Download Effects of smart specialization on regional economic resilience in EU

Effects of smart specialization on regional economic resilience in EU
Vinko Muštra, PhD
Assistant professor
Faculty of Economics Split, University of Split
Cvite Fiskovića 5, Split, Croatia
[email protected]
Blanka Šimundić, PhD
Senior teaching and research assistant
Faculty of Economics Split, University of Split
Cvite Fiskovića 5, Split, Croatia
[email protected]
Zvonomir Kuliš, MA
[email protected]
ABSTRACT
The term regional economic resilience has became popular due to the recent global financial
crisis, especially in order to understand how local and regional economies are faring due to
the consequences of recent economic crises (e.g. Bristow, 2010; Hassink, 2010;
Christopherson et al, 2010; Simmie & Martin 2010; Martin 2012; Martin and Sunley, 2014,
Capello et al., 2015). Although regional economic resilience has been in focus of the
researcher recently, the literature offers ambiguous or no answers on several crucial question
related with this phenomena. In first place, up to the knowledge of the authors, there is lack of
the empirical investigation on the determinants of the regional economic resilience.
Therefore, this paper establishes empirical model focusing on the role of smart specialization
for regional economic resilience. The motivation can be found in paper by Martin and Sunley
(2014) and in paper by McCann and Ortega-Argilés (2015). In first paper regional economic
resilience is “characterized by a fuller and more productive use of its physical, human and
environmental resources” indicating the importance of effective and efficient use of the
regional potentials. At same time this has been recognized as significant characteristic of the
smart specialization concept which emphasizes issues of economic potential, and the
mechanisms whereby such potential is most likely be realized (p. 1293, McCann and OrtegaArgilés, 2015). Therefore it is likely to indicate possible importance of the smart
specialization for regional economic resilience.
Empirical analysis is performed for NUTS 2 regions in EU 27 based on a panel data approach
on EUROSTAT data for a 14-years period from 2001 to 2014. The dynamic panel data model
is formed by including the control variables based on the paper Martin and Sunley (2014) and

e-mail address of the correspondent author
1
by taking into consideration empirical concerns from the papers by Rodriguez-Pose et al
(2014) and Sensier et al. (2016). The results are indicating that smart specialization has
significant and positive effect on regional economic resilience.
This research is important because it provides for the first time investigation of the
relationship between smart specialization and regional economic resilience. It indicates
possible channels through which smart specialization might affect regional economic
resilience and provides empirical evidence on the effect of smart specialization on regional
economic resilience.
Keywords: regional resilience, smart specialization, EU
JEL classification: R10, R11 O31, B52
2
1. INTRODUCTION
The term regional economic resilience received substantial interest lately, in first place to
understand how local and regional economies are faring due to the consequences of recent
economic crises (e.g. Bristow, 2010; Hassink, 2010; Christopherson et al, 2010; Simmie &
Martin 2010; Martin 2012; Bristow & Healy 2014; Martin and Sunley, 2014). However,
recent economic crisis is not the only reason why researches are focusing on the concept of
resilience recently. Martin (2012) indicates that, beside the recent financial and economic
crises, there are at least three additional reasons: a) the impact of natural and man-made
disasters that have afflicted local communities; b) the influence of other disciplines, such as
ecology, where the main interest is in how ecosystems respond to shocks and c) recognition
that major disruptions can affect the whole economic landscape. Evidently, resilience is
multidimensional phenomena that can be defined, used and managed differently, according to
different objectives (Modica and Reggiani, 2015).
The present paper tries to synthesize the knowledge from several different sources, in order to
empirical investigate the determinants of regional economic resilience, in first place role of
smart specialization as one of the cornerstones of the new place-based approach
characterising regional development interventions in the EU ( Rodriguez-Pose et al., 2014)
The paper is structured as follows: in Section 2 we discuss and provide the various definitions
of resilience and their characteristics and links with the smart specialization. Section 3
introduces the empirical investigation the role of smart specialization for regional economic
resilience for NUTS II regions in EU during the period 2001-2014. Finally, section 4
concludes the paper with some recommendations for future research, highlighting the need for
more empirical studies on clarifying the determinants of regional economic resilience.
3
2. LITERATURE REVIEW
We start this section by offering review of the main definitions of regional economic
resilience, in order to provide the theoretical-methodological background for investigation the
role of smart specialization for regional resilience.
