1
1,03
0 0,5 1
share of manufacturing workers in region 1 (lambda1)
rela
tive
re
al w
ag
e (
w1
/w2
)
A
DC
B
E
F
The Tomahawk diagram
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Panel a
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Panel a
NEG Extensions
• See previous slide (Tomahawk), extreme set of equilibria:
change menu of aggl. and spreading forces
[e.g. no labour mobility, congestion, non-tradable goods (housing)……..]
→ get rid of the Tomahawk
• Empirical implications? Yes ……..
• What’s the relevance for economic growth?
How to position basic NEG model?
• International trade theory? (NEG adds factor
mobility)
• Urban & Regional economics? (NEG adds
general equilibrium analysis of spatial linkages)
• (Evolutionary) economic geography? (same Q
but different tools)
• Economic growth theory? (basic NEG model is
static=allocation model, thus …→ ………..→
NEG and growth: 3 questions
• Why is (even static) NEG useful foranalysis of economic growth?
• How to turn NEG into a growth model?
• (Discussion) What’s the value added of NEG for innovation and growth analysis?
Q1: Why NEG might be useful
• 5 stylized facts on economic growth(“Kaldor” 2009):
1) Gdp per capita increases over time
2) Differences in growth are persistent
3) Scale matters (within-between country)
4) Growth process is lumpy (stagnation/ growth spurts
5) Changes in gdp rankings: “leapfrogging”
1) Per capita income increases over time; 2) (increasing) dispersion (no convergence)
a. His togram of ln(incom e per capita)
0
5
10
15
20
25
5.7 6.6 7.5 8.4 9.3 10.2 11.1
ln(income per capita)
nu
mb
er
of
co
un
trie
s
1950
Qatar
Kuw aitGuinea
Bissau
b. His togram of ln(incom e per capita)
0
5
10
15
20
25
5.7 6.6 7.5 8.4 9.3 10.2 11.1
ln(income per capita)
nu
mb
er
of
co
un
trie
s
1968
Qatar
Malaw i
Burundi
Chad
Guinea
1) Per capita income increases over time; 2) (increasing) dispersion (no convergence)
c. His togram of ln(incom e per capita)
0
5
10
15
20
25
5.7 6.6 7.5 8.4 9.3 10.2 11.1
ln(income per capita)
nu
mb
er
of
co
un
trie
s
1986
USA
Tanzania
Chad
Guinea
d. His togram of ln(incom e per capita)
0
5
10
15
20
25
5.7 6.6 7.5 8.4 9.3 10.2 11.1
ln(income per capita)
nu
mb
er
of
co
un
trie
s
2003
USAZaire
Regional income inequality in the EU NUTSII regions: 2) persistent differences?
EU regional income inequality; Lorenzcurves 1995 and 2004
0
1
0 1cumulative share of population
cu
mu
lati
ve
sh
are
of
inc
om
e
19952004
diagonal
Regional income inequality in the EU: 3) scale matters
EU regional income inequality: Theil index and Gini coefficient
0.02
0.03
0.04
0.05
0.06
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
0.08
0.12
0.16
0.20
0.24
Theil within countries (left hand scale)
Theil between countries (left hand scale)
Gini coefficient (right hand scale)
Leaders and laggards in the world economy, 1-2003: 4) lumpiness and 5) leapfrogging
10001 1500 1600 18001700 200019000
100
200
300
500
400
income per capita (% of world average)
year
Italy
IraqIran
Netherlands
UK
Australia
USA
Switzerland
IndiaChina
oAfrica
W Offshoots
New Zealand
AustraliaMany Many
Italy
10001 1500 1600 18001700 200019000
100
200
300
500
400
10001 1500 1600 18001700 200019000
100
200
300
500
400
1 1500 1600 18001700 200019000
100
200
300
500
400
income per capita (% of world average)
year
Italy
IraqIran
Netherlands
UK
Australia
USA
Switzerland
IndiaChina
oAfrica
W Offshoots
New Zealand
AustraliaMany Many
Italy
Now return to Q1: standard growth models do focus on 1 and 2
(and 3), NEG emphasizes 3,4 and 5
1) Gdp per capita increases over time
2) Differences in growth are persistent
3) Scale matters (within-between country)
4) Growth process is lumpy (stagnation/ growth spurts
5) Changes in gdp rankings: “leapfrogging”
The Tomahawk diagram
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Panel a
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Sustain points
Break point
Transport costs T1
0
1
λ1
0.5
Stable equilibria
Unstable equilibria
B
S0
S1
Basin of attraction for spreading equilibrium
Basin of attraction for agglomeration in region 1
Basin of attraction for agglomeration in region 2
Panel a
The LUMPY Evolution of agglomeration: STATIC model fo 12 regions
Herfindahl index
0
0.1
0.2
0.3
0.4
0.5
0 200 400 600 800
reallocation
STATIC NEG model (12 region Krugman model) already shows LEAPFROGGING!!
Evolution of share of manufacturing
0
0,1
0,2
0,3
0,4
0,5
0 200 400 600 800
reallocation
3
6
9
Q2: How to turn NEG into growth model?• Following i.a. Baldwin and Forslid (2000) many
(endogenous) NEG growth models
[→ see Baldwin&Martin (2004) for a survey]
Endogenous growth ingredients: (localized) knowledge spillovers +…….
NEG ingredients: spatial linkages (geography)
What does this growth-NEG “marriage” look like??
Stability in the
Baldwin-Forslid
economic growth-NEG
model
0.2 0.4 0.6 0.80 1
1.5 1.26 1.14 1.06 1∞
Freeness
of trade
Knowledge
spillovers
Implied T
1
Agglomeration stable; spreading unstable
Agglomeration stable; spreading stable
Agglomeration unstable; spreading stable
Finally Q3: value added of NEG growth view
• Within (!) mainstream endogenous models of growth & innovation:
-focuses on spatial linkages (NEG addsgeography)
-offers richer explanation of “growth facts”
-enables “what if” analysis of (policy) experiments or shocks……………….
The “what if” bonus
• Recall the Tomahawk for the basic idea
• Real “shocks” (wars & disasters); economic integration shocks (examples: EU integration; increased labor betweenChinese cities; lower trade barries in SSA)
• Highly stylized example: EU integrationwith Krugman (1991) model
Multi-region simulations of Krugman model for NUTSII regions (EU 15)
Multi-region simulations of Krugman model for NUTSII regions (EU 15)
Summing Up/Discussion….
• NEG adds geography to mainstreameconomics
• Incorporation of NEG into growth models yields new theoretical and empiricalinsights
• A discussion about the relevance of the NEG perspective is a discussion about the relevance of mainstream economics
• EEG and NEG: complements/substitutes?