Migrations, communities-on-the-move and knowledge flows: An empirical analysis of the Spanish provinces in the last decade. Anna D'Ambrosio1,4, Sandro Montresor2, Mario Davide Parrilli3, Francesco Quatraro1,4 October 2016 1
University of Turin, Department of Economics and Statistics “Cognetti de Martiis” (IT) 2
3
4
Kore University of Enna (IT) University of Bournemouth (UK) BRICK, Collegio Carlo Alberto (IT) Abstract This paper investigates the role of regional immigrants and emigrants for the occurrence of innovation-based knowledge
flows between regions and foreign countries. Drawing on the evolutionary economic geography literature on knowledge
diffusion and on the role of proximity for it, we posit that immigrants (emigrants) act as a transnational knowledge
bridge between the host (home) regions and their origin (destination) countries, and thus facilitate the occurrence of
innovation-based knowledge flows between them. We also argue that the social capital of both the hosting and moving
communities reinforce such a knowledge bridging role, as well as their commonality of language and migrants' human
capital . By combining OECD patent data with official national data on residents and electors abroad, we test our
hypotheses by applying a gravity model to the knowledge flows between the Spanish provinces (NUTS3) and a number
of foreign countries in different periods of the last decade. As expected, both immigration and emigration are found to
contribute to the innovation-based knowledge flows at hand; the effect is stronger for more skilled migrants; andsocial
capital of both the moving and of the hosting community is found to positively moderate the effect that migration has
on knowledge flows. The impact of migration is stronger with non-Spanish speaking countries, pointing to a languagebridging role of migrants. Overall, individual and community aspects combine in accounting for the impact of
migrations on knowledge flows. Key-words: Migrations; communities-on-the-move; innovation-based knowledge flows; social
capital. JEL codes: R11; F22; O33. 1. Introduction As regions become increasingly more permeable to the mobility of goods, services, capital and
people, the migration of individual workers has been accompanied by the rise of transnational
ethnic communities spanning across local contexts (Portes, 2000, Vertovec, 1999; Castles, 2002;
Saxenian, 2006; Sonderegger and Taube, 2010). Regional studies have, from different perspectives,
found evidence of the manifold impact of immigration on the workforce composition, the industrial
base, and the economic performance of regions (e.g. Piore, 1986; Ottaviano and Peri, 2006; Peri and
Sparber, 2008; Niebuhr, 2010; Lewis, 2013; De Arcangelis et al., 2013; Gagliardi, 2015). These
studies usually model that the effect of immigration on regional economies proceeds as an injection
of (more or less skilled) labour or as a source of cognitive diversity. Another mechanism that has
been identified in the literature could be labelled as a “diaspora effect” in the sense that it refers to
the activation to economic ends of the persistence of strong ties to the homeland (Brubaker, 2006).
Specifically, migration flows and diasporas have been claimed to play the role of “information
brokers” between the host and the home regions, allowing for knowledge flows, which are deemed
to explain the benefits they get from engaging in international trade and FDI (Rauch and Trinidade,
2002; Wagner et al., 2002; Peri and Requena-Silvente, 2010; Felbermayr et al., 2014). The knowledge flows brought to local economies by migration have recently attracted more direct
attention. The analysis has concentrated on innovation-based knowledge flows linked to skilled
migration at the local level, such as those revealed by cross-regional patent citations and coinventorships (Almeida and Kogut, 1999; Agrawal et al, 2006, 2008; Breschi and Lissoni, 2009,
2006a, 2006b). This kind of knowledge flows does also represent the focus of the present paper. Its
aim is to investigate the extent at which the innovation-based knowledge flows that regions
exchange with foreign countries are affected by the immigrants and the emigrants of the regions,
coming from and going to the correspondent countries, respectively. In so doing, we aim at
extending the standard analysis of the issue at stake in at least two respects. First of all, by drawing
on the economic geography literature and combining it with recent local/regional studies on
innovation systems (Autio, 1998, Cooke et al., 2004, Breschi and Lissoni, 2001, Paci and Usai,
2000, Tödtling and Trippl, 2005, Boschma, 2005), we extend the analysis of the “carriers” of the
knowledge flows at hand beyond that of high skilled workers and mobile inventors, and look at
migration in broader terms. We argue that, even if with lower levels of formal education and
training, immigrants and emigrants embody a tacit kind of knowledge and experience of their
