Presentation at the 59th Annual Meeting of the Italian Economic Society ( Regione Emilia-Romagna, Bologna - 25-27 october 2018)

1. The geography of eco-innovation in Italy.
Evidence from patent data at firm level and
the role of family involvement
Francesco Aiello - Paola Cardamone
Lidia Mannarino - Valeria Pupo
Department of Economics, Statistics and Finance “Giovanni Anania”
University of Calabria
I-87036 Arcavacata di Rende (CS), Italy

2. Agenda
• Scope & Motivation
• Literature review
• Paper contribution
• Empirical setting
– Data source
– Green patenting in a nutshell
– Modelling firm green patenting
• Basiline results
• Sensitivity analysis
• Some conclusions
• Going ahead

3. The scope
• A wide literature investigates the determinants
of eco-innovation. Recent exhaustive surveys: Barbieri et al, (2016); de
Jesus Pacheco et al (2016); del Río et al. (2016); Hojnik and Ruzzier (2016)
• However, little is known about the role of family
involvement in generating green technology
• We address this issue by analyzing how family
and nonfamily firms differ in terms of introducing
green technologies. Eco-innovation is gauged by
green patents

4. Motivation
• Gaining a deeper understanding of the association between
family ownership and eco-innovations is a relevant for two
reasons (at least):
– Family firms constitute a significant part of industrial structure
of many economies (IFERA, 2013)
– Policy-makers and international organizations have shown
increasing concern for the impact of economic activity on the
natural system (e.g., United Nations , European Commission)
• Therefore, the understanding of the conditional effect of
the most prevalent form of business ownership on firm
green patenting activity may prove helpful insights in
optimizing innovation and sustainability related policies
(Bammens and Hunermund, 2018)

5. Literature review: 2 backgrounds
• Two (wide) theoretical arguments suggest
that family firms differ from non-family firms
in terms of environmental behavior:
– Risk aversion: family firms induced by the family
as owners could have a negative influence on their
environmental innovation activities
– Socioemotional Wealth (SEW) (Gómez-Mejía et
al, 2007): specific characteristics of family
businesses can either foster or hinder green
innovation

6. Literature review/Risk aversion
– Members of family firms pursue the accumulation and
conservation of wealth
• future family generations (Zellweger et al., 2012)
• family name
• firm reputation, which has often been built up over several
generations (Dyer and Whetten, 2006; Lumpkin et al., 2010)]
– These aspects are indicative of a more general long-
term orientation of family firms, which makes risky
behaviour less likely (Gómez-Mejía et al., 2007; Le
Breton-Miller and Miller, 2006; Lumpkin et al., 2010)

7. Literature review/SEW1/2
• Family owners are concerned about not only the financial
returns but also on the socio-emotional wealth that are
derived though their ownership shares. The latter category
includes a variety of non-financial aspects of firm ownership
that can influence corporate social responsibility - that
comprises ecological concerns - such as the desires to obtain
high social status in a local community (Block, 2010) and to
fulfil needs related to organizational and family identification
(Le Breton-Miller et al., 2011; Zellweger et al., 2010)
• Following these arguments, family firms should be more
interested than non-family firms in generating green
technology

8. Literature review/SEW2/2
• Recent works, however, show that SEW can be considered as
double-edged sword that can either reveal its bright or dark side
(Cruz et al., 2014; Kellermanns et al., 2012; Kim et al., 2016;
Samara et al 2018):
– Bright side of SEW: the family firms value their reputation and role
in a community more than the nonfamily firms and thus behave more
responsibly
– Dark side of SEW: due to their concern with preserving the business
financial stability and a sense of financial responsibility for preserving
family wealth across generations, family firms are less likely to invest in
the protection of the environment, as investments in environmental
sustainability are a net cost

9. Literature review: empirics
• There are few empirical studies on
environmental behaviour of family firms.
• Results are contradictory
• These studies may be classified in two groups:
– Literature on the innovative behavior of family
firms
– Literature on the influence of family ownership on
corporate social responsibility (CSR) (the
environment is a dimension of CSR)

10. Few empirics from the literature on
the innovative behavior of family firms
• Family firms pursue inherently uncertain
activities less than non-family firms, and
therefore invest fewer resources into R&D and
innovation (Block, 2012; Duran et al., 2015)
• This could also concern inherently uncertain
innovation with environmentally highly
beneficial products

