Regional Convergence and Innovation

Performance in Europe
Outline:
1. Introduction
2. Some Theoretical Considerations
3. Empirical Trends of Convergence and Disparities
4. β−Convergence and the Recession
5. Economic Growth: Innovation and Entrepreneurship
6. Conclusion and Perspectives

Trends in Applied Economics 1

Some Issues to Address

Do we observe convergence for income per capita among the

regions in Europe or will Europe split up in an A-team and a
B-team?
How does innovation performance affect the potential of
economic growth?
Can innovation performance incorporate in the production
function?
Will the A-team be concentrated in the larger cities or urban
areas?
Will the B-team be on the countryside only?
Is it best to live in the city?
Can we solve some of these issues by use of land planning?

Trends in Applied Economics 2

Literature:

Andreas P. Cornett and Nils Karl Sørensen (2008), International vs. Intra-national
Convergence in Europe – an Assessment of Causes and Evidence. Investigaciones
Regionales. Vol. 13 pp. 35−56.

Andreas P. Cornett and Nils Karl Sørensen (2012), Innovation and regional disparities
– a survey of regional growth drivers and economic performance. Chapter 5 in: Charlie
Karlsson, Börje Johansson and Roger R. Stough (editors): Innovation, Technology and
Knowledge. Routledge Press.

Andreas P. Cornett and Nils Karl Sørensen (2014), Regional GDP Convergence in the
European Regions in the light of the Economic Recession. From: P. Linde (editor)
(2014): Symposia in Applied Statistics, page 111-117, University of Copenhagen and
Statistics Denmark, ISBN 978-87-501-2111-4.

Trends in Applied Economics 3

1. Introduction
In the two first articles it is claimed that:
Large intra-regional disparities do not necessarily lead to
lower economic growth on the national level than smaller
disparities do
Contrary polarization of economic growth can lead to excess
growth and an increase in overall convergence
Innovation performance may have different implications on
the regional growth and convergence performance

What are the implications:

Overall convergence is present, but not at the regional level
At the regional level patterns are very different and
divergence is frequently observed
Innovation performance has very different impact
This has different implications for the policy on growth
Trends in Applied Economics 4
2. Some Theoretical Considerations
(Cornett and Sørensen (2012) appendix I)

Convergence imply that the growth rate in GDP per capita

in an open economy framework will stabilize and the same
rate for all regions in the long run
In the short run there will be a situation where the growth
rate in poorer nations will exceed that in rich nations
This will also be present at the regional level
So poor regions with a low level of GDP per capita will
“catch-up” on the rich regions with high level of GDP
This process is based on the neo-classical growth theory i.e.
the Solow model

Trends in Applied Economics 5

Let us look at the neo-classical Solow model:
Here, I use the model to look for a decrease in savings:
 Savings moves downward and increases investment
 The decrease will reduce demand for investment goods and growth will decrease
 Growth will be negative, and we reach a lower level of wealth and growth
 The model will not predict continuous growth. Here we must introduce IRS in the production
function

Y
y= = income per capita
N
y = f(k)
y*
(n+d)k
y**
s0 y
E0
s1 y

EL E1

K
k= = capital/labour ratio
N
k** k*

Trends in Applied Economics 6

Growth pattern will look as:
y s E1
E0 s0
y*
s1
y** E0 EL
E1 EL

t0 tL t0 tL

A problem with this model is that growth is not modeled such that it
is persistent in time
The unique balanced growth pattern can be written as:

y k A
   g y  gk  g A
y k A
Notice that relative to the basic model technology has been dynamic
This just says that all growth rates in the long run are similar
Trends in Applied Economics 7
 If convergence is similar for all, a steady state growth rate called *
should be present.
~ ~
Denoting y  Y / AL and k  K / AL as the output and capital
efficiency of labor
Then we can Taylor expand around the steady state path and
obtain: ~
k ~* ~
~   (log k  log k )
k
Solving the differential equation, we obtain:

log ~
y (t )  (1  e  t ) log ~
y *  e  t log ~
y (0)
or
log y (t )  log y (0)
 (1  e  t ) ln A(0)  gt  (1  e  t ) ln y (0)  (1  e  t ) ln y *