country of origin, which can be facilitate the exchangeof more formal and codified knowledge
between the inventors of the host and home countries. Our arguments can be grounded in Boschma's (2005) proximity framework. Assuming that foreign
inventors embody knowledge that is relevant for domestic inventors in a given region, we can argue
that geographic, cognitive and cultural (i.e. institutional, Boschma's framework) distance hamper
access to relevant knowledge (see also Picci, 2010). Social and institutional proximity within the
transnational networks of co-ethnics may allow immigrants accessing part of this knowledge. This
is more likely to occur by higher immigrants' human capital, which reduces the cognitive distance
between immigrants and foreign inventors in the home country. In the host region, the geographic
proximity between immigrants and the local community facilitates the creation of connections
between domestic investors and immigrants. Initial institutional distance will be lower for
immigrants who speak the language of the host region; cognitive distance will be lower for more
skilled immigrants; but for all immigrants, geographic proximity over time is likely to contribute to
reduce the social and institutional distance (cfr. Balland et al., 2015). Nonetheless, in a highly
segregated society, geographic proximity may fail to have an effect on other kinds of proximity. We
argue that the social capital of both the host and the moving community facilitates the reduction of
social and ultimately of institutional distance, allowing for knowledge, including innovationrelevant knowledge, to flow. We operationalize the above arguments by systematically investigating the impact of migration on
knowledge flows measured by coinventorships and by disentangling the mediating effects of social
capitallanguage commonality and human capital. We focus on Spanish provinces in their relationships with an ample set of foreign countries,
measuring the effects of both immigration and emigration over different periods of time (depending
on the considered flow) in the last decade. As usual, the choice of the context is jointly motivated
by relevance and convenience. According to Eurostat, Spain scored the highest migration rates in
Europe in the 2000s; this situation led Spanish innovation systems to pursue a set of radical
innovation strategies to compete with other advanced economies (Rodriguez-Pose, 2014; Alberdi et
al., 2014). On the other hand, a dataset built up by the authors for related investigations
(D’Ambrosio and Montresor, 2016; Quatraro and Usai, 2016) was available and integrated ad-hoc
for the present one. The empirical analysis has been carried out by using a gravity model of
knowledge flows – based on co-invention and patent citation data – quite standard in the literature
(e.g. Maurseth and Verspagen, 2002; Paci and Usai, 2009; Picci, 2010), which we have originally
specified at the province-country dyadic level. Our results support the expectation that both immigrants and emigrants contribute to innovationbased knowledge flows between the investigated local contexts and their origin and destination
countries, and that social and human capital positively moderate this effect. The impact is found to
be stronger for migration with non-Spanish speaking countries. The remainder of the paper is organized as follows. In Section 2, we review the background
literature and put forward our research hypotheses. In section 3, we present the data and the
econometric strategy of our empirical application. Section 4 discusses the results and Section 5
concludes. 2. Background literature and research hypotheses Economic geography has long recognized the mobility of workers among the crucial determinants
of regional economic performance, development and growth, as well as among the drivers of
regional convergence (Bairoch, 1997; Helpman and Krugman, 1985; Krugman, 1991; Fujita et al.,
1999; Combes et al., 2008). In a recent stream of research, across regional and innovation studies, a link has also been
established between the innovation performance of regions and a particular kind of migration flow,
represented by the mobility of skilled human capital and inventors towards local contexts (see, for
example, Faggian and McCann (2006) and Gagliardi (2015), for evidence on UK regions; Niebuhr
(2010) with respect to German ones; and Trippl (2013) on a global scale). The underlying argument
maintains that skilled migrants inflowing in regions embody advanced knowledge generated in their
country of origin, which they can exchange with skilled locals and set at work in the host region,
thus increasing its innovation outcomes. A wide set of studies incepted by the seminal work by Jaffe et al. (1993) has sought to investigate
the mechanisms of knowledge transmission. One stream of this literature has focussed on mapping
inter-regional knowledge flows through patent citations and the collaboration of researchers and
