Transcript Economic Convergence

Economic
Convergence
A Theoretical Review and Application to
European Regions
Lorenzo Ferrari
Nightlights in Europe
Why are we interested in
convergence?
• Half the GDP in Europe is produced in a transnational
geographic region, called the “core”
• The core is characterised by higher per-capita income
and lower unemployment rate
• The reduction of regional disparities is a main goal for the
EU, already stated in the Treaty of Rome (1957)
• Has convergence taken place at the regional level in
the EU?
What convergence are we
looking for?
Poor economies grow
faster than rich ones
Faster growth when far
from steady state
ABSOLUTE
CONDITIONAL
• The neoclassical model only predicts conditional
• We need the two economies to have similar steady
states, implied by similar tastes and access to
technology
• This motivates the investigation of regional
convergence
Two Concepts of
Convergence
Catching up
process according
to which poor
countries tend to
grow faster than
rich ones
Reduction in the
cross-sectional
dispersion of percapita income
β-convergence
σ-convergence
Necessary but not sufficient
condition
β-Convergence
log 𝑦𝑖𝑡 𝑦𝑖,𝑡−1 = 𝑎𝑖𝑡 − 1 − 𝑒 −β ∙ log 𝑦𝑖,𝑡−1 + 𝑢𝑖𝑡 ,
• β > 0 convergence
• β < 0 divergence
Unexpected changes
in production
conditions or
preferences
• Intuition: if β > 0, then 1 − 𝑒 −β is positive and
− 1 − 𝑒 −β is negative. This implies that there exists a
negative relation between the initial level of per
capita income and its growth
σ-Convergence
• The variance of per-capita income at time t is
2
2
𝜎𝑡2 = 𝑒 −2β 𝜎𝑡−1
+ 𝜎𝑢𝑡
• This is a first-order differential equation with solution
2
2
𝜎
𝜎
𝑢
𝑢
−2βt +
𝜎𝑡2 = 𝜎02 −
𝑒
1 − 𝑒 −2β
1 − 𝑒 −2β
σ-Convergence
• The steady-state variance of per-capita income is
2
𝜎
𝑢
𝜎2 =
1 − 𝑒 −2β
• The steady-state dispersion of per-capita income
decreases in β but increases in 𝜎𝑢2 , the variance of
the random disturbance.
• This is why we say that β-convergence is a
necessary but not sufficient condition for σconvergence
And now…some
empirical evidence
from EU regions…
Empirical Analysis
• I study convergence among European regions in
terms of per-capita GDP
• Data: Per-capita GDP in PPS (Purchasing Power
Standards) for NUTS2 and NUTS3 from 1995 to 2010
• Convergence if a negative coefficient for log 𝑦𝑖,𝑡−1
(I do not estimate β directly)
log 𝑦𝑖𝑡 𝑦𝑖,𝑡−1 = 𝑎𝑖𝑡 − 1 − 𝑒 −β ∙ log 𝑦𝑖,𝑡−1 + 𝑢𝑖𝑡 ,
Nomenclature of territorial
units for statistics
NUTS 2 - Regions
Source: Lorenzo Ferrari (2013).
NUTS 3 - Provinces
Beta-Convergence - EU27 regions
NUTS3 – 1995/2010
0
-.5
0
.5
.5
1
1
1.5
1.5
NUTS2 – 1995/2010
8
8.5
9
9.5
lngdp1995
growth1995_2010
10
Fitted values
-0.2239435***
Source: Lorenzo Ferrari (2013).
10.5
8
9
10
lngdp1995
growth1995_2010
Fitted values
-0.155746***
11
β -Convergence - EU27 regions
Estimation of the coefficient for the
single observations sorted from the
poorest to the richest.
Recursive Rolling
Regression
NUTS2
DIVERGENCE
NUTS3
CONVERGENCE
DIVERGENCE IN THE CASE OF THE FIRST OBSERVATIONS!
Source: Lorenzo Ferrari (2013).
8
9
10
11
0
-.2
FINANCIAL
-.1
NUTS2
0
.2
.4
NUTS2
.6
.1
.8
.2
β -Convergence - EU27 regions
9
9.5
lngdp2000
growth2000_2007
10
10.5
lngdp2007
growth2007_2010
Fitted values
11.5
Fitted values
-0.0520234***
-.5
0
0
NUTS3
.4
CRISIS
.2
NUTS3
.6
.5
.8
1
1
-0.1961544***
11
8
9
10
lngdp2000
growth2000_2007
11
12
Fitted values
-0.1658848***
Source: Lorenzo Ferrari (2013).
8
9
10
lngdp2007
growth2007_2010
11
Fitted values
-0.0558489***
12
β -Convergence -New Entrants
Regions
•
•
•
•
•
•
•
•
•
•
•
•
Source: Lorenzo Ferrari (2013).