Modica and Reggiani, (2015) stress that common foundation for all definitions of resilience,
and therefore also of the regional economic resilience, is it etymology. More precisely, the
word resilience comes from Latin “resilire” which means to rebound or leap back and
therefore researchers try to imbed in their definitions (e.g. Martin, 2012). Also, it should be
stressed that term resilience has been introduced in other sciences long time ago, in first place
in physics and ecology, and this has led to first differentiation in definition (Modica and
Reggiani, 2015).
Same authors (Modica and Reggiani, 2015) citing papers by Markusen (1999) and Swanstrom
(2008) indicate that in economic science resilience has been introduced as a fuzzy concept or
just conceptual framework. However, due to already mentioned reasons, resilience in
economics and especially in regional economics received huge interest of the scientists
recently (Christopherson et al, 2010; Simmie & Martin 2010; Martin 2012; Bristow & Healy
2014; Martin and Sunley, 2014, Modica and Reggiani, 2015 ).
Literature offers several different definitions of the regional economic resilience that has been
widely explained in the papers by Martin and Sunley (2014) and by Modica and Reggiani
(2015) . First definition lies on the ‘bounce back’ version of the resilience concept. This
concept is founded in the paper by Holling (1973) in which resilience is examined as system’s
speed of recovery or return to its pre-shock position. Considering that this definition is mostly
used in many physical and engineering sciences, it has been called ‘engineering resilience’.
This concept implies “return to normal” and the “normal” can be interpreted as identifiable
long-run, path-dependent developmental trajectories. This definition and ability of systems to
maintain or regain stability after external perturbations and disturbances has been challenged
by implementing new concept defined as ‘ability to absorb“ implying resilience as the “the
capacity of a system to absorb disturbance and reorganize while undergoing change so as to
still retain essentially the same function, structure, identity and feedbacks” (Walker et al,
2006, p 2.). However, as Martin and Sunley (2014) indicate “this definition is not without
ambiguity, since it remains unclear just how much ‘reorganisation’ and ‘change’ is permitted
for the system to be regarded as still having ‘essentially the same structure, identity and
feedbacks’. This doubts offered space for the new concept named „adaptive resilience’
4
defining resilience as “the ability of the system to withstand either market or environmental
shocks without losing the capacity to allocate resources efficiently” (Perrings, 2006, p.418),
This idea of resilience has led some authors to refer to it as ‘evolutionary resilience’, defined
in terms of ‘bounce forward’ concept rather than ‘bounce back’ (Simmie and Martin, 2010).
Taking three main interpretations into consideration it is obvious that defining regional
economic resilience is not straightforward. However, Martin and Sunley (2014) introduce the
definition that tackles all previous mentioned challenges. They define regional economic
resilience as the capacity of a regional or local economy to withstand or recover from market,
competitive and environmental shocks to its developmental growth path, if necessary by
undergoing adaptive changes to its economic structures and its social and institutional
arrangements, so as to maintain or restore its previous developmental path, or transit to a new
sustainable path characterized by a fuller and more productive use of its physical, human and
environmental resources. As the Martin and Sunley (2014) underline, this definition has five
crucial elements: vulnerability (the sensitivity or propensity of a region’s firms and workers to
different types of shock); shocks (the origin, nature and incidence of a disturbance, and the
scale, nature and duration thereof), resistance (the initial impact of the shock on a region’s
economy); robustness (how a region’s firms, workers and institutions adjust and adapt to
shocks, including the role of external mechanisms, and public interventions and support
structures); and recoverability (the extent and nature of recovery of the region’s economy
from shocks, and the nature of the path to which the region recovers). All these five elements
indicate that regional economic resilience is linked with the regional (economic)
characteristics, by shaping and by being shaped by the reaction of the region’s economy as it
has been already indicated in the paper by Simmie and Martin (2010).
Considering that regional (economic) characteristics are noticeably important for fully
understanding the regional resilience phenomena and that the literature offers ambiguous
answers on this matter, the paper tries to identify and empirical investigate the influence of
some specific feature of regional structure on regional economic resilience. In other words,
we try to empirical investigate what makes some regions “characterized by a fuller and more
productive use of its physical, human and environmental resources”, Martin and Sunley
(2014).
The “fuller and more productive use of resources” has been recognized as significant attribute
of the smart specialization concept which emphasizes issues of economic potential, and the
mechanisms whereby such potential is most likely to be realized (p. 1293, McCann and
5
Ortega-Argilés, 2015). By taking into consideration the characteristics of smart specialization
and regional economic resilience it is likely to indicate possible importance of the smart
specialization for regional economic resilience.