inventors (e.g. Maurseth and Verspagen, 2001; Paci and Usai, 2009; Quatraro and Usai, 2016), and
on highlighting their determinants and possible obstacles (e.g. Picci, 2010;Montobbio and Sterzi,
2013; Miguelez, 2016), that is, the issue at stake in the present paper. A parallel research strand has investigated the effects of proximity on knowledge flows by
focussing on the relationship between mobile inventors and patent citations (Thompson and FoxKean, 2005; Agrawal et al., 2006, 2011; Breschi and Lissoni, 2009; Singh, 2005). These studies
show two general results. First of all, spatial proximity increases the occurrence of co-inventorships
and/or patent citations by easing inventors' face-to-face interaction and their exchange of tacit
knowledge,. Second, the transaction costs associated to the mobility of inventors can be
compensated by their social proximity, as guaranteed by their common belonging to the same
disciplinary and/or professional network. Sharing common experience, mutual appreciation and
trust have a beneficial impact on knowledge exchanges. These results on the role of skilled migration in inter-regional knowledge flows can be extended,
thorough this work, in two respects. First and most relevant, a knowledge exchange between
regions could be helped by a further type of mobility, i.e. the mobility of their immigrants and
emigrants. In their relationship with the foreign investors, domestic inventors in a region could
benefit from the proximity to immigrantsfrom that country (e.g. by means of understanding new
cultural practices and preferences that lead to explore new inventions/patents, processes and
products). Secondly, in the exchange of knowledge with domestic inventors, foreign inventors
could benefit from their proximity to emigrants (e.g. through collective support, motivation, and
services). Table 1 provides a schematic illustration of the above mechanisms and of their difference
with respect to the standard analysis of skilled migration and knowledge flows. Table 1 – Migration an innovation-­‐based knowledge flows Domestic region (D)  Innovation-­‐based knowledge flow  Inventor D  Co-­‐inventorship, co-­‐
citation, … I) Standard analysis (skilled migration): Inventor F moves to D, or vice-­‐versa, Inventor D moves to F Domestic region (D)  Innovation-­‐based knowledge flow  Inventor D  Co-­‐inventorship, co-­‐
citation, …  Inventor F  Foreign country (F) Inventor F Foreign country (F)  … Inventor D  Co-­‐inventorship, co-­‐
citation, …  Inventor F II) Our analysis (broad migration): Inventors F and D do not move, while other F and/or D workers do Domestic region (D)  Innovation-­‐based Foreign country (F) knowledge flow  Inventor D  Co-­‐inventorship, co-­‐
Inventor F   citation, …    Immigrants from F Emigrants from D The impact that broad migration flows have on the innovation-based knowledge flows investigated
in this paper is hardly addressed by the literature on regional studies. However, migrants can play
an important role as facilitators of innovation-based knowledge flows, given their multiple
involvements in both their home country and the host society (Coe and Bunnell, 2003; Williams,
2007; Basch et al. 1994). Assuming that there exists a set of potential opportunities of knowledge
exchange (e.g. co-inventorships), which do not materialize because of cultural and communication
barriers between inventors (Picci, 2010), we argue that migrants contribute to the realization of
these opportunities by lowering those barriers/costs in the interaction with local inventors in loco.
For instance, migrant workers of a regional firm can interact with its local inventors and help them
in getting familiar with the research and university systems of their country of origin. This would
increase the probability that the local inventors co-invent with foreign inventors or exchange patentbased knowledge with them through citations. More generally, migrants’ familiarity with both the
host and origin cultures and languages may help home inventors to connect their innovative ideas
with those of distant inventors in their country of origin (D’Ambrosio, 2015). One could go even
further and, following Saxenian (2006), argue that transnational co-inventorship and patent citation
ties are enabled by circular migration pathways, embodying tacit, non-scientific knowledge gained
elsewhere that promotes joint innovation activities with contacts in other countries1.
The previous arguments constitute the basis of our first introductory and overarching research
hypothesis, which is:twofold: H1.a) Immigrants facilitate the occurrence of knowledge flows between their hosting
regions and their countries of origin. H1.b) Emigrants facilitate the occurrence of knowledge flows between their home regions
and their destination countries. In practice, we hypothesize that im(e)migrants act like catalysers of positive reactions
between distant elements (inventors in this case) in their knowledge exchanges and joint creations.