11. Empirics from CSR
– Some scholars argue that family-owned firms are more
environmentally responsible than their nonfamily counterparts (e.g.
Berrone et al., 2010; Block and Wagner , 2014; Marques et al., 2014),
while others raise some doubts (Cruz et al., 2014; Kellermanns et al.,
2012). Finally, Calza et al. (2016) find that ownership concentration -
as is the case for family firms - negatively determine proactive
environmental firm behavior
– Doluca et al (2018) provide a more complex overview of the
differences regarding the development of firms’ environmental
behavior.
• The story they tell is as follows: family firms are less likely than other firms to
implement environment-related activities and innovations in early diffusion phases
but catch up with non-family firms later. They also find that family firms’
environmental behaviour is less volatile and more stable over time compared with
non-family firms

12. Paper contribution
• Nobody has investigated the environmental behaviour of
family firms in terms of the production of green patents (meant
as a proxy of new “environmental” knowledge). We focus on
how Italian family firms differ from non-family firms paying
attention to the propensity to green patent.
• Several studies hypothesize that EMS can promote eco
innovations (Wagner, 2007; Dangelico et al 2016; Montobbio
and Solito, 2018). Theoretical arguments suggest that family
firms differ from non-family firms regarding the role of EMS
certification, which merits separate and comparative analysis.
BUT no research has so far investigated how the implementation
of environmental management systems can spur green
innovation for family and nonfamily firms
• Does patenting accumulation impact differently?
• Policy implications. In period of scarcity of resources to devote
to the environment, investigating the role of the family firms
can help policy-makers to implement more effective instruments
to promote eco-innovation: policy-makers should better
support family firms in acting pro-environmentally and better
exploit the potential of environmental management systems to
promote eco innovations

13. Empirical setting. Dataset construction
• The analysis is based on a rich panel dataset built by combining multiple data sources on administrative patent data,
firm information, patent characteristics and environmental management systems (ISO 14001).
• Our final sample consists of about 4500 Italian manufacturing firms in the period 2009-2017.
• We selected these firms from a panel of 26000 firms in the Amadeus (Bureau van Dijk) database based on the
criterion that they should have applied for at least one patent at the EPO between 1981 and 2017. The focus on
patenting companies (either in the considered period or before the period) allows relying on a homogeneous
population of potentially innovative firms for which patenting is (or has been) a relevant tool to protect their
inventions/innovations.
• The firm level databases by Bureau van Dijk are unique in two important ways. First, they provide information on the
ownership structure of the firms. Second, they contain firm-level balance sheet data in an internationally comparable
format. Using the ownership information, we are able to identify the ultimate owner for each firm in the sample.
• We extract patents from the Orbis dataset provided by Bureau van Dijk, which has recently been linked to the
European Patent Office’s (EPO) PATSTAT dataset. The main advantage of using the Orbis-PATSTAT dataset is that it
provides us with an unique firm’s identifier that will allow us to match firm-level patents to balance sheet and income
statement data at a later stage. This allows us to link patents of the applying firms to the comprehensive information
contained in Bureau van Dijk's Amadeus database via common identifiers.
• We count the number of granted patents per firm per year, including only priority patents and excluding equivalent
patent filings. The fact that we focus on granted patents of firms’ registered in Orbis implies that our sample is not
likely to include the lowest quality patents (such as non-successful applications)
• In addition green patents in Orbis are identified using the WIPO Green Inventory
• The WIPO Green Inventory includes all the IPC classes that are associated with environment friendly technologies in a
variety of fields. In particular, it includes seven technological fields (alternative energy production, transportation,
energy conservation, waste management, agriculture/forestry, and administrative regulatory and nuclear power
generation
• We merged financial data for the whole list of firms from 2009 to 2017 and the patent portfolio from the Amadeus
database. We obtained a final panel spanning from 2009 to 2017, reporting observations on about 4500 italian firms
in the manufactoring sectors. We then merged these 4500 firms with the ones contained in the Accredia Register,
updated to 2017, in order to identify the certified firms whit ISO 14001, merging first tax code.