The development of smart specialization concept started within the broad empirical literature
examining the transatlantic productivity gap, the productivity gap between the US and
Europe, that has been evident since 1995 (van Ark et al. 2008). The empirical analysis of the
factors which have generated these differences was especially related to the differences in
labour markets (Gu et al., 2002; Gu and Wang, 2004; Bloom et al., 2005; Gordon and DewBecker, 2005; Gomez-Salvador et al., 2006; Crespi et al., 2007) and the differences in
industrial performance (McCann and Ortega-Argilés, 2015). The differences in industrial
performance have been explained via two broad themes. The first one originating from the
differences in industrial structure where according to Mathieu and van Pottelsberghe de la
Potterie (2008) European traditional, middle and low-tech sectors are far more ineffective in
translation of R&D in productivity gains. The second theme goes deeper in sectoral analysis
finding out that even within the same sectors European firms exhibit a lower ability to
translate R&D into productivity gains or other type of investment (Erken and van Es, 2007).
The common theme that has emerged from these explanations is that information and
communication technologies (ICTs) play a critical role in productivity gap explanations
(McCann and Ortega-Argilés, 2015). Finally, revealed technological disadvantage of EU
relative to the US has been seen in ICT-related explanations, proxied by ICT-based R&D
investment (Foray, David and Hall, 2009). Nevertheless, the shift from the role of R&D
intensity to the role of dissemination of new technologies across the wider economy in
explaining the growth potentials has emerged recently in the literature (McCann and OrtegaArgilés, 2015). The rationale for smart specialisation strategy is seen in encouraging
investment in programs that will complement the country’s other productive assets to create
future domestic capability and interregional comparative advantage (Foray, David and Hall,
2009). Thus the importance of adoption, dissemination and the adaptation of new ICTs across
the wider economy has been emphasized (Jorenson, 2001; Jorgenson et al., 2005; Stiroh,
2002; Timmer and van Ark, 2005; O'Mahoy and van Ark, 2003; Gordon, 2004; Draca et al.,
2006; Wilson, 2009). This resulted in the adaptation of smart specialisation concept originally
developed in sectoral dimension to the regional dimension where the straightforward shift
from smart specialisation concept to a regional context is quite difficult and rather more
complex that it at first appears (McCann and Ortega-Argilés, 2015). Rodriguez-Pose et al.
6
(2014) explain that the development of smart specialisation concept on a regional basis is
based on the recognition that successful development and innovation strategies cannot be
replicated mechanically in each and every different regional context. They also argue that
rather than trying to generate ‘technology miracles’ in economically disadvantaged areas, the
best way to close the gap between less innovative regions and technology hubs is to try to
identify the unique assets that make the potential for innovation in a peripheral region
(Rodriguez-Pose et al., 2014). According to Foray, David and Hall (2009) national public
policy has an important role in supporting and accompanying emerging trends in smart
specialisation. The role is seen in supplying incentives to encourage entrepreneurs and other
organizations (higher education, research laboratories) to become involved in the discovery of
the regions’ respective specialisations. Thus, in contrast of the traditional linear model of
devoting more resources into R&D, research and innovation strategies for smart specialisation
(RIS3) require a much higher effort in planning, as policy measures are not transposed from
the best performers, but they are the result of a careful examination of a region’s weaknesses
and potential strengths (Rodriguez-Pose et al., 2014). Consequently the initialisation of the
usual policy tools to support R&D and innovation should begin after depicting of the potential
of a particular technology to regenerate the targeted economic domain (production or
services) through the co-invention of applications, and after examining whether the size of
this domain is large enough (the size refers here not to GDP but to the size of the relevant
sectors in the economy, that is, those sectors that could potentially benefit from the
knowledge spillovers from the initial development of applications (Foray, David and Hall,
2009). Finally, according to McCann and Ortega-Argilés (2015) an increasing emphasis on
enhancing the linkages between knowledge generation processes in all their forms (including
R&D) and the promotion and dissemination of entrepreneurship and innovation across all
sectors, activities and occupations within the context of global value chains is present
nowadays in regional and smart specialisation development strategies.
Obviously, by taking into consideration the attributes of smart specialization and regional
economic resilience it is likely to indicate possible importance of the smart specialization for
regional economic resilience. However, although literature indicate the possible link between
smart specialization and regional economic resilience, theoretical and especially empirical
evidence are quite limited. Moreover, for the analyzing the effects of the smart specialization
on regional economic resilience, up to the knowledge of the authors, there is no (empirical)
investigations that tackles this issue.