Trends in Applied Economics 8

 After this brief (and hardly understandable) manipulation we can set
up the following on types of convergence:
β-convergence: Reformulation of the equation
log y (t )  log y (0)   0  1 ln y (0)  x' 
or
 log yt   0  1 ln y0
where y = GDP per capita, 0 = start time and t = end time (t−0)
if β1 < convergence (DRS, CRS stable model)
β1 = no deterministic pattern (knife-edge model)
β1 > divergence (IRS and unstable model)

This equation can be estimated by use of OLS

β0 = constant including the change in A i.e. the technology

Trends in Applied Economics 9

An alternative to this type of convergence is:

σ-convergence: The standard deviation of a cross section for each

year for all regions in a given country divided by the mean

The σ-convergence is then the cross-section coefficient of

variation defined as:
sR
 R  CVR 
XR
If the standard deviation and CV is decreasing convergence is
observed
It is evident that the two concepts are closely related. It can be
shown that β-convergence is a necessary, but not sufficient
condition for σ-convergence
Trends in Applied Economics 10
3. Empirical Trends of Convergence & Disparities

Data ranges from 1994 to 2005 and covers GDP per capita by
European regions
The source is EUROSTAT
For most countries, statistics are collected at the NUTS 2 level,
but for Denmark at NUTS 3
In total we have 369 regions/observations
Initially inspect some maps and Box-plots to look at some
overall trends

Trends in Applied Economics 11

GDP per capita growth 1994 to 2005 and level 2004

Trends in Applied Economics 12

Trends in Applied Economics 13
Consider now some Box-plots of Regional Disparities 2004 (all in €)

15000 20000 25000 30000 35000 40000 45000 50000 15000 20000 25000 30000 35000 40000 45000 50000
Denmark Germany

15000 20000 25000 30000 35000 40000 45000 50000 0 5000 10000 15000 20000 25000
Sweden Portugal

15000 20000 25000 30000 35000 40000 45000 50000 0 5000 10000 15000 20000 25000
Finland Czech Republic

15000 20000 25000 30000 35000 40000 45000 50000 0 5000 10000 15000 20000 25000

Italy Poland

Trends in Applied Economics 14

Aggregate result for β–convergence:
14,00 Growth rate 1995-
2004
12,00
y = 22.97 - 1.97x
10,00 2
R = 0.49
8,00

6,00

4,00

2,00

0,00
6,00 6,50 7,00 7,50 8,00 8,50 9,00 9,50 10,00 10,50 11,00
Ln GDP per capita 1995

Trends in Applied Economics 15

β-Convergence and Divergence in 19 EU-countries
and Norway
Constant, β0 β1-coefficient R2 Standard Obs.
Coef. Std.dv. P-value Coef. Std.dv. P-value Error
EU Total C 22.97 0.97 0.00 –1.97 0.10 0.00 0.49 1.73 369
Small EU members:
Denmark I –12.65 33.17 0.72 1.38 3.25 0.69 0.04 0.97 6
Sweden D –22.93 12.52 0.11 2.65 1.26 0.07 0.39 0.40 9
Norway C 29.71 12.86 0.07 –2.63 1.29 0.10 0.45 0.69 7
Finland D –5.07 3.85 0.24 0.91 0.34 0.07 0.52 0.16 7
Netherlands I –5.46 5.75 0.36 0.93 0.58 0.13 0.15 0.31 16
Belgium I 1.27 5.00 0.80 0.14 0.51 0.79 0.01 0.51 14
Austria C 17.31 3.06 0.00 –1.50 0.31 0.00 0.68 0.21 13
Greece I 26.83 24.00 0.28 –2.42 2.67 0.38 0.05 1.69 17
Portugal I 14.93 17.71 0.84 –1.09 1.96 0.60 0.04 0.98 9
Large EU members:
Germany C 14.68 3.43 0.00 –1.34 3.87 0.00 0.23 0.59 51
France C 9.63 3.93 0.00 –0.70 0.40 0.09 0.09 0.50 34
Italy C 11.04 3.14 0.00 –0.69 0.33 0.05 0.15 0.47 27
Spain I 10.43 3.02 0.00 –0.54 0.32 0.11 0.11 0.31 24
UK D –3.22 4.54 0.48 1.02 0.47 0.04 0.09 0.65 49
New EU members:
Poland D –13.04 8.56 0.14 2.48 1.09 0.03 0.20 0.74 23
Czech rep. D –26.30 10.78 0.04 4.07 1.30 0.02 0.58 0.78 9
Hungary D –20.16 8.42 0.04 3.64 1.05 0.01 0.60 0.66 10
Bulgaria D –16.89 4.83 0.01 3.46 0.69 0.00 0.78 0.32 9
Slovenia I 18.86 8.20 0.08 –1.34 1.00 0.25 0.31 1.22 6
Romania D –33.92 10.62 0.01 5.59 1.43 0.00 0.58 0.88 13