The two hypotheses represent our first extension of the literature on the impact of skilled migration
for knowledge flows that can be looked at in terms of the substitution between different forms of
proximity, on which evolutionary economic geography has recently concentrated (Boschma, 2005;
Ponds et al., 2007). The geographical distance between inventors can be compensated by the spatial
1
As we will say, these patterns of circular migration will not be addressed in the empirical application, as we
are only able to capture a rather simplistic dual origin-destination migration flow.. proximity between local inventors and migrants. Through spatial and social proximity to local
inventors, migrants can interact with them and transmit important procedural/tacit knowledge that
the latter can use to open an innovation-led collaboration with foreign inventors of the immigrants’
country of origin. A second contribution of our paper regardrs linking transnational knowledge flows with the social
capital of the hosting and the moving communities. This is taken as the value created by their social
networks that bond similar people and bridge diverse people (Putnam, 2000; Dekker and Uslaner,
2001; Adler and Kwon, 2002). Since the seminal contribution by Portes and Sensenbrenner (1993),
the prevailence of bridging over bonding-types of social capital of both the migrating communities
and the host society is taken to lead to greater socio-economic integration; hence, to make the host
context more permeable to the knowledge exchanged between the community-on-the-move and the
host community. This was reaffirmed by Putnam (2000) who emphasized the capacity of
open/bridging associations to promote democracy and economic integration more effectively than
inward-oriented and isolated associations. Through bridging social capital, in the form of
generalized trust and reciprocity, immigrants and the emigrants can gain easier access to localized
native in-group resources, support by other members of the local community, and rely on more
predictable transactions. Through these mechanisms, the knowledge bridging role of immigrants
and emigrants can be reinforced. This argument can be supported by the substitution/compensation
between different forms of proximity. The social capital of the hosting (moving) community adds to
the spatial proximity between regional (foreign) inventors and immigrants (emigrants) a social
proximity that can help local (foreign) inventors to bridge the physical distance to foreign (local)
inventors. An encounter of work ethics and of expectations about the cooperative behaviour of the
other parties is obviously more likely to occur if both the moving and the hosting communities are
marked by social capital of a bonding type. In brief, the impact of migrants on the occurrence of
knowledge flows could be amplified by social capital, as we state in our twofold second hypothesis: H2.a) The social capital of the host community positively moderates the impact of immigrants on
knowledge flows between host region and the immigrants’ countries of origin.
H2.b) The social capital of the moving community positively moderates the impact of emigrants on
the knowledge flows between their home regions and their destination countries. The extent to which knowledge flows between regions are connected to migrations of
communities-on-the-move .can be addressed by considering cognitive factors that may hamper the
effectiveness of the knowledge transfer at stake. The first one is language commonality. Even if we
deal with knowledge- and skill-intensive flows, sharing the same language may still ease
communication flows between local inventors and migrants, and indirectly with foreign investors
based in the migrants’ country. This leads us to our twofold third hypothesis: H3.a) Language commonality between the host regions and the immigrants’ countries positively
moderates their impact on the knowledge flows. H3.b) Language commonality between the emigrants’ destination countries and their home regions
positively moderates their impact on knowledge flows. Finally, we study the role of the migrants' human capital. On the one hand, it is immediate that a
higher level of education and training of the migrants can help reducing the cognitivedistance that
can be generally claimed to exist between them and the local inventors. On the other hand, the
cognitive distance of immigrants with foreign investors will be lower, the more skilled the
communities-on-the-move. Our last twofold hypothesis reads as follows: H4.a) A higher level of immigrants’ human capital positively moderates their impact on the
knowledge flows between their hosting regions and their countries of origin. H4.b) A higher level of emigrants’ human capital positively moderates their impact on the
knowledge flows between their home regions and their destination countries. 3. Empirical application Our empirical application covers the migration and innovation-based knowledge flows of 50