7
3. EMPIRICAL RESEARCH
In the empirical part of the paper the authors introduce the empirical model that tries to tackle
the relationship between the smart specialization and regional economic resilience.
Data and methodology
Before explaining the results of the empirical analysis, we will first introduce the data and
methodology.
Data for GDP, education, infrastructure has been collected from QoG EU Regional dataset
(Charron et al., 2016), labor force participation and data for calculation of specialization
indexes were collected from Eurostat database. Data for governance indicators were collected
from World Bank dataset (WGI, 2015). Data covers period from 2001 till 2014.
Considering the goal of the paper, two group of variables are particularly important for this
study: measure of regional economic resilience and smart specialization. For selecting
indicators for these two variables and for other control variables we have considered the
papers by Martin and Sunley (2014), Rodriguez-Pose et al (2014), Moddica and Reggiani
(2015) and Sensier et al. (2016).
Taking into consideration that there is no ambiguous definition of regional economic
resilience indicates the problematic aspect of measuring this phenomenon. The literature
offers several different ways, ranging from descriptive, interpretative case studies to
econometric models (e.g. Simmie and Martin, Martin, 2012, Fingleton et al.,2012, Cowell,
2013, Sensier et al., 2016). Each of this methods and approaches has a limitation(s) and in
principle there is no reason why different methods could not be combined (Martin and Sunley
(2014). However, descriptive methods and approaches could not be used effectively to
measure and compare resilience across different regions in EU and therefore we focus on
more statistical measures.
More precisely, in panel data model we use as an indicator for dependent variable (regional
economic resilience) sensitivity index (RES)1 formula introduced in papers by Martin (2012)
and Finleton et al. (2012). The sensitivity index has been usually implemented on
employment or GDP data. Considering that literature indicates that employment is a more
meaningful measure than GDP as it counts the number of people employed in a region and is
1
In previous studies it is marked as βrAlso, in studies by Martin (2012) and Finleton et al. (2012) they compare
regional employment lost comparing to the national average lost. 8
less prone to revision (see Coyle, 2014 and Sensier et , 2016), we adopt the Sensitivity index
so that it gauges the percentage change in employment lost in a region (Er) compared with the
EU-27 average (EEU‐27). Thus, formula for our Sensitivity index goes as follows:
∆
(1)
∆
In equation (1) ∆
while
stands for employment loss in region i in period t compared to period t-1
is employment in region i in period t. The symbol ∆EU‐27 stands for employment loss
in EU 27 in period t compared to period t-1, while EU 27 represents employment in period t.
Second important variable in our empirical research is smart specialization (SS). For defining
the proxy for smart specialization we have decide to follow approach implemented in paper
by Rodriguez-Pose et al. (2014) where the natural logarithm of patents application filled to the
European Patent Office (EPO) per million inhabitants in region is proxy for SS. For robustness
check, as indicator it is also used natural logarithm of patent applications to the EPO, per million of
active population.
Taking assumption that regional economic resilience is influenced only by the smart
specialization is rather restrictive and results can potentially suffer from the omission of other
(possibly) significant determinants. Thus we test whether the relationship between regional
economic resilience and smart specialization holds when we include additional explanatory
variables. To do so, we firstly examined the relevant literature, or more precisely, we
considered all possible determinants indicated in papers by Fingleton and Palombi (2013),
Brakman et al (2014), Di Caro (2014), Diodato and Waterings (2014) and Martin and Sunley
(2014).
It would be ideally to include all potential determinants indicated in the abovementioned
literature. But it isn’t straightforward task. Firstly, regional data on these aspects are rarely
available and/or of poor quality and therefore we had to be focus on a limited number of
variables. In addition, considering that this paper represents unique empirical study which
analyses the importance of smart specialization for regional economic resilience, it was not
possible to follow similar variable selection procedure.
Taking into consideration this limitations, we have consider the following variables: level of
GDP, labor force participation, education, institutional quality, infrastructure and
specialization indexes. Indicator for level of GDP (GDPpc) is gross domestic product per
9
capita (GDP at current market prices, PPS per inhabitant). As proxy for labor force
participation (PART) is used labor force participation rate (or economic activity rate population from 15 to 64 years) while education (EDU) is approximated by percentage of
people with tertiary education in population between 25 and 64 years. Indicator for
infrastructure (RAIL) is total railway lines (kilometre/1000 square km).
