Environmental performance, innovation and regional spillovers - - PowerPoint PPT Presentation

IEFE, Milan, february 11, 2011

Environmental performance, innovation and regional spillovers

Valeria Costantini (Università di Roma III) Massimiliano Mazzanti (Università di Ferrara e CERIS-CNR) Anna Montini (Università di Bologna e CERIS-CNR)

Research questions

Environmental performances (emission intensity, EM/VA),

the role of:

regional productive structures (sectoral) efficiency sectoral labour productivity sectoral technological innovation sectoral technological innovation technological/innovation spillovers environmental regulation environmental spillovers

Geo-framework:

Italian regions (20) 2 V.Costantini, M.Mazzanti, A.Montini - Environmental Performance and Regional Innovation Spillovers

Framed in Environmental Kuznets Curves / IPAT models (Martinez Zarzoso,

2009)

• Study of sectoral drivers, time effects
• Merge with innovation, trade, policy data
• Dynamics of (joint) economic and environmental productivities

Italian NAMEAs (1990-2007 for air emissions) are a good playground

• Mazzanti & Zoboli (2009), Ecological Economics
• Mazzanti, Montini and Zoboli (2008), Economic System Research

Mazzanti and Montini (2010), Ecological Economics

Sectoral decoupling analysis using NAMEA

Mazzanti, Montini and Zoboli (2008), Economic System Research

• Mazzanti and Montini (2010), Ecological Economics
• Marin and Mazzanti (2011), J. of Evolutionary Economics
• EU
• Femia, Moll, Watson (2006), shift-share in NAMEA EU + input-output + emissioni indirette
• Forthcoming: A full Eurostat EU27 NAMEA 2000-2006
• (Actually: Eurostat Namea x19 countries x12air emissions x80sectors

Studies using Reg-NAMEA are still less developed

(Stauverman, 2007)

In Italy:

E.g. recent applications to Lazio and Emilia Romagna (Bonazzi -

Sansoni, 2008; Mazzanti, Montini and Zoboli, 2007, EFEA; Mazzanti

Regional NAMEA analyses

Sansoni, 2008; Mazzanti, Montini and Zoboli, 2007, EFEA; Mazzanti & Montini, 2010, Ecological Economics)

Costantini, Mazzanti and Montini, 2010, FEEM, provide spatial and

innovation related analyses on ISTAT 2005 Regional NAMEA using shift-share and econometrics

ISTAT will release full 2000 REG-NAMEA by early 2011,

thus allowing panel based and spatial analysis

Methods

Shift-Share analysis (2005, 20 regions, 24 sectors, 10 pollutants)

to decompose the source of change of the emission intensity (EM/VA)

into:

regional specific productive structures/specializations (the share) the efficiency feature (the shift between regional and national efficiency) a covariance effect between the previous two a covariance effect between the previous two

Modelling emission intensity (cross-section 2005, 19 regions, 11 sectors,

GHG, ACID)

Emission-demand model to identify inn. and env. drivers Environmental spillovers Technological/innovation spillovers (patents based indicator) 5

Modelling emission intensity

r k r k r k r k r k r k

p t Y e ε β β β α + + + + =

− − − 3 2 1 ln

[adapted from Cole et al.2005; Medlock and Soligo (2001)] where:

• e kr, pollutant emissions for k-sector in the r-region scaled with

r k r k r k r k r k r k r k r k

p ts t es lp e ε β β β β β α + + + + + + =

− − − + − 5 4 3 2 1 6

• e kr, pollutant emissions for k-sector in the r-region scaled with

region/sector value added

• α kr, region and sector specific effects
• lp kr, labour productivity
• es kr, environmental spillovers
• t kr, internal (regional) technology
• ts kr, innovation spillovers (inter-regional and intra-sector)
• p kr, environmental policy
• ε kr, error term

Emissions’ intensities

(Main) Data

Italian regional NAMEA data, 2005 (Istat, 2009) 20 regions 24 productive sectors environmental pressures, 10 pollutants (direct emissions)

