1. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Spillover Diffusion, Agglomeration and
Distance
a Spatial Extension of the Knowledge Production Function
Approach
Giovanni Guastella1
1 MSc in Economics and Geography
Utrecht University
Thesis Dissertation, July 2010
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
2. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Motivation
NGT (Romer [20], Lucas [13]) stresses the role of
knowledge spillovers as source of increasing returns (IR).
Altough IR are likely to cause divergence, it is argued that
spillovers diffusion may also contribute to convergence,
depending on the degree of localization of these
externalities (Grossman and Helpman [8]).
One problem ...
If one one side knowledge cannot be contained within walls, on
the other side it is not accessible from everywhere and
everyone.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
3. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Motivation
NGT (Romer [20], Lucas [13]) stresses the role of
knowledge spillovers as source of increasing returns (IR).
Altough IR are likely to cause divergence, it is argued that
spillovers diffusion may also contribute to convergence,
depending on the degree of localization of these
externalities (Grossman and Helpman [8]).
One problem ...
If one one side knowledge cannot be contained within walls, on
the other side it is not accessible from everywhere and
everyone.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
4. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Motivation
NGT (Romer [20], Lucas [13]) stresses the role of
knowledge spillovers as source of increasing returns (IR).
Altough IR are likely to cause divergence, it is argued that
spillovers diffusion may also contribute to convergence,
depending on the degree of localization of these
externalities (Grossman and Helpman [8]).
One problem ...
If one one side knowledge cannot be contained within walls, on
the other side it is not accessible from everywhere and
everyone.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
5. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Motivation
... and another problem
Altough the literature on innovavation and geography
(Audretsch and Feldman [2]) suggests that spillovers are higher
in agglomerated areas and the intensity decreases with
distance, it is not easy to establish a direct link between
geography, agglomeration and spillover diffusion. This paper
attempts to study the way geography, agglomeration and
spillovers cause innovative activities.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
6. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
7. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
8. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
9. Introduction and Literature
Data and Methods
Results
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
10. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
11. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
How do spillovers fit in economic theories
There is no doubt that spillovers determine increasing
returns, and this idea is maintained also in this work. What
is diffuclt is to define and identify spillovers.
Mainstream view: knowledge is a public good accessible
from everyone. Social returns from innovative investments
are higher than private ones.
Evolutionary view: there are geographical, social and
cultural barriers to knowledge flows. Physical and
technological distances are considered among the most
important obstacles to spillover diffusion.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
12. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
How do spillovers fit in economic theories
There is no doubt that spillovers determine increasing
returns, and this idea is maintained also in this work. What
is diffuclt is to define and identify spillovers.
Mainstream view: knowledge is a public good accessible
from everyone. Social returns from innovative investments
are higher than private ones.
Evolutionary view: there are geographical, social and
cultural barriers to knowledge flows. Physical and
technological distances are considered among the most
important obstacles to spillover diffusion.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
13. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
How do spillovers fit in economic theories
There is no doubt that spillovers determine increasing
returns, and this idea is maintained also in this work. What
is diffuclt is to define and identify spillovers.
Mainstream view: knowledge is a public good accessible
from everyone. Social returns from innovative investments
are higher than private ones.
Evolutionary view: there are geographical, social and
cultural barriers to knowledge flows. Physical and
technological distances are considered among the most
important obstacles to spillover diffusion.
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
14. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
15. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
16. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
17. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
18. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
19. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
20. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
21. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Definition of spillovers
knowledge cannot be entirely codified (explicit vs tacit)
knowledge transfer is costly
Distance is important because
it allows face-to-face contacts
it reduces costs of transmission
physical distance, cognitive distance, institutional distance,
...
... far more complex than NGT models would predict
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
22. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
23. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
24. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
25. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
26. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
27. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
28. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
29. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
The KPF Approach (Griliches, [7])
More efforts we put, more output we get
Ii = f (X1i , X2i , ..., Xni ) (1)
Empirical evidences are stronger at aggregate level
Localized Knowledge Spillovers
Labor mobility
Entrepreneurship and spin-off
Inter-firms collaborations
Pure vs pecuniary externalities?