Trends in Applied Economics 16

Estimates of σ-convergence in 19 EU-countries (CV = σ/mean)
CV 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
EU total C 54.1 52.3 50.3 49.9 50.0 50.3 48.9 48.3 47.7 47.7
Small EU members:
Denmark D 14.2 14.7 15.0 14.4 16.1 17.0 16.9 17.9 17.3 18.7
Sweden D 12.4 14.1 15.9 16.6 17.6 17.6 16.9 17.2 16.3 16.7
Finland D 16.8 18.5 17.8 20.3 22.2 19.1 22.2 20.6 19.2 18.0
Netherlands D 14.0 15.9 15.7 15.6 15.9 16.1 15.6 15.8 15.4 15.5
Belgium I 36.0 36.4 35.7 35.4 36.1 36.2 36.4 36.6 35.8 35.6
Austria C 20.4 20.6 19.8 19.6 19.3 18.9 19.0 19.0 18.5 17.8
Greece D 17.5 18.3 16.9 16.1 15.0 19.5 19.7 20.3 20.6 21.0
Portugal I 19.2 18.9 20.0 20.6 18.0 19.4 18.4 19.8 19.1 19.1
Large EU members:
Germany I 24.1 23.6 23.7 24.0 24.0 24.2 24.6 24.0 23.4 23.1
France C 21.3 21.7 21.1 20.6 20.6 21.6 20.6 20.4 20.7 20.6
Italy C 26.5 26.6 25.6 25.9 25.2 25.4 24.8 24.4 24.7 25.0
Spain I 20.2 19.7 20.1 20.1 20.2 21.9 21.1 20.4 19.7 19.3
UK D 25.0 25.4 26.7 27.8 28.2 29.8 28.8 29.5 29.5 29.4
New EU members:
Poland D 14.5 16.4 17.5 18.5 20.6 20.1 21.2 20.8 21.0 20.5
Czech republic D 26.5 26.5 29.3 33.4 35.7 37.3 39.9 40.9 40.9 40.5
Hungary D 23.6 25.4 27.2 27.5 29.9 31.8 31.8 34.5 33.4 32.6
Bulgaria D 17.8 17.6 16.9 15.5 19.1 16.0 18.8 22.0 22.4 24.5
Slovenia C 59.9 56.3 55.2 56.4 60.7 56.7 56.3 56.2 54.0 52.2
Romania D … … … 20.5 20.8 33.8 34.9 34.1 30.0 30.6

Trends in Applied Economics 17

We observe that (especially for β-convergence) I:
The over all rate of convergence for EU total is 1.97 percent
(close the famous 2 percent found by many other studies)
Many countries has internal divergence, but contribute to
overall convergence
This is especially true for the new EU-members. Here the rural
areas becomes more rural! This stress the need for a regional
policy on cohesion. Here high speed of divergence or
convergence is observed
The large EU-members experience convergence with low speed
Many of the small EU-members are inclusive or has
divergence. This is especially true for some Nordic countries
like Sweden or Finland. Is the “Nordic model” under pressure?
The trends are confirmed for σ-convergence as well
Trends in Applied Economics 18
We observe that (especially for β-convergence) II:
Our results can be summarized in the following table