GHG (CO , N O, CH ) - globally distributed

• GHG (CO2, N2O, CH4) - globally distributed
• ACID (NOx, SOx, NH3) – local (neighb. regions)

economic data (ValAd, HouEx, FullTimeEJ) Electricity consumption by sector (TERNA) Indirect emissions (considered in the model)

7

Region CO2 Region SO

X

Trentino Alto Adige 136 Trentino Alto Adige 39 Campania 141 Valle d’Aosta 45 Valle d’Aosta 153 Abruzzo 69 Piedmonte 185 Campania 78 Lazio 204 Lomba rdy 99 Marche 206 Lazio 101 Lombardy 209 Marche 108 Abruzzo 258 Piedmonte 108 Veneto 267 Calabria 123 Emilia Romagna 270 Basilicata 224

CO2 and Sox emission intensity (kgx1M€ of value added, increasing order)

8

Emilia Romagna 270 Basilicata 224 Tuscany 278 Emilia Romagna 226 ITALY 301 Molise 276 Calabria 307 Veneto 300 Umbria 342 ITALY 315 Friuli Venezia Giulia 353 Tuscany 349 Basilicata 430 Umbria 373 Liguria 472 Friuli Venezia Giulia 539 Sicily 547 Puglia 859 Molise 689 Liguria 886 Sardinia 824 Sicily 1,347 Puglia 971 Sardinia 1,530

Innovation (spillovers)

The role of regional innovation (technological) spillovers

• tech. learning and knowledge spilloves have a centripetal force fostering

agglomeration patterns (Kyriakopoulou and Xepapadeas, 2009)

Domestic (internal) effect (t) Inter-regional intra-sector effect (ts)

Measure

Sectoral innovation intensities: patents (wo specific env. purposes – further research) to VA ratios

9

research) to VA ratios Five years average (2000-04) for patents by sector (proxy of the innovation stock at sectoral level) Geographical distances and economic structure similarity matter Localization economies associated with the concentration of a particular sector in the (neighbouring) regions

Data

• Patents (REGPAT-Eurostat from OECD PATSTAT)

ad hoc codification of IPC codes according to NACE (manufacturing) codes

Innovation spillovers (measure)

∑ ∑

= =

=

n k IT k IT k n k r k r k r k

t t t t RSI

1 1

The relative specialization index (RSI):

r k

t

ITk

t

∑

=

=

q r r k ITk

t t

1

where is the five-years average of patents to valued added ratios for each k-th sector and r-th region, while is the same measure at the national level, as 10 V.Costantini, M.Mazzanti, A.Montini - Environmental Performance and Regional Innovation Spillovers The bilateral innovation spillovers ( ) for each k-th sector from the s-th Region to the r-th Region un-weighted by the geographical distance are expressed as:

rs k

ts

s k s k r k s k r k rs k

t RSI RSI RSI RSI ts ⋅         + − =

−1

r s ≠ ∀

The resulting (20 x 20) matrix of spillovers for each k-th sector (with a vector of 0 in the diagonal dimension )

r s = ∀ .

Innovation spillovers (spatial weights)

Several alternative criteria to transform geo-distances into spatial weights

Binary contiguity concept (D1): assumes interregional spillovers take place only between direct neighbours (comon border) ( )

∑

≠ =

=

n r s s rs rs k r k

w ts ts D

, 1 1

with wrs = 1 only if s neighbouring r

11

k-nearest neighbours (D2): thresold distance, 300km inverse distances (D3): the intensity of influences between any two regions diminishes continuosly with increasing distances

Environmental spillovers

The role of extra regional environmental regulation on regional environmental performance

environmental policy acts as a centrifugal force - increasing compliance costs reduce the convenience to localize industrial activities in that region emissions produced by neighbouring Regions may represent the role

• f economic agglomeration phenomena in explaining environmental

12

• f economic agglomeration phenomena in explaining environmental

performances (Gray and Shadbegian, 2007) concentration of dirty activities into circumscribed geo-areas

Measure

Emission intensity of the surrounding regions environmental spillovers as the sum of sectoral emissions per unit of value added from the other regions (eks) valid for ∀s ≠ r weighted by distances D1, D2, D3

Environmental regulation

Proxied by

The stringency of the environmental regulatory framework

• the incidence of environmental regulation on average regional income

(Costantini and Crespi 2008)

• Public expenditure 4env.protection may be considered as a proxy of the

(regional) WTP of citizens to preserve natural environment

Measure(s)

13

Measure(s)

3 alternative public expenditure (regional) measures

• Current exp. 4environmental protection activities
• Capital exp. 4environmental protection activities
• R&D exp. 4environmental protection activities

Data

• Expenditures 4environmental protection activities (Istat, 2007)
• by region

Diagnostic checks

Spatial dependence

LM lag and LM error tests

Potential endogeneity of regional innovation

Hausman test

14

Hausman test

The geographical distribution

• f environmental performance

in the Italian regions

Empirical evidence (i)

in the Italian regions

i. Shift-Share analysis (only direct emissions)

• industry mix effect (other things being equal):
• industrial regional specialization matters
• more industrialized N regions are (obv.) penalized

16

• efficiency effect (given an homogeneous industry mix across

regions):

• NW regions perform well
• some N regions perform bad (es. FVG)

Shift-Share: productive specialization (industry mix) component

• 17

Note: Below zero values indicate positive performances

Shift-Share: efficiency component

• 18

Note: Below zero values indicate positive performances

Value Added per worker GHG emissions per Value Added ACID emissions per Value Added

Reg.distribution of Value added per worker, GHG and ACID emissions for NAMEA Sector #9 (NACE codes: DF-DG)

Reg.distribution of Value added per worker, GHG and ACID emissions for NAMEA Sector #12 (NACE codes: DK-DL-DM)

Value Added per worker GHG emissions per Value Added ACID emissions per Value Added

Driving forces of the geographical distribution of environmental performances

Empirical evidence (ii)

environmental performances

MODEL ESTIMATION [adapted from Cole et al.2005; Medlock and Soligo (2001)]

r k r k r k r k r k r k r k r k

p ts t es lp e ε β β β β β α + + + + + + =

− − − + − 5 4 3 2 1 22

DATASET (qxn)x1 q regions (q=20) n sectors (n=11 manufacturing sectors)

Drivers for environmental performances (GHG)

Dep var GHG (1) (2) (3) (4) (5) (6) (7) Labour productivity

• 0.756***
• 0.671***
• 0.688***
• 0.714***
• 0.501***
• 0.542***
• 0.522***

(-4.13) (-3.85) (-4.05) (-4.12) (-2.94) (-3.17) (-3.09) Internal Innovation

• 0.009
• 0.001

0.002 0.005 0.009 0.003 0.014 (-0.33) (-0.04) (0.01) (0.16) (0.32) (0.11) (0.50) Energy Intensity 0.645*** 0.541*** 0.531*** 0.549*** 0.567*** 0.557*** 0.583*** (14.67) (11.64) (12.23) (10.63) (11.41) (12.31) (10.18) Dirty Sector dummy 1.331*** 0.996*** 0.925*** 1.033*** 0.976*** 0.894*** 0.997*** (12.81) (7.33) (6.64) (7.17) (7.08) (6.31) (6.67)

• Environ. Spillovers D1

0.243*** 0.236*** (3.84) (3.57)

• Environ. Spillovers D2

0.289*** 0.288*** (4.40) (4.40)

• Environ. Spillovers D3

0.229*** 0.216*** (3.05) (2.74)

• Tech. Reg. Spillovers D1
• 0.125***

(-2.97)

• Tech. Reg. Spillovers D2
• 0.097**

23

• Tech. Reg. Spillovers D2
• 0.097**

(-2.57)

• Tech. Reg. Spillovers D3
• 0.152***

(-2.98) Constant 4.121*** 4.083*** 2.77*** 4.014*** 3.013*** 2.184*** 3.01*** (6.77) (6.72) (5.01) (6.80) (4.67) (3.69) (4.91) Regional dummies Yes Yes Yes Yes Yes Yes Yes No obs. 209 209 209 209 209 209 209 Adj R-sq 0.78 0.80 0.81 0.79 0.81 0.81 0.81 F-stat 32.22 42.3 44.92 40.35 39.6 45.31 41.55 Root MSE 0.63 0.61 0.60 0.62 0.60 0.59 0.60 Hausman Chi-sq 0.23 (0.63) 0.02 (0.89) 0.05 (0.82) Average VIF value 1.54 1.45 1.73 LM (lag) 0.03 (0.86) 0.01 (0.94) 0.01 (0.97) 0.01 (0.97) 0.12 (0.72) 0.02 (0.89) 0.15 (0.69) LM (error) 3.88 (0.05) 3.19 (0.07) 3.40 (0.07) 2.50 (0.11) 3.31 (0.07) 3.34 (0.07) 3.18 (0.07) Robust LM (error) 4.95 (0.03) 3.64 (0.06) 3.94 (0.05) 2.90 (0.09) 3.33 (0.07) 3.67 (0.06) 3.12 (0.08)

Drivers for environmental performances (ACID)

Dep var ACID (1) (2) (3) (4) (5) (6) (7) Labour productivity

• 1.543***
• 1.383***
• 1.301***
• 1.313***
• 1.201***
• 1.139***
• 1.051***

(-6.16) (-5.32) (-5.73) (-4.76) (-4.61) (-4.94) (-3.93) Internal Innovation

• 0.019
• 0.017
• 0.013
• 0.010
• 0.006
• 0.010

0.004 (-0.53) (-0.47) (-0.36) (-0.28) (-0.17) (-0.29) (0.10) Energy Intensity 0.404*** 0.373*** 0.358*** 0.352*** 0.398*** 0.389*** 0.392*** (8.97) (7.60) (8.01) (6.88) (7.59) (8.15) (7.18) Dirty Sector dummy 2.559*** 2.272*** 2.034*** 2.155*** 2.247*** 2.008*** 2.084*** (20.76) (9.05) (7.03) (8.55) (9.03) (6.97) (8.46)

• Environ. Spillovers D1

0.109 0.106 (1.35) (1.31)

• Environ. Spillovers D2

0.195** 0.191** (2.16) (2.12)

• Environ. Spillovers D3

0.163* 0.162** (1.90) (1.96)

• Tech. Reg. Spillovers D1
• 0.134**

(-2.40)

24

(-2.40)

• Tech. Reg. Spillovers D2
• 0.111**

(-2.29)

• Tech. Reg. Spillovers D3
• 0.204***

(-3.12) Constant 4.596*** 4.423*** 3.489*** 4.228*** 3.281*** 2.833*** 2.865*** (5.41) (5.21) (4.47) (4.89) (3.51) (3.46) (3.28) Regional dummies Yes Yes Yes Yes Yes Yes Yes No obs. 209 209 209 209 209 209 209 Adj R-sq 0.77 0.77 0.77 0.77 0.78 0.78 0.79 F-stat 47.96 49.87 54.32 49.8 48.30 54.90 50.78 Root MSE 0.77 0.77 0.77 0.77 0.76 0.76 0.75 Hausman Chi-sq 0.04 (0.84) 0.01 (0.93) 0.05 (0.83) Average VIF value 1.70 1.83 1.98 LM (lag) 0.03 (0.86) 0.02 (0.89) 0.01 (0.92) 0.01 (0.93) 0.10 (0.76) 0.07 (0.79) 0.21 (0.65) LM (error) 0.68 (0.41) 0.71 (0.40) 0.79 (0.37) 0.44 (0.50) 0.87 (0.35) 1.12 (0.29) 1.11 (0.29)

ii. Econometric model for emission intensity

• labour productivity gains associated with env. efficiency gains
• environmental spillovers (positive and significant)
• major effect with regions in the range of 300km (D2)
• convergence in production processes and techniques?
• internal innovation (pat/va) plays no role
• tech interregional spillovers play a role
• higher for more localised pollutants (ACID)
• more likely the availability of env-friendly technologies
• Spatial dependence
• GHG: weak evidence
• ACID: no evidence