...what standard methodologies [...] suggest to be pure
externalities, will turn out to be, at a more careful scrutiny,
knowledge flows that are mediated by market mechanisms...
Breschi and Lissoni [5]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
30. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
31. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
32. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
33. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
34. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
35. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
36. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
37. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
38. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
39. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Agglomeration and spillovers
Concentration of knowledge sources pushes the creation
of new knowledge (Jaffe, [11])
Geography is still a Black Box (Distance is Exogenous!!!)
However...
Externalities have not only positive effects
congestion costs
spatial and cognitive lock-in
What we define agglomeration economies is ...
Marshall’s specialization [14]
Porter’s competition [19]
Jacob’s diversity [10]
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
40. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
41. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-aggregate level
Use of WR&D to proxy spatial spillovers
elasticity to external R&D is about .07 (.04 to .11) and
spatial spillovers are more important of technological ones
(Bottazzi and Peri, [3])
elasticity to external R&D is about .025 and spillovers are
bounded within 300 km (Bottazzi and Peri, [4])
elasticity to external R&D is about .04; sipllover are
bounded within 176 miles and there are no spillovers
among technological neighbors (Greunz, [6])
the majority of spillovers are confined within regional
borders and, in any case, within 350 km from the origin
region (Moreno et al., [16])
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
42. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-aggregate level
Use of WR&D to proxy spatial spillovers
elasticity to external R&D is about .07 (.04 to .11) and
spatial spillovers are more important of technological ones
(Bottazzi and Peri, [3])
elasticity to external R&D is about .025 and spillovers are
bounded within 300 km (Bottazzi and Peri, [4])
elasticity to external R&D is about .04; sipllover are
bounded within 176 miles and there are no spillovers
among technological neighbors (Greunz, [6])
the majority of spillovers are confined within regional
borders and, in any case, within 350 km from the origin
region (Moreno et al., [16])
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
43. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-aggregate level
Use of WR&D to proxy spatial spillovers
elasticity to external R&D is about .07 (.04 to .11) and
spatial spillovers are more important of technological ones
(Bottazzi and Peri, [3])
elasticity to external R&D is about .025 and spillovers are
bounded within 300 km (Bottazzi and Peri, [4])
elasticity to external R&D is about .04; sipllover are
bounded within 176 miles and there are no spillovers
among technological neighbors (Greunz, [6])
the majority of spillovers are confined within regional
borders and, in any case, within 350 km from the origin
region (Moreno et al., [16])
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
44. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-aggregate level
Use of WR&D to proxy spatial spillovers
elasticity to external R&D is about .07 (.04 to .11) and
spatial spillovers are more important of technological ones
(Bottazzi and Peri, [3])
elasticity to external R&D is about .025 and spillovers are
bounded within 300 km (Bottazzi and Peri, [4])
elasticity to external R&D is about .04; sipllover are
bounded within 176 miles and there are no spillovers
among technological neighbors (Greunz, [6])
the majority of spillovers are confined within regional
borders and, in any case, within 350 km from the origin
region (Moreno et al., [16])
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
45. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-specific level
concentration of economic activities vary across industries,
industrial specialization has positive effects and spillovers
happen between regions specialized in similar industries
(Moreno et al.,[15]
positive interregional spillovers and positive effect of
specialization (no diversity)
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
46. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
At industry-specific level
concentration of economic activities vary across industries,
industrial specialization has positive effects and spillovers
happen between regions specialized in similar industries
(Moreno et al.,[15]
positive interregional spillovers and positive effect of
specialization (no diversity)
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
47. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
48. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
49. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
50. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
51. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
52. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
53. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
54. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
55. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Enlarged geographical scope - 250 NUTS II regions
Explicit role for geography (agglomeration, specialization,
competition and diversity)
Industry-specific analysis (13 manufacturing industries)
Interregional and inter-industry spillovers
Differentiation among different regimes based on
Human Geography
Physical Geography
Economic Geography
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
56. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Main idea: use aggregate data to find stronger evidence of
spillover
My idea: split as much a possible to find evidence of pure
spillovers and separate R&D spillovers from other
externalities
Externality Positive Effect Negative Effect
Interreg within industry spillovers industrial competition among regions
Inter-ind between industries spillovers regional competition amond industries
Agg market potential ongestion costs
Spec labor market pooling and low cognitive distance cognitive lock-in
Comp more incentives to innovate big firms invest more in research
Div cross-industry knowledge exchange too much cognitive distance
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
57. Introduction and Literature Introduction
Data and Methods Economic Theories, Agglomeration and Spillovers
Results Previous Findings
Conclusion and Policy Implications Research Hypothesis
My contribution
Main idea: use aggregate data to find stronger evidence of
spillover
My idea: split as much a possible to find evidence of pure
spillovers and separate R&D spillovers from other
externalities
Externality Positive Effect Negative Effect
Interreg within industry spillovers industrial competition among regions
Inter-ind between industries spillovers regional competition amond industries
Agg market potential ongestion costs
Spec labor market pooling and low cognitive distance cognitive lock-in
Comp more incentives to innovate big firms invest more in research
Div cross-industry knowledge exchange too much cognitive distance
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
58. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
59. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Model specification and assumptions
Iij =α0 + α1 R&Dij + α2 UNIi + α3 GOVi +
β1 WR&Dij + β2 R&Di,k=j +
(2)
γ1 AGGi + γ2 SPECij + γ3 COMPij + γ4 DIVi +
εi
α1 to α3 : home made investments by firms, universities
and governments
β1 : interregional spillovers
β2 : interindustry spillovers
γ1 to γ4 : externalities
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
60. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Model specification and assumptions
Iij =α0 + α1 R&Dij + α2 UNIi + α3 GOVi +
β1 WR&Dij + β2 R&Di,k=j +
(2)
γ1 AGGi + γ2 SPECij + γ3 COMPij + γ4 DIVi +
εi
α1 to α3 : home made investments by firms, universities
and governments
β1 : interregional spillovers
β2 : interindustry spillovers
γ1 to γ4 : externalities
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
61. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Model specification and assumptions
Iij =α0 + α1 R&Dij + α2 UNIi + α3 GOVi +
β1 WR&Dij + β2 R&Di,k=j +
(2)
γ1 AGGi + γ2 SPECij + γ3 COMPij + γ4 DIVi +
εi
α1 to α3 : home made investments by firms, universities
and governments
β1 : interregional spillovers
β2 : interindustry spillovers
γ1 to γ4 : externalities
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
62. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Model specification and assumptions
Iij =α0 + α1 R&Dij + α2 UNIi + α3 GOVi +
β1 WR&Dij + β2 R&Di,k=j +
(2)
γ1 AGGi + γ2 SPECij + γ3 COMPij + γ4 DIVi +
εi
α1 to α3 : home made investments by firms, universities
and governments
β1 : interregional spillovers
β2 : interindustry spillovers
γ1 to γ4 : externalities
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
63. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
64. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
65. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
66. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
67. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
68. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
69. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
70. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Measuring issues
POPi
AGGi = Areai
R&Dij j R&Dij
SPECij = /
i R&Dij i j R&Dij
FIRMSij
COMPij = EMPLOYEESij
2
1
DIVi = j R&Dij − J j R&Dij
Choice of W
Great circle distance. Which d?
K -nearest neighbors. Which k?
Physical contiguity. What about islands?
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
71. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Weighting spillovers
Do spillover depend on the source?
I made no differentiation of the source, meaning that all
neighbors and all industries contribute with the same weight!!!
Be care with the interpretation!!!