Trends in Applied Economics 19

4. β−Convergence and the Recession
How did the recession in 2008 affect the rate of convergence
among the regions in the European Union?
The overall rate of β−convergence decreased and reached a
level equal to about −1.7 over the period 2008 to 2010
Since the recession growth has been much more diversified
among the European regions
No countries experiences overall convergence by regions
For most countries the pattern of convergence in inconclusive

Trends in Applied Economics 20

During the first
period regions can be
classified into two
groups

During the second

period volatility has
increased

Trends in Applied Economics 21

 Period 1995 to 2006 Period 2007 to 2010 Comparison Obs.
constant β0 coefficient R2 constant β0 coefficient R2 1995 - 2007 –
β1 β1 2004 2010

EU Total 29.55*** −2.63*** 0.67 15.67*** −1.60*** 0.15 C C 356

Small EU members:
Denmark −10.65 1.29 0.21 −24.65* 2.39* 0.64 I D 6
Sweden −15.88 1.98* 0.33 17.62 −1.63 0.03 D I 9
Finland −4.90* 0.90* 0.78 16.57 −1.58 0.02 D I 7
Netherlands −7.37 1.13 0.16 −44.36** 4.37** 0.36 I D 17
Belgium 3.78 −0.09 0.01 6.97 −0.58 0.03 I I 14
Austria 16.71*** −1.40*** 0.65 5.57 −0.46 0.09 C I 13
Greece 32.59* −2.93 0.14 −12.74 1.23 0.09 I I 17
Portugal 14.41 −1.01 0.03 8.48 −0-85 0.06 I I 9
Large EU members:
Germany 18.23*** −1.66*** 0.33 8.11 −0.71 0.02 C I 51
France 19.19*** −1.67*** 0.28 8.89 −0.87 0.07 C I 34
Italy 10..88*** −0.68*** 0.32 −2.64 0.21 0.01 C I 27
Spain 13.44*** −0.84* 0.17 −0.40 −0.02 0.01 C I 23
UK −3.05 0.95 0.04 −13.20* 0.79** 0.06 I D 49
New EU members:
Poland −14.16 2.80** 0.21 −9.32* 1.37** 0.26 D D 23
Czech Republic −4.90* 3.54** 0.54 1.23 0.25 0.02 D I 8
Hungary −20.06** 3.56*** 0.65 −3.56 0.30 0.01 D I 10
Bulgaria −30.61*** 5.59*** 0.90 −26.34*** 3.67*** 0.82 D D 9
Slovakia 42.56 −4.25 0.40 1.89 0.29 0.04 I I 6
Romania −51.32*** 8.47*** 0.67 −13.64* 1.56* 0.23 D D 13

Trends in Applied Economics 22

5. Economic Growth: Innovation and Entrepreneurship
The parameter A in the production function can be made
operational by inclusion of indicators related to innovation
and entrepreneurship (we look at other ways later in the course)
In this way the β0 term i.e., the constant in the convergence
regression can in a similar way be given an interpretation
The Regional Innovation Scoreboard 2023 – see link:
https://op.europa.eu/en/publication-detail/-/publication/c849333f-25db-11ee-a2d3-
01aa75ed71a1/language-en/format-PDF/source-289680093
The scoreboard has been published by the EU commission
since 2005 (most times every second year)
Initially there were 5 indicators based on 25 sub indicators, but
the number of main indicators has been extended
Trends in Applied Economics 23
Illustration of the web-page

European Commission, Directorate-General for Research and Innovation,

Hollanders, H., Es-Sadki, N., Regional Innovation Scoreboard 2023, Publications
Office of the European Union, 2023, https://data.europa.eu/doi/10.2777/70412

Trends in Applied Economics 24

RIS 2023 – Interactive statistical tool

Interactive statistical tool to

be found at:

https://ec.europa.eu/research
-and-
innovation/en/statistics/perfo
rmance-indicators/european-
innovation-scoreboard/eis#

Trends in Applied Economics 25

Trends in Applied Economics 26
Results from the RIS 2023
Aa

Trends in Applied Economics 27

The Innovation Scoreboard ranging from 0 to 1 is found as a
weighted index of the subcomponents and divided into 4
categories