14. Green patenting in Italy
Some descriptive evidence

21. Small family

23. Intensity in patenting by firm group

24. Dependent variable
• Dummy variable: 1 when firm i-th has at least
one green patent at time t; 0 otherwise.
Green patents are identified using the WIPO
Green Inventory
• Robustness: number of green patents by firm
(three-years accumulation)

25. Family involvement
• There is still no agreement on the
definition of family business (for a recent
review, see Hernàndez-Linares et al, 2018)
• Family firm = individuals or family
members record a direct ownership of
over 50%

26. Accumulation of green patenting
Stock of Green and nonGreen: Kk
it = PATk
it + (1 −d)Kk
i,t−1
– The firm patent stock is computed for the period 1981–2017. We use
the perpetual inventory method (Cockburn and Griliches 1988).
– k = G, NG (G = green, NG = non green).
– knowledge depreciation rate (δ) = 10%
– firm past experience has been disentangled between green and non-green
technologies in order to control overlaps between the two categories
• Expectations
– The higher the past knowledge in green technologies, the higher
the amount of green inventions (positive internal within-
learning effect)
– The higher the past knowledge in non green technologies, the
higher the amount of green inventions (positive crossing
learning effect)

27. Green certification
• The Environmental Management System (EMS) is measured
through the certification from the International Organization
for Standardization’s ISO14001 standard
• ISO 14001 is the most widespread standard that supports
organizations in the implementation and maintenance of their
EMS defining a list of requirements to improve environmental
performance (source ACCREDIA, the Italian Accreditation
Body)
• Expectations: From Wagner (2007) Dangelico et al (2016) and
Montobbio and Solito (2018) we learn that:
– Reputational and image-related reasons promote family
firms’ behaviour towards the adoption of EMS certification.
– But, given that family firms are smaller than their non-family
counterparts (Block and Wagner, 2014; Zahra et al., 2004),
the lack of human and financial resources hinders the
adoption green certifications

28. BASELINE RESULTS

29. ALL FIRMS - Probit RE (2008-2017)
VARIABLES Coefficients AME Coefficients AME -
ISO (-1) 0.3178*** 0.0083*** 0.3036*** 0.0078***
FAMILY -0.1952*** -0.0040*** -0.2038*** -0.0042***
NO GREEN Stock of Patenting 0.0078*** 0.0002*** 0.0078*** 0.0002***
GREEN Stock of Patenting (-1) 0.3017*** 0.0063*** 0.2947*** 0.0061***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0001 0.0000 0.0000 0.0000
AGE -0.0042** -0.0001** -0.0046** -0.0001**
Constant -2.6544*** -2.5474***
Year dummies no no yes yes
Observations 26,109 26,109 26,109 26,109
Number of firms 3,903 3,903
log likelihood -1338 -1330
Wald chi2 185.3 200.8
p-value 0 0
*** p<0.01, ** p<0.05, * p<0.1

30. ALL FIRMS - Probit RE
VARIABLES Coefficients AME Coefficients AME -
ISO (-1) 0.3178*** 0.0083*** 0.3036*** 0.0078***
FAMILY -0.1952*** -0.0040*** -0.2038*** -0.0042***
NO GREEN Stock of Patenting 0.0078*** 0.0002*** 0.0078*** 0.0002***
GREEN Stock of Patenting (-1) 0.3017*** 0.0063*** 0.2947*** 0.0061***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0001 0.0000 0.0000 0.0000
AGE -0.0042** -0.0001** -0.0046** -0.0001**
Constant -2.6544*** -2.5474***
Year dummies no no yes yes
Observations 26,109 26,109 26,109 26,109
Number of firms 3,903 3,903
log likelihood -1338 -1330
Wald chi2 185.3 200.8
p-value 0 0
*** p<0.01, ** p<0.05, * p<0.1

31. ALL FIRMS - Probit RE
VARIABLES Coefficients AME Coefficients AME -
ISO (-1) 0.3178*** 0.0083*** 0.3036*** 0.0078***
FAMILY -0.1952*** -0.0040*** -0.2038*** -0.0042***
NO GREEN Stock of Patenting 0.0078*** 0.0002*** 0.0078*** 0.0002***
GREEN Stock of Patenting (-1) 0.3017*** 0.0063*** 0.2947*** 0.0061***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0001 0.0000 0.0000 0.0000
AGE -0.0042** -0.0001** -0.0046** -0.0001**
Constant -2.6544*** -2.5474***
Year dummies no no yes yes
Observations 26,109 26,109 26,109 26,109
Number of firms 3,903 3,903
log likelihood -1338 -1330
Wald chi2 185.3 200.8
p-value 0 0
*** p<0.01, ** p<0.05, * p<0.1