At region level there is no database on regional institutional quality that would be completely
appropriate for our panel analysis for period 2001-2014. Therefore, we decided to implement
commonly used proxy for institutional quality at national level, World Governance Indicators
(WGI) provided by World Bank. We adopt national WGI by calculating the average of the
values for six dimensions of WGI2, and this values has been presented as proxy for
institutional quality at regional level in that country (for example, Austrian WGI is assigned to
all Austrian regions in the analyse).
For the regional economic resilience, specialization can been also indicated as a significant
variable and therefore we have introduce the index of specialisation (Sij) expressed as follows
(European Commission, 2012):
∑
∑
∑
∑
where i refers to the region, j to the sector and E stands for employment in period t. In case
that region i has higher level of employment concentration in sector j than on EU27 average,
implies that this region is specialized in this sector and that value of this index is higher than
one. The sectors, or sector groups, chosen to define regional specialisation are chosen by
NACE classification3.
Finally, we assumed that regional economic resilience is dynamic relationship which means
that its current value depends on its past values. Thus we introduce modified equation which
includes dynamic behavior of dependent variable characterized by the presence of lagged
dependent variable among the repressors:
2
Voice and Accountability (VOI), Political Stability and Absence of Violence (POL), Government Effectiveness
(GOV), Regulatory Quality (REG), Rule of Law (LAW) and Control of Corruption (COR)
3
Agriculture, hunting, forestry and fishing (AGR), Industry (IND), Construction (CON), construction (CON),
distributive trades, transport, accommodation and food services; information and communication services;
financial and insurance services; real estate activities; professional, scientific, technical, administrative and
support services (market services) (DIS), public administration, defence, education, human health and social
work; arts, entertainment, recreation, other services and activities of household and extra-territorial organisations
and bodies (public administration and other services) (PUB).
10
RESit = µ + γRESi.t‐1 + β1SSit + β2GDPpcit + β3EDUit + β4PARTit + β5WGIit + β6RAILit + SIitδ + αi + εit
(2)
where i stands for region i and t is one year period. µ is an intercept, γ is a parameter of
lagged dependent variable and β1, β2,…, β6 are the parameters of exogenous variables. SIit =
[SIit1, SIit2…SIitK] where 1xK is matrix of control variables of specialization indexes. δ is Kx1
vector of parameters. It is assumed that εit are IID(0,σ ). αi is unobservable individual-specific
effect that is time invariant and it accounts for any individuals.
This analysis poses several advantages. For example, Pablo-Romero and Molina’s (2013)
highlight that panel data methodology usage allows larger number of explanatory variables,
larger sample of countries, longer time periods under analysis and greater depth in the
relationships between variables. Furthermore, Seetaram and Petit (2012) specify that one of
the most important advantages is that panel data modelling allows for the control of
heterogeneity in the sample. However, in order to control heterogeneity even better, all
variables that were not given in relative ratios, we made their logarithm (except WGI which is
special as is given in mark between -2.5 and 2.5 and it is impossible to make negative
logarithms).
Descriptive statistics presented in the Table 1 should provide intuition and foundation for
understanding our empirical investigation. It can be noticed that regional economic resilience
measured by Sensitivity index (RES) has mean of -3.25 and enormous standard deviation
(136.22) which suggests that this variable has high variations.
Table 1: Descriptive statistics
Variable
Mean
Std. Dev.
Min
RES
-3.252076
136.2204
-2699.88
SSmil
98.00758
122.316
0.018
SSact
204.6432
382.0706
.05
GDPpc
23797.21
10988.56
3900
EDU
24.37204
9.735761
6.2
PART
70.6038
6.456295
46.3
WGI
1.119955
0.4576923
-0.046492
RAIL
66.71371
74.78184
0
AGR
1.19039
0.7218024
0
IND
1.004222
0.3188535
0.1200788
CON
1.037063
0.1829671
0.4695925
TRA
0.9605747
0.1994347
0.5803223
ADM
1.026421
.1828255
.6588255
Source: compiled by the authors using software Stata 13.0
Max
1692.08
1018.935
3341.41
148000
61
85.3
1.98539
708
5.34161
1.947419
2.081549
2.607515
2.286818
i
N
3662
3050
3102
3800
3609
3710
3770
1970
1907
1907
1907
1644
1644
290
268
268
273
267
290
290
180
220
220
194
194
194
To test multicollinearity, as suggested by Baltagi (2008) we present pair wise correlation
matrix in Table 2. Gujarati and Porter (2008) explain that problem of multicollinearity exists
11
when Pearson's correlation coefficients between independent variables in the model exceed
the 0.8 threshold (it is only correlation of 0.9201 between two smart specialization indicators,
but which are analyzed in separated models anyway). As presented in Table 2, all other
coefficients are lower than critical value of 0.8. and therefore it is possible to continue with
the empirical analysis.