Trends in Applied Economics 28

Regional Innovation Scoreboard 2017

Trends in Applied Economics 29

Regional Scoreboard Main Variables
Notice that the selection of variables are not clear relative to
theory
Variable: Measurement:
Knowledge workers Science and technology – core per cent of population

Life-long learning Participation in life-long learning per 100 population

aged 25–64

Medicine and high-tech manufacturing Employment out of total workforce

High-tech services Employment in high-tech sectors in per cent of total

workforce

Public research and development Public R&D in per cent of GDP

Business research and development Private R&D in per cent of GDP

Patents EPO patents per million population

Trends in Applied Economics 30

Process of flow/classification of RIS
using the 2023 variables

Trends in Applied Economics 31

The is a weak negative relation between the level of σ-
convergence and the scoreboard. This means that countries
with a low scoreboard has a higher rate of catch-up
The scoreboard is available for 202 regions at the NUTS II
level and the rate of β-convergence is equal to −2.69 (against
−1.97 for the full data set

Trends in Applied Economics 32

Scoreboard Classification and Convergence

Trends in Applied Economics 33

Regression Model Set Up
Dummy Model:

 log y R.t   0. LEADER  1 ln y0. R

  2 DFOLLOW . R   3 DMODERATE . R   4 DCATCH UP. R
Convergence Model:

 log y R.G .t   0  1 ln y0. R.G G  1, 2,3, 4

Where: R= Regions
G= Leaders (1), Followers (2), Moderate (3)
and Catch-up (4)
Trends in Applied Economics 34
What is observed
Upper part: β−convergence is fixed and constant term varies
Overall rate of convergence equals −2.49
For followers (1.53) and moderate (1.34) there is a mark-up
relative to the leaders
The mark-up is not significant for catching-up
Lower part: β−convergence and constant term varies:
All variables are significant, but explanatory power varies
The rate of convergence is the lowest for the leaders and
highest for the followers
Catching−up has a rate slightly above moderate
The potential for development is the highest for the followers
Trends in Applied Economics 35
Relation between Scoreboard and GDP

Trends in Applied Economics 36

Regression Model Set Up
In levels of GDP:
GDPR   0  1KWR   2 LLLR   3 MHTM R   4 HTS R   5 PR & DR   6 BR & DR
  7 DFOLLOW . R  8 DMODERATE . R   9 DCATCH UP. R

In changes of GDP replace GDPR with ∆GDPR

Model with city dummy:

GDPR   0  1KWR   2 LLLR   3 MHTM R   4 HTS R   5 PR & DR   6 BR & DR
  7 DCITY . R

Trends in Applied Economics 37

What is observed
Model to the left is in levels, whereas model to the right is in
growth terms
No surprising patents are important for development (the
negative sign on the growth model is due to the fact that the
highest concentration of patents is among the leaders)
Surprisingly public as well as private R&D is not significant
Med and high tech-services are overall positive significant
Life-long learning and knowledge workers are not significant
for levels, but significant for growth
These variables are then viewed as important development
parameters for regions without patents
The dummies confirm earlier results, but is positive for
followers
Trends in Applied Economics 38
Relation between Scoreboard, Cities and GDP

Trends in Applied Economics 39

What is observed
A dummy variable has been included in order to examine the
impact of larger city regions using the Losch approach (largest
city center in region)
In both cases the city dummy is significant
In levels the city variable takes a negative coefficient due
asymmetry between the city regions and the remaining rural
regions
OR: Large and wealthy cities in the European hemisphere
have lower degree of internal convergence (when ranked by
decile)
In growth terms: Large cities have positive impact on growth

Trends in Applied Economics 40

6. Conclusion and Perspectives
The research question unequivocal cannot be answered unequivocal
Overall, we observe convergence, but not at the disaggregate level
In many small countries or new EU-members divergence is found
In 3 out of 5 large countries in EU convergence is observed
Metropolitan (city) areas are often following a different pattern than
rural areas
Innovation has different impact, and this calls for different
development strategies
In the long run this may lead to a Europe in different speed
This may give rise to an A-team and a B-team
This may give rise to problems of redistribution by regions
Within each team convergence will be present but at different levels

Trends in Applied Economics 41