administrative provinces (NUTS3) of Spain with 73 countries worldwide over the periods 19982011 (for immigrants) and 2006-2011 (for emigrants) (see Section 3.1 for details). Spain is notably
interested by the phenomenon we are investigating. On the one hand, although with a natural
heterogeneity among them, Spanish regions have invested substantially in their research and
university systems in the recent two decades, above all by increasing their openness to external
knowledge flows (Rodriguez-Pose 2014). On the other hand, still with variation across regions, the
country combines a long-lasting experience of emigration (still growing at an average rate of 4.17%
over the 1997-2012 period) with a relatively recent boom of immigration ( growing over the same
period at an average rate of 17,9%). With the 2008-2009 crisis, the yearly growth rate of
immigration stocks stagnated, while emigrant stocks growth accelerated. These dynamics introduce
variation in the data, which we can conveniently exploit in our empirical analysis2. As shown in previous studies (e.g. Blanes Cristobal 2008; Peri and Requena, 2010; D'Ambrosio and
Montresor, 2016), Spanish provinces benefit from significant immigrants' and emigrants' network
effects with foreign countries, with a positive impact on their economic performances, especially
international trade. These results are attributed precisely to migrants' information and knowledge
bridging effect, thus providing an extra-motivation for their direct analysis. Furthermore, to a large
extent, the countries of destination of Spanish emigration can be deemed as knowledge-intensive
economies3; the countries of origin of immigrants in Spain are mainly located in the EU, Latin
America and Northern Africa. Of course, the phenomenon at stake in the paper is relevant also for
other countries, and the choice of Spain was also motivated by the availability of a dataset (see the
next Section) that the authors of this paper have built up for other previous studies of theirs
(D’Ambrosio and Montresor, 2016; Quatraro and Usai, 2016) and integrated for the present one. 3.1 Data The database for our application originates from the intersection of different data sources (see
Appendix 1 for a detailed description). The data on patent co-inventions link 50 of the 52 Spanish
provinces (excluding Ceuta and Melilla) with 73 world countries, leading to 3,650 dyads. Patent
citation data, instead, are limited to 26 EU countries only, leading to a total of 1,300 dyads, and will
be accordingly used for the sake of robustness check (Appendix 3). Table A2.1 in Appendix reports
the countries included in the analysis. 3.2 Model and variables The model we use to test our research hypotheses is a gravity model for the analysis of knowledge
flows (Maurseth and Verspagen, 2002; Paci and Usai, 2009; Picci, 2010; Maggioni et al, 2011;
Montobbio and Sterzi, 2013; Cappelli and Montobbio, 2013). The main argument for the choice of
2
In 2010, the first 15 emigration countries for Spanish nationals were Argentina, France, Venezuela,
Germany, Brazil, Switzerland, Mexico, Cuba, USA, UK, Uruguay, Belgium, Chile, Andorra, and
Netherlands (OECD DIOC data). 3
In 2010, the first 15 emigration countries for Spanish nationals were Argentina, France, Venezuela,
Germany, Brazil, Switzerland, Mexico, Cuba, USA, UK, Uruguay, Belgium, Chile, Andorra, and
Netherlands (OECD DIOC data). this model – based on the distance between the exchanging pairs and on their mass – is the wellknown feature of imperfect public good of knowledge (Arrow, 1962) and its mixed nature in terms
of codifiability and tacitness (Nelson, 1959). Given that an important part of knowledge for
innovation is tacit and localized, its diffusion across regions is hindered by their geographical and
non-geographical (in particular, cognitive, cultural and technological) distance (Criscuolo and
Verspagen, 2009; Picci, 2010, Montobbio and Sterzi, 2013). On the other hand, the regions
involved in the exchange are also marked by a “mass” of knowledge, which can work as an
“attractor” of the same knowledge flows, for example in terms of R&D, absorptive capacity, and
human capital (Montobbio and Sterzi, 2013: p.14). Given the structure of our data, the model is applied to the dyads constituted by each Spanish
province i and each foreign partner country j, in each and every year t. According to the standard
gravity logic, all factors that increase the costs of exchanging knowledge between provinces and
countries are taken to increase their, broadly defined, “distance” and to negatively affect the related
flows. Following the literature on the pro-trade effects of immigrants, and relying on their
knowledge bridging role we have hypothesized, we include migrants between each provincecountry dyad among the factors that reduce bilateral costs. Hence, our theoretical model is the
following “augmented” gravity model: Knowldge_flowijt = Gt
Massit!!!! ∗Massjt!!!! Migrantsijt!!!!
Distanceij!!