Table 2: Pair wise correlations matrix
Variable
RES
L2.RES
L.logSSm
L.logSSa
1.0000
RES
-0.0101
1.0000
L2.RES
0.1183
0.1144
1.0000
L.logSSm
0.1248
0.1152
0.9201
1.0000
L.LlogSSa
0.0706
0.0904
0.7964
0.6989
L.logGDPpc
0.0406
0.0627
0.5580
0.4691
L.PART
0.0214
0.0129
0.3397
0.3168
L.EDU
0.1126
0.1342
0.7470
0.6020
L.WGI
0.0829
0.0461
0.1927
0.2572
L.logRAIL
-0.0513
-0.0574
-0.0665
-0.1675
L.AGR
0.0259
0.0134
0.1486
0.1738
L.IND
-0.0643
-0.0597
0.0058
-0.0816
L.CON
0.0107
0.0113
0.0933
0.1309
L.TRA
0.0202
0.0280
-0.0098
-0.0351
L.ADM
Variable
L.WGI L.logRAIL
L.AGR
L.IND
1.0000
L.WGI
0.0921
1.0000
L.logRAIL
0.0178
-0.5012
1.0000
L.AGR
0.0274
-0.1067
0.1050
1.0000
L.IND
0.0617
-0.1273
0.2714
0.0268
L.CON
-0.0103
0.4407
-0.5330
-0.3866
L.TRAD
0.0077
0.1234
-0.2456
-0.4458
L.ADM
Source: compiled by the authors using software Stata 13.0
L.logGDPpc
L.PART
L.EDU
1.0000
0.5592
0.4303
0.6067
0.1943
-0.1907
-0.0552
-0.0062
0.3375
0.0167
L.CON
1.0000
0.4649
0.6443
0.0704
-0.0392
0.1076
-0.0197
0.1446
-0.2509
L.TRA
1.0000
0.4060
-0.0460
-0.1520
-0.0603
-0.1044
0.1970
0.0484
L.ADM
1.0000
0.0793
0.0430
1.0000
0.0492
1.0000
Blundell and Bond (1998) GMM (generalized methods of moments) is an improved version
of Arellano and Bond (1991) GMM estimator and is the appropriate choice considering our
sample characteristics (data sample containing more cross than time observations and big
number of regions). Blundell and Bond two step estimator was chosen over one step estimator
because the latter assumes the error terms to be independent and homoscedastic across
countries and over time while two step estimator relaxes the assumption of independence and
homoscedasticity.
Finally, following Blundell-Bond (1998), panel data model with lagged variables will be
estimated in following manner:
RESit =µ+γ1RESi.t‐2 +γ2RESi.t‐1 +β1logSSi,t‐1+β2logGDPpci,t‐1+β3EDUi,t‐1+β4PARTi,t‐1+β5WGIi,t‐1+
β6logRAILi.t‐1+SIi,t‐1δ+αi+εit
(3)
The model in equation (3) will be tested through 2 different options. In both models all
previous mentioned control variables are included (level of GDP - GDPpc, labor force
12
participation - PART, education - EDU, modified institutional quality - WGI, infrastructure
quality - RAIL and specialization indexes - SI).
Difference between two options is our proxy variable for smart specialization. Option 1
includes patent per million inhabitant as indicator of SS while in Option 2, as additional
robustness check, indicator for SS is per million of active population.
Table 3: Estimation Results (Blundel and Bond GMM System Estimator) for model of
regional resilience
Cons.
L.RES
L2.RES
L.logSSinh
L.logSSact
L.logGDPpc
L.EDU
L.PART
L.WGIE
L.logRAIL
L.AGR
L.IND
L.CON
L.DIS
L.ADM
Number of observations
Number of groups
Number of instruments
Sargan test (P‐value)
m1‐test (P‐value)
m2‐test (P‐value)
Standard errors in parentheses
* p < 0.1, ** p < 0.05, *** p < 0.01
Source: compiled by the authors using software Stata 13.0
Option 1
-70.59
-0.0693***
-0.0735***
-4.889***
-47.04***
-0.817*
-7.235***
95.08***
-18.55***
167.3***
319.1***
14.08
505.9***
22.06
908
138
88
0.2689
0.0000
0.7968
Option 1
-59.31
-0.0686***
-0.0727***
-5.014***
-47.90***
-0.785*
-7.221***
94.11***
-18.67***
165.5***
320.0***
14.07
508.2***
20.71
908
138
88
0.2668
0.0000
0.7819
Before we introduce the empirical results we test the validity of models, or to be more
precisely, we implement Sargan test and tests for serial correlation (Table 3.) The results of
tests (in Table 3) indicate that model is well specify.