Equal weight to all
neighbors Equal weight to all
industries
Row standardization
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
72. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Differentiating across regimes
η = (α0 , α1 , α2 , α3 , β1 , β2 , γ1 , γ2 , γ3 , γ4 )
η = η1 AC + η2 AWC + η3 NAC + η4 NAWC
η = η5 CORE + η6 INTER + η7 PERIP
η = η8 NONLAG + η9 POTLAG + η10 LAG
Source ESPON project (Copiright ESPON 2006 -
http://www.espon.eu)
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
73. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Differentiating across regimes
η = (α0 , α1 , α2 , α3 , β1 , β2 , γ1 , γ2 , γ3 , γ4 )
η = η1 AC + η2 AWC + η3 NAC + η4 NAWC
η = η5 CORE + η6 INTER + η7 PERIP
η = η8 NONLAG + η9 POTLAG + η10 LAG
Source ESPON project (Copiright ESPON 2006 -
http://www.espon.eu)
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
74. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Differentiating across regimes
η = (α0 , α1 , α2 , α3 , β1 , β2 , γ1 , γ2 , γ3 , γ4 )
η = η1 AC + η2 AWC + η3 NAC + η4 NAWC
η = η5 CORE + η6 INTER + η7 PERIP
η = η8 NONLAG + η9 POTLAG + η10 LAG
Source ESPON project (Copiright ESPON 2006 -
http://www.espon.eu)
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
75. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
76. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
77. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
78. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
79. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
80. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
81. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
82. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
83. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
84. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Patent counts as measure of regional innovation
PA are not a good proxy for innovations
PA underestimate innovation in small firms (Pakes and
Griliches, [17])
Big firms tend to overpatenting innovations
Patents do not reflect the economic value of innovation
(Hall et al., [9])
Literature based measures better proxy real innovations
(Pavitt et al., [18], Kleinknecht, [12])
All successfull innovations are considered
Are costly to be produced
Comparison depends on how data are collected
Does it make the difference at aggregate level?
NO!!! Acs et al., [1] provide evidence that in a KPF framework
both measures lead to identical conclusions
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
85. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
R&D data
R&D data at regional industry-specific level are not
available
Regional data are derived from national levels using
symplifying assumption
R&Dij EMPij
= (3)
NAT − R&Dj NAT − EMPj
NOTE!!!
The share of R&D per worker is costant across regions in the
same country for each industry
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
86. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
R&D data
R&D data at regional industry-specific level are not
available
Regional data are derived from national levels using
symplifying assumption
R&Dij EMPij
= (3)
NAT − R&Dj NAT − EMPj
NOTE!!!
The share of R&D per worker is costant across regions in the
same country for each industry
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
87. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
R&D data
R&D data at regional industry-specific level are not
available
Regional data are derived from national levels using
symplifying assumption
R&Dij EMPij
= (3)
NAT − R&Dj NAT − EMPj
NOTE!!!
The share of R&D per worker is costant across regions in the
same country for each industry
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
88. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Reconciling SIC codes with IPC classes
Schmoch et al., [21] provided a table to reconcile 4-digit
IPC with SIC industries
PA data are provided by Eurostat at 3-digit IPC class
It may happen that one IPC code belongs to more than
one SIC industries
I counted the times every IPC appears in a SIC. The share
of the count wrt total is the proportion of patents attributed
to the SIC
Industry SIC IPC
Food DA: food A01 C12 C13 A21 A23 A24
Textile DB: textile D04 D06 A41
Leather DC: leather A43 B68
Wood DD: wood B27 E04
Paper DE:paper, pub. and print. B41 B42 B44 D21
Fuels DF: petroleum and nuclear fuel C10 G01
Chemical DG: chemicals A01 A61 A62 ...
... ... ...