32. ALL FIRMS - Probit RE
VARIABLES Coefficients AME Coefficients AME - Coefficients AME Coefficients AME
ISO (-1) 0.3178*** 0.0083*** 0.3036*** 0.0078*** 0.3044*** 0.0079*** 0.2973*** 0.0076***
FAMILY -0.1952***
-
0.0040*** -0.2038***
-
0.0042*** -0.2047***
-
0.0042*** -0.2059***
-
0.0042***
NO GREEN Stock of
Patenting 0.0078*** 0.0002*** 0.0078*** 0.0002*** 0.0075*** 0.0002*** 0.0076*** 0.0002***
GREEN Stock of Patenting (-
1) 0.3017*** 0.0063*** 0.2947*** 0.0061*** 0.2942*** 0.0061*** 0.2931*** 0.0061***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
AGE -0.0042** -0.0001** -0.0046** -0.0001** -0.0046** -0.0001** -0.0044** -0.0001**
high_tech 0.0468 0.0010 0.0537 0.0012
med_high_tech -0.0319 -0.0007 -0.0169 -0.0004
med_low_tech -0.1176 -0.0023 -0.1125 -0.0022
centre -0.0271 -0.0006
south 0.2699** 0.0071*
Constant -2.6544*** -2.5474*** -2.5033*** -2.5233***
Year dummies no no yes yes yes yes yes yes
Observations 26,109 26,109 26,109 26,109 26,109 26,109 26,109 26,109
Number of firms 3,903 3,903 3,903 3,903
log likelihood -1338 -1330 -1329 -1327
Wald chi2 185.3 200.8 203.2 207.8
p-value 0 0 0 0
*** p<0.01, ** p<0.05, * p<0.1

33. Family vs Non-family - Probit RE
VARIABLES
Coefficients
- Family
AME -
Family
Coefficients
- Family
AME -
Family
Coefficients
- Non-family
AME - Non-
family
Coefficients -
Non-family
AME - Non-
family
ISO (-1) 0.3524** 0.0077 0.3500** 0.0077 0.3212*** 0.0102*** 0.3145*** 0.0100***
NO GREEN Stock of Patenting 0.0380** 0.0006** 0.0374** 0.0006** 0.0068** 0.0002** 0.0069** 0.0002**
GREEN Stock of
Patenting (-1)
0.184** 0.0029* 0.1891** 0.0030* 0.3085*** 0.0082*** 0.3058*** 0.0081***
EMPLOYEES (-1) -0.0004 -0.0000 -0.0004 -0.0000 0.0000 0.0000 0.0001 0.0000
ROE (-1) -0.0003 -0.0000 -0.0003 -0.0000 0.0001 0.0000 0.0001 0.0000
AGE -0.0060* -0.0001* -0.0059* -0.0001* -0.0038 -0.0001 -0.0035 -0.0001
high_tech 0.0662 0.0011 0.0701 0.0012 0.0208 0.0006 0.0362 0.0010
med_high_tech -0.0613 -0.0010 -0.0369 -0.0006 -0.0175 -0.0005 -0.0076 -0.0002
med_low_tech -0.0450 -0.0007 -0.0301 -0.0005 -0.1927 -0.0048 -0.1938 -0.0048
centre 0.0875 0.0015 -0.1248 -0.0031
south 0.2788 0.0057 0.2665 0.0088
Constant -2.5282*** -2.5555*** -2.6134*** -2.6252***
Year dummies yes yes yes yes yes yes yes yes
Observations 13,934 13,934 13,934 13,934 12,175 12,175 12,175 12,175
Number of BVD_id 2,131 2,131 1,772 1,772
log likelihood -515.3 -514.1 -803.4 -801.9
Wald chi2 54.99 58.39 129 131.5
p-value 3.57e-06 3.72e-06 0 0
*** p<0.01, ** p<0.05, * p<0.1