Finally, taking into consideration the empirical results presented in Table 3, it is first possible
to notice that all independent variables have significant impact on regional resilience as all
variables are statistically significant on significance level of at least 5%. The only exceptions
are two specialization indexes (for construction and for administration). Higher level of initial
GDP, education, labor force participation and infrastructure quality has negative effect on the
sensitivity index and therefore positive impact for regional economic resilience. At same time,
institutional quality has negative impact, but this result should be considered taking into
consideration that institutional quality has been proxied with the value of national institutional
quality (WGI).
13
Smart specialization (SS), as our variable of special concern, has negative effect on the
sensitivity index (RES) in both options. In other words, higher level of implementation of
smart specialization decreases sensitivity of regional economy and therefore increases
regional economic resilience. This results indicates that smart specialization allows “fuller
and more productive use of resources” that is crucial for regional economic resilience. The
intuition for this positive impact of smart specialization on regional economic resilience still
remains fuzzy or unexplored. However, this empirical analysis doesn’t not have a goal to
provide and explain the theoretical background for this relationship, but just to provide the
empirical evidence that may stimulate research to pay more attention on theoretical
explanation for possible channels through which smart specialization might affect regional
economic resilience.
This could lead policy makers to better understanding of regional economic resilience and
therefore make regional economic resilience more robust by delivering more effective
regional policy measures.
14
4. CONCLUSION
This research is important because it provides for the first time investigation of the
relationship between smart specialization and regional economic resilience. It offers empirical
evidence of the influence of the smart specialization on regional economic resilience, or to be
more precisely, the results indicate higher level of implementation of smart specialization
decreases sensitivity of regional economy and therefore increases regional economic
resilience.
The results of this research should firstly stimulate research to pay more attention on
theoretical background of this empirical relation. After that, it should help policy makers to
predict and boost positive effects of smart specialization on regional economic resilience and
therefore make regional economic resilience more robust. Namely, by identifying the
significant effect of smart specialization on regional economic resilience it could help in
delivering more effective regional policy measures.
At same time, considering that the overall magnitude of the effects generated by the recent
economic crisis is not yet fully understood or even less clear, our investigation may stimulate
research to pay more attention to the spatial distribution of the costs and the impact of the
different type of crisis.
15
LITERATURE:
1. Arellano, M. & S. Bond (1991), "Some tests of specification for panel data: Monte
Carlo evidence and an application to employment equations", The Review of
Economic Studies, 58(2), pp. 277-297.
2. Baltagi, B. (2008), Econometric analysis of panel data. Chichester: John Wiley &
Sons Gujarati, D. N. & D.C. Porter (2008), Basic Econometric, Vol 5. New York:
McGraw-Hill/Irwin
3. Blundell, R. &. S. Bond (1998), ''Initial conditions and moment restrictions in
dynamic panel data models''. Journal of econometrics, 87(1), pp. 115-143.
4. Brakman, S. & Harry Garretsen & Charles van Marrewijk, 2014. "The Crisis and
Regional Resilience in Europe: On the Importance of Urbanization and
Specialization," CESifo Working Paper Series 4724, CESifo Group Munich.
5. Bristow, G. (2010) Resilient Regions: Re-place-ing Regional Competitiveness,
Cambridge Journal of Regions, Economy and Society, 3, 1, pp. 153-167.
6. Capello, R, Caragliu, A & Fratesi, U 2015, ‘Spatial heterogeneity in the costs of the
economic crisis in Europe: are cities sources of regional resilience?’, Journal of
Economic
Geography,
Available
from:
http://joeg.oxfordjournals.org/content/early/2015/01/20/jeg.lbu053
7. Charron, Nicholas, Stefan Dahlberg, Sören Holmberg, Bo Rothstein, Anna Khomenko
& Richard Svensson. 2016. The Quality of Government EU Regional Dataset, version
Sep16. University of Gothenburg: The Quality of Government Institute,
http://www.qog.pol.gu.se
8. Christopherson, S., Michie, J. and Tyler, P. (2010) Regional Resilience: Theoretical
and Empirical Perspectives. Cambridge Journal of Regions, Economy and Society, 3,
pp. 3-10.