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
89. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Reconciling SIC codes with IPC classes
Schmoch et al., [21] provided a table to reconcile 4-digit
IPC with SIC industries
PA data are provided by Eurostat at 3-digit IPC class
It may happen that one IPC code belongs to more than
one SIC industries
I counted the times every IPC appears in a SIC. The share
of the count wrt total is the proportion of patents attributed
to the SIC
Industry SIC IPC
Food DA: food A01 C12 C13 A21 A23 A24
Textile DB: textile D04 D06 A41
Leather DC: leather A43 B68
Wood DD: wood B27 E04
Paper DE:paper, pub. and print. B41 B42 B44 D21
Fuels DF: petroleum and nuclear fuel C10 G01
Chemical DG: chemicals A01 A61 A62 ...
... ... ...
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
90. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Reconciling SIC codes with IPC classes
Schmoch et al., [21] provided a table to reconcile 4-digit
IPC with SIC industries
PA data are provided by Eurostat at 3-digit IPC class
It may happen that one IPC code belongs to more than
one SIC industries
I counted the times every IPC appears in a SIC. The share
of the count wrt total is the proportion of patents attributed
to the SIC
Industry SIC IPC
Food DA: food A01 C12 C13 A21 A23 A24
Textile DB: textile D04 D06 A41
Leather DC: leather A43 B68
Wood DD: wood B27 E04
Paper DE:paper, pub. and print. B41 B42 B44 D21
Fuels DF: petroleum and nuclear fuel C10 G01
Chemical DG: chemicals A01 A61 A62 ...
... ... ...
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
91. Introduction and Literature
Data and Methods The model
Results A Regional Innovation dataset
Conclusion and Policy Implications
Reconciling SIC codes with IPC classes
Schmoch et al., [21] provided a table to reconcile 4-digit
IPC with SIC industries
PA data are provided by Eurostat at 3-digit IPC class
It may happen that one IPC code belongs to more than
one SIC industries
I counted the times every IPC appears in a SIC. The share
of the count wrt total is the proportion of patents attributed
to the SIC
Industry SIC IPC
Food DA: food A01 C12 C13 A21 A23 A24
Textile DB: textile D04 D06 A41
Leather DC: leather A43 B68
Wood DD: wood B27 E04
Paper DE:paper, pub. and print. B41 B42 B44 D21
Fuels DF: petroleum and nuclear fuel C10 G01
Chemical DG: chemicals A01 A61 A62 ...
... ... ...
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
92. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
95. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
96. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Simple Spatial Lag
Industry Spillovers Externalities
Home Interreg Inter-ind Agg Spec Comp Div
Food +
Textile + + -
Leather +
Wood + -
Paper + - + +
Fuels - + +
Chemical + - -
Rubber + + + +
Non Metal + + + - +
Metal + - + - +
Machinery + - + +
Electrical + + +
Transport + - +
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
97. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Outline
1 Introduction and Literature
Introduction
Economic Theories, Agglomeration and Spillovers
Previous Findings
Research Hypothesis
2 Data and Methods
The model
A Regional Innovation dataset
3 Results
Basic Results
Spatial lag
Spatial lag and Spatial Regimes
4 Conclusion and Policy Implications
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
98. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Density regime - Agglomerated with Centres
Industry Spillovers Externalities
Home Interreg Inter-ind Agg Spec Comp Div
Food + -
Textile + +
Leather +
Wood +
Paper + +
Fuels - + - +
Chemical + -
Rubber + + +
Non Metal + + + -
Metal + +
Machinery + + +
Electrical + +
Transport + + +
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance
99. Introduction and Literature
Basic Results
Data and Methods
Spatial lag
Results
Spatial lag and Spatial Regimes
Conclusion and Policy Implications
Density regime - Agglomerated Without Centres
Industry Spillovers Externalities
Home Interreg Inter-ind Agg Spec Comp Div
Food
Textile +
Leather + -
Wood - + + + -
Paper
Fuels
Chemical
Rubber + -
Non Metal +
Metal + -
Machinery + + -
Electrical +
Transport +
Giovanni Guastella Spillover Diffusion, Agglomeration and Distance