(1)
In analogy with gravity models of international trade, Gt is likely to be an “unconstant” (Baldwin
and Taglioni, 2007) capturing the bilateral costs of exchanging knowledge between province i and
all other partners different from j in the world, and between country j and all other provinces
different from i in the world in each year t; it also includes the “inventive masses” of all other
partners around the world. These factors imposing “multilateral resistance” to exchanging
knowledge need to be included in the model in order to accurately estimate the causal effect of
migrants on knowledge flows (Anderson and Van Wincoop, 2003). As we said, the knowledge flows between pairs of provinces and foreign countries, ij, are proxied
by looking at patent data. More precisely, we follow the recent evolution of the extant literature and
consider as our main dependent variable the number of co-inventorships that patents reveal between
the two geographical contexts at hand. Since the seminal work by Jaffe et al. (1993), the number of
patent citations has been for long retained the most direct proxy for the so-called knowledge
spillovers. However, following an important debate on their shortcomings(e.g. Jaffe et al, 1998;
Thomson and Fox Kean, 2005; Thomson, 2006; Belenzon and Schankerman, 2013), some recent
works have partially shifted the original focus and studied patent co-inventions as an additional
measurement (e.g. Montobbio and Sterzi, 2013; Quatraro and Usai, 2016). While they also suffer
from some limitations, co-inventorships have actually been claimed to capture a “thicker” kind of
knowledge flow than patent citations. Indeed, unlike citers, coinventors need to confront their
learning routines and practices and exchange tacit knowledge (Montobbio and Sterzi, 2013); these
are possibly more directly affected by the bridging role of migrants that we are investigating. On the basis of the previous arguments, our main estimates will refer to the log of patent coinventorships, while the log of patent citations will be used as a robustness check (Section 4.2). To
construct our co-invention measure, based on the inventors’ addresses, we selected all patent
applications where at least one inventor resides in a Spanish province and one abroad. We thus
constructed a dyadic co-invention measure, which counts the number of co-invention ties between
each province (NUTS3 region) and each country in each year between 1998 and 2011. In case of
multiple inventors, a proportional share is assigned to each region; hence, cells are not made of
integers. We consider the priority year of the patent application as the year of the co-invention,
meant to approximate the date to the invention. As far as patent citations are concerned, we both
looked at inward and outward patent citations, from country j to province i, and from province i to
country j, respectively, by aggregating citations by Spanish NUTS3 region and partner country for
every year, yielding a similar dyadic variable to co-inventorships. Coming to the independent variables, following the literature on gravity models with panel data (e.g.
Head and Mayer, 2014; De Benedictis and Taglioni, 2011; Baldwin and Taglioni, 2007), we
mainly account for the masses of the involved pairs through time-varying fixed effects. The
advantage of including fixed effects with respect to explicitly including attractor variables is that
they allow a more accurate estimation of the causal effect of our variables of interest: fixed effects
account not only for the “knowledge mass” of each partner, but also for the average amount of
knowledge it exchanges each year with the whole world (cfr. for instance Anderson and Van
Wincoop, 2003, Baldwin and Taglioni, 2007, and Head and Mayer, 2014 on trade flows). More
specifically, on the one hand, we include time-varying, country-level dummies, ψjt, which proxy for
the mass factors that typically attract knowledge in countries, like their aggregate R&D efforts,
stock of patents, quality of the educational system, as well as their GDP, along with their variations.
On the other hand, we do the same to proxy provincial knowledge attractors but, in order to avoid
problems of model saturation with time-varying, province (NUTS3)-level dummies (Bratti et al.
2014), we rather include a set of them at the (NUTS2) region (r) level, φrt. Being provinces our
focal units of analysis, however, we add to these fixed effects two variables that account for the
province-specific capacity to attract knowledge, that is, the lagged log of the provinces’ GDP and of
their stock of patent applications. In absence of more reliable disaggregated data, these can actually
be taken as proxies of the provinces’ absorptive capacity (R&D data are not available at this level)
and of their human inventive capital, respectively.4 As to distance-related regressors, we include the great-circle distance in km between each province
and the capital city of each partner country and, to account for factors other than geographical
proximity, which could influence the bilateral costs of exchanging knowledge – like, for instance,
the similarity of the technological portfolio across pairs - we add a bilateral fixed effect χrj at the
country j-NUTS2 region r level. Finally, we add the log stocks of migrants (immigrants and emigrants) between province-country
pairs. More precisely, we both consider, first alternatively and then jointly, immigrants from
country j to province i, and emigrants from province i to country j, by retaining their stock at time t
- 1. The reference to stocks, rather than to flows, derives from our interest in the knowledge impact