9. Cowell, M. (2013) Bounce back or move on: regional resilience and economic
development planning, Cities, 30, pp.212-222.
10. Di Caro, Paolo, 2014. "Testing and explaining economic resilience with an application
to Italian regions," MPRA Paper 60298, University Library of Munich, Germany.
11. Diodato, D. and A.B. Weterings (2014), The resilience of regional labour markets to
economic shocks: exploring the role of interactions among firms and workers, Journal
of Economic Geography, J Econ Geogr (2014) doi: 10.1093/jeg/lbu030
12. Euroepan Comission. (2012). Analysis of the main factors of regional growth: An indepth study of the best and worst performing european regions. An Inception Report
for the European Commission, DG Regional Policy.
13. Eurostat (2016) regional database. The QoG EU
http://qog.pol.gu.se/data/datadownloads/qogeuregionaldata
Regional
dataset
14. FINGLETON B., GARRETSEN H. and MARTIN R. (2012) Recessionary shocks and
regional employment: evidence on the resilience of UK regions, Journal of Regional
Science 52, 109–133.
15. Fingleton, B. and S. Palombi (2013), Spatial panel data estimation, counterfactual
predictions, and local economic resilience among British towns in the Victorian era,
Regional Science and Urban Economics, 43: 649-660.
16
16. Griliches, Z. (1990) Patent statistics as economic indicators: a survey. Journal of
Economic Literature, 28: 1661-1707.
17. Hassink, R. (2010) Regional Resilience: A Promising Concept to Explain Differences
in Regional Economic Adaptability? Cambridge Journal of Regions, Economy and
Society, 3, 1, pp. 45-58.
18. Holling CS (1973) Resilience and stability of ecological systems. Annu Rev Ecol Syst
4:1–23
19. Martin, R.L. and Sunley, P. (2014) 'On the Notion of Regional Economic Resilience:
Conceptualisation and Explanation'. Journal of Economic Geography.
http://joeg.oxfordjournals.org/content/early/2014/07/02/jeg.lbu015.full
20. Markusen A (1999) Fuzzy concepts, scanty evidence, policy distance: the case for
rigour and policy relevance in critical regional studies. Reg Stud 37:701–717
21. McCann, P. & Ortega-Argilés, R. (2015) Smart Specialization, Regional Growth and
Applications to European Union Cohesion Policy, Regional Studies, 49:8, 1291-1302,
DOI: 10.1080/00343404.2013.799769
22. Martin, R 2012, ‘Regional Economic Resilience, Hysteresis and Recessionary
Shocks’, Journal of Economic Geography, vol. 12, no. 1, pp. 1–32.
23. Marianne Sensier, Gillian Bristow & Adrian Healy (2016) Measuring Regional
Economic Resilience across Europe: Operationalizing a complex concept, Spatial
Economic Analysis, 11:2, 128-151, DOI: 10.1080/17421772.2016.1129435
24. Modica, Marco & Reggiani, Aura (2015) "Spatial Economic Resilience: Overview and
Perspectives," Networks and Spatial Economics, Springer, vol. 15(2), pages 211-233,
June.
25. Pablo-Romero, M.P. & Molina, J.A. (2013), "Tourism and economic growth: A
review of empirical literature", Tourism Management Perspectives, 8, pp. 28-41.
26. Roodman, D. (2009), ''A note on the theme of too many instruments'', Oxford Bulletin
of Economics and statistics, Vol. 71 No. 1, pp. 135-158.
27. Rodriguez-Pose, A. & Di Cataldo, M. & Rainoldi, A.(2014) "The Role of Government
Institutions for Smart Specialisation and Regional Development," JRC-IPTS Working
Papers JRC88935, Institute for Prospective Technological Studies, Joint Research
Centre.
28. Seetaram, N. & S. Petit (2012), "Panel Data Analysis". In: Dwyer, L., Gill, A. and N.
Seetaram (eds.).Handbook of Research Methods in Tourism: Quantitative and
Qualitative Approach, pp.127-144. Cheltenham: Edward Elgar
29. Simmie, J & Martin, R 2010, ‘The economic resilience of regions: towards an
evolutionary approach’, Cambridge Journal of Regions, Economy and Society, vol. 3,
no. 1, pp. 27–43.
30. Swanstrom T (2008) Regional resilience: a critical examination of the ecological
framework. Working paper 2008–07. Institute of Government Studies, University of
California, Berkeley
31. World Bank (2015). Worldwide Governance Indicators (www.govindicators.org),
available at: http://info.worldbank.org/governance/wgi/index.aspx#home
17