Bulgaria
Cyprus
Czech Republic
Estonia
Hungary
Latvia
Lithuania
Malta
Poland
Romania
Slovakia
Slovenia
β -Convergence -New Entrants
Regions
NUTS3 – 2000/2010
0
.2
.4
.5
.6
1
.8
1
1.5
NUTS2 – 2000/2010
8
8.5
9
lngdp2000
growth2000_2010
9.5
Fitted values
-0.1918162***
10
7.5
8
8.5
9
lngdp2000
growth2000_2010
9.5
Fitted values
-0.1977416***
• Lower coefficient than EU27  Capital vs. Rural regions.
• The process of transition affected the quality of data and the
significance of the regression from the 1995 to the 2010.
Source: Lorenzo Ferrari (2013).
10
The effect of being a New
Entrant on growth
Introducing the dummy variable it is possible to
evaluate the contribution that belonging to a New
Entrant has on growth.
The results show a positive effect on
growth from the 2000 to the 2007.
0.2370913***
Source: Lorenzo Ferrari (2013).
This positive coefficient however
significantly reduces after the crisis of
the 2007.
0.0711268***
Beta-Convergence – Regions of the
Founders of the EU
• Belgium
• France
• Germany
• Italy
• Luxembourg
• Netherlands
Source: Lorenzo Ferrari (2013).
β -Convergence – Regions of the
Founders of the EU
NUTS3 – 2000/2010
0
-.2
.1
0
.2
.2
.3
.4
.4
.6
NUTS2 – 2000/2010
9
9.5
10
lngdp2000
growth2000_2010
10.5
Fitted values
11
9
9.5
growth2000_2010
0.0075194
NOT SIGNIFICANT
Source: Lorenzo Ferrari (2013).
10
lngdp2000
10.5
11
Fitted values
-0.0321639**
There is no evidence
of any convergence
between these
regions.
Mixed growth
performances
The effect of being a
Founder on growth
Introducing the dummy variable «eu_found» it is
possible to estimate the the contribution that being a
founder has on growth.
The results show a negative effect on
growth from the 2000 to the 2007
This effect becomes however slightly
positive in the afternight of the
financial crisis
-0.0361925***
0.0268983***
Good performances of Germany and absence of the data for
Italian provinces.
Source: Lorenzo Ferrari (2013).
β -Convergence - Regions of
the PIGS
• Greece
• Ireland
• Spain
• Portugal
Only NUTS3 level
Source: Lorenzo Ferrari (2013).
β -Convergence - Regions of the
Poor Four
NUTS3 – 2000/2010
.6
.8
Lower coefficient than EU27
-.2
0
.2
.4
• These regions have a
more homogeneous
distribution of per-capita
GDP.
8.5
9
9.5
lngdp2000
growth2000_2010
10
Fitted values
-0.1559148***
Source: Lorenzo Ferrari (2013).
10.5
• The most important
contribution to
convergence came from
the New Entrants.
β -Convergence - Regions
of the PIGS
Introducing the dummy variable «poor» it is possible to
estimate the the contribution that belonging to a Poor
country has on growth.
The results suggest a positive effect
when the years from the 1995 to the
2010 are considered
It appears that belonging to a Poor
region becomes a disadvantage when
the estimation refers to the 20002010 time interval
0.0510249***
-0.0402644***
This could be explained remembering how these country experienced a
sustained growth until the explosion of the financial crisis.
Source: Lorenzo Ferrari (2013).
σ-Convergence across
countries
Reduction
starting from
very high levels.
Slight
increase
.
Slight
reduction
after a peak
in 1999.
The New Entrants are
slightly converging but
disparities between
regions are still enormous
The EU funders are
slightly diverging: some
regions have grown
continuously, others
lagged behind
Substantial
Stability.
Financial Crisis
Source: Lorenzo Ferrari (2013).
The financial crisis has
not any appreaciable
effect on dispersion.
σ-Convergence - regions within
the same country (1)
EU FOUNDERS
There is evidence of a
slight convergence
between the regions
of the EU founders
for what regards the
dispersion of their
per-capita GDP.
Source: Lorenzo Ferrari (2013).
σ-Convergence - regions within
the same country
NEW ENTRANTS
The regional
dispersion of the
per-capita GDP is
very high and
generally increasing
for the countries
that recently joined
the Union.
Most of the percapita GDP is
concentrated in
the capital region.
Source: Lorenzo Ferrari (2013).
σ-Convergence - regions within
the same country
POOR FOUR
Mixed Evidence
Source: Lorenzo Ferrari (2013).
Conclusions
• European Regions appear to have converged both
from the point of view of the growth of their percapita GDP and its regional dispersion.
• This result however masks some country-specific
trends and dynamics and the effects of the crisis
that risks to jeopardise this process.
Source: Lorenzo Ferrari (2013).
Conclusions
• The eastern enlargements altered the allocation of
the European budget for social and territorial
cohesion.
Future
enlargements
will
further
complicate the puzzle.
• The existence of a core-periphery pattern in Europe
makes cohesion spending necessary for ensuring a
balanced growth of the peripheral regions.
Source: Lorenzo Ferrari (2013).
Practical Session
• We use a dataset downloaded from EUROSTAT
http://ec.europa.eu/eurostat/data/database
• For β-convergence we use STATA
• .do file
• dataset
• For σ -convergence we use EXCEL