of communities on the move, as the above-mentioned “diaspora effects” is more likely to occur by
larger and more connected migrant communities; flows are not informative from this point of view
(Rauch and Trinidade, 2002). Taking into account the previously defined variables, we thus estimate the following empirical
model:5 ln(1 + Co-inventorshipjit) = ψjt + φrt + χrj +
+ β1*ln(GDPit-1) + β2*ln(1 + Patent_Stockit-1) +
+ β3*ln(1 + Immigrantsjit-1) + β4*ln(1 + Emigrantsijt-1) + β5*ln(Distanceij) + εjit
(2)
In order to address our first research hypotheses (H1.a and H1.b), we run different specifications of
our model. In a first set of specifications, we exclusively focus on immigrants and fully exploit the
length of our panel (1998-2011). In a second set of specifications, we only look at emigrants, as it
this shrinks the length of the panel to 5 years only, 2006-2011. In a third set of specifications, we
include immigrants and emigrants stocks jointly. This allows comparing the effects of the two
migration types over the period 2006-2011. 4
We also run our regressions substituting the stocks of patents with the stock of inventors, obtaining similar
results, available from the authors on request. 5
As usual, in order not to miss their logarithm, the variables assuming possible 0 values are
transformed by adding 1 to them. The test of our second hypothesis, about the role of social capital of the hosting and moving
communities, is carried out by adding to equation (3) the (log of the) social capital (Social_capital)
of province i at time t – 1, and by interacting the stock of immigrants (H2.a) and of emigrants (H2.b)
with a set of dummy variables representing high social capital (and with their complement to one)
so as to check whether the combination of migration and SC has an extrinsic explanatory power on
co-inventions beyond their direct and individual impact. Inclusive equals 1 if the province scores
social capital growth levels greater than the median for the year and zero otherwise. Bridging equals
1 if the country of origin scores high levels of bridging social capital, measured as the share of
respondents declaring to volunteer in “bridging” associations in the EVS/WVS, i.e. in groups
devoted to cultural activities, youth work, and ecumenical religions (e.g. Knack and Keefer, 1997;
Cortinovis et al., 2016); Bonding equals one if the share of respondents declaring to volunteer in
group-specific and “bonding” kinds of associations (professional associations, political
parties/groups, trade unions) exceeds the median. Similarly, in order to test our hypotheses about the role of language commonality, we interact both
the immigrant (H3.a) and the emigrant stocks (H3.b) with a dummy (Spanish), equal to 1 if the
partner country of the considered pairs is Spanish-speaking (either as an official language or
because Spanish is currently spoken) and 0 otherwise – and with its complement to one.6 Finally, the hypothesis on the role of the immigrants’ human capital (H.4a) will be tested by
interacting their stock with a dummy variable (Qualified), which takes value 1 if the immigrant
population is comparatively qualified (the ratio between ISCED 3-6 codes and ISCED 0-2 exceeds
the median) and zero otherwise – and with its complement to one.7 In the Appendix, Table 3 reports the summary statistics of our variables, and Table 4 the correlation
matrix. Before presenting our results, we should noted that our model faces a common estimation issue
associated with gravity modelling with panel data: on the one hand, our dependent variable is a
count variable, which would have to be treated through Maximum Likelihood Estimators; on the
other hand, our specification requires the inclusion of a large set of fixed effects, included both for
theoretical reasons and to absorb the wide heterogeneity across units. This makes it practically
6
Given that Spain is the only Spanish-speaking EU country, we could not run this analysis for inward and
outward citations, which is another reason for keeping it as a robustness check. 7
The information about the average immigrants' qualification is not available for Cyprus, Estonia, Kazakstan,
Latvia, Madagascar, Malta, Malaysia, Netherlands, Serbia, Uzbekistan in the version at our current disposal, which
reduces the observations to a maximum of 42,700 when the high-qualification variable is included. impossible to apply a Poisson Pseudo-Maximum Likelihood (Poisson PML) estimator, as many
fixed effects hamper model convergence (cfr. Bratti et al, 2014). This also rules out the possibility
to employ a Tobit model, on grounds of the incidental parameters problem (Greene, 2004). Hence,
we apply a fixed effects OLS regression and deal with the zeros by taking the natural log of the
variables to which we add one unit. This yields a log-log model where the estimated coefficients
can be interpreted as elasticities. 4. Results We begin our analysis from the base-line specification of the model, with no interacting terms.
Starting with immigration stocks, for which we have a longer time period of observation (19982010), column 1 of Table 1 displays the results of the standard gravity specifications, without
inserting the immigration variable. to enable comparisons with the specifications using patent
citations as a dependent variable, The robustness checks on the subsample of EU countries and
adopting patent citations are reported in Appendix Tables A3.1 and A3.2. In all the specifications,
the “mass-attractor” variables at the province level (GDP and stock of patents) have the expected
significant and positive effect on knowledge flows. The magnitude of the coefficients suggests a