34. Robustness check
Balanced data and Probit RE

35. Robustness check: balanced panel - ALL FIRMS - Probit RE
VARIABLES Coefficients AME Coefficients AME Coefficients AME Coefficients AME
ISO (-1) 0.3153*** 0.0072** 0.3170*** 0.0071** 0.3158*** 0.0071** 0.3133*** 0.0070**
FAMILY -0.3380*** -0.0059*** -0.3461*** -0.0059*** -0.3480*** -0.0060*** -0.3512*** -0.0060***
NO GREEN Stock of Patenting 0.0127*** 0.0002*** 0.0126*** 0.0002*** 0.0121*** 0.0002*** 0.0122*** 0.0002***
GREEN Stock of Patenting (-1) 0.2670*** 0.0050*** 0.2660*** 0.0049*** 0.2718*** 0.0050*** 0.2696*** 0.0050***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0017* 0.0000 0.0016 0.0000 0.0016 0.0000 0.0016 0.0000
AGE -0.0036 -0.0001 -0.0041 -0.0001 -0.0039 -0.0001 -0.0037 -0.0001
high_tech 0.2279 0.0050 0.2329 0.0051
med_high_tech 0.0596 0.0011 0.0682 0.0013
med_low_tech -0.0049 -0.0001 -0.0076 -0.0001
centre -0.0386 -0.0007
south 0.2660 0.0062
Constant -2.7721*** -2.7634*** -2.8154*** -2.8325***
Year dummies Yes Yes Yes Yes Yes Yes
Observations 16,120 16,120 16,120 16,120 16,120 16,120 16,120 16,120
Number of BVD_id 2,015 2,015 2,015 2,015
log likelihood -756.7 -749.1 -747.7 -746.7
Wald chi2 113.8 126 128.9 130.3
p-value 0 0 0 0
*** p<0.01, ** p<0.05, * p<0.1

36. Robustness check: balanced panel – Family vs Non-family- Probit RE
VARIABLES
Coefficients
- Family
AME -
Family
Coefficients
- Family
AME -
Family
Coefficients
- Non-family
AME -
Non-family
Coefficients
- Non-family
AME -
Non-family
ISO (-1) 0.4106** 0.0074 0.4199** 0.0076 0.2763** 0.0080** 0.2670** 0.0076*
NO GREEN Stock of
Patenting 0.0243 0.0003 0.0237 0.0003 0.0126*** 0.0003*** 0.0127*** 0.0003***
GREEN Stock of Patenting (-
1) 0.5291*** 0.0062 0.5120** 0.0060 0.2757*** 0.0070*** 0.2749*** 0.0069***
EMPLOYEES (-1) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
ROE (-1) 0.0032** 0.0000** 0.0032** 0.0000** 0.0009 0.0000 0.0009 0.0000
AGE -0.0056 -0.0001 -0.0054 -0.0001 -0.0027 -0.0001 -0.0024 -0.0001
high_tech 0.3275 0.0054 0.3165 0.0051 0.1654 0.0047 0.1790 0.0051
med_high_tech 0.1596 0.0019 0.1509 0.0018 -0.0047 -0.0001 0.0111 0.0003
med_low_tech -0.0347 -0.0004 -0.0465 -0.0005 -0.0072 -0.0002 -0.0049 -0.0001
centre -0.1482 -0.0015 0.0080 0.0002
south 0.0722 0.0009 0.3442 0.0114
Constant -2.6846*** -2.6631*** -2.9009*** -2.9331***
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
Observations 7,448 7,448 7,448 7,448 8,672 8,672 8,672 8,672
Number of BVD_id 931 931 1,084 1,084
log likelihood -189 -188.7 -551.7 -550.7
Wald chi2 34.84 35.23 81.96 83.66
p-value 0.00418 0.00884 7.36e-11 1.95e-10

37. Robustness check
Probit MLM estimations
A. 2-level hierarchy (firms-city)
B. 3-level hierarchy (firms-city-province)
C. 4-level hierarchy (firms-city-province-
region)

38. 2-level hierarchy: firms are embedded in province (NUTS3)

40. Concluding remarks

41. Green intensity in patenting by area

42. Specialization in green patenting
Balassa index by region and time

43. Green intensity in patenting over time
Micro-sized firms

44. Green intensity in patenting over time
small-sized firms

45. Green intensity in patenting over time
Medium-sized firms

46. Green intensity in patenting over time
Large-sized firms

47. Green Patents in Italy by IPC codes