25

The role of environmental regulation

GHG ACID (1) (2) (3) (4) (5) (6) Labour productivity

• 0.501***
• 0.542***
• 0.522***
• 1.201***
• 1.139***
• 1.051***

(-2.94) (-3.17) (-3.09) (-4.61) (-4.94) (-3.93) Internal Innovation 0.009 0.003 0.014

• 0.006
• 0.01

0.004 (0.32) (0.11) (0.50)

• (0.17)

(-0.29) (0.10) Energy Intensity 0.567*** 0.557*** 0.583*** 0.398*** 0.389*** 0.392*** (11.41) (12.31) (10.18) (7.59) (8.15) (7.18) Dirty Sector dummy 0.976*** 0.894*** 0.997*** 2.247*** 2.008*** 2.084*** (7.08) (6.31) (6.67) (9.03) (6.97) (8.46)

• Environ. Spillovers D1

0.236*** 0.106 (3.57) (1.31)

• Environ. Spillovers D2

0.288*** 0.191** (4.40) (2.12)

• Environ. Spillovers D3

0.216*** 0.162** (2.74) (1.96)

• Tech. Reg. Spillovers D1
• 0.125***
• 0.134**
• Tech. Reg. Spillovers D1
• 0.125***
• 0.134**

(-2.97) (-2.40)

• Tech. Reg. Spillovers D2
• 0.097**
• 0.111**

(-2.57) (-2.29)

• Tech. Reg. Spillovers D3
• 0.152***
• 0.204***

(-2.98) (-3.12)

• Env. Reg. Current Exp.
• 0.105
• 0.62**

(-0.81) (-2.05)

• Env. Reg. Capital Exp.
• 0.005
• 0.272**

(-0.03) (-2.03)

• Env. Reg. R&D Exp.
• 0.163**
• 0.288**

(-2.58) (-2.27) Constant 2.95*** 2.187*** 2.527*** 4.738*** 4.143*** 1.84** (4.54) (3.54) (3.85) (5.78) (5.90) (2.11) Regional dummies Yes Yes Yes Yes Yes Yes No obs. 209 209 209 209 209 209 Adj R-sq 0.81 0.81 0.80 0.77 0.78 0.79 F-stat 39.60 45.31 41.55 48.30 54.90 50.78 Root MSE 0.60 0.59 0.60 0.76 0.76 0.75

26

ii. Econometric model for emission intensity the role of environmental regulation

• environmental regulation
• plays a role more effective for ACID emissions
• plays a role in GHG emissions too
• BUT only R&D public expenditures 4env protection
• BUT only R&D public expenditures 4env protection

27

Concluding remarks

Labour productivity, innovation efforts and region-specific regulatory framework matter North vs South different performances within N and S different performances

28

existence of agglomeration effect at the sectoral level higher conc. of polluting firms adopting dirtier prod. processes centripetal forces associated to inn.spillovers among regions a first influence of the tech adopted in the production processes limit to distance: 300km

Concluding remarks

4global pollutants (GHG): the agglomerative impact associated to env. efficiency externalities overwhelm the clustering effect due to general inn. spillovers more collective action is necessary

29

4localized pollutants (ACID): the opposite occurs a geo-circumscribed collective action is effective

Thank you !

costanti@uniroma3.it costanti@uniroma3.it ma.maz@iol.it anna.montini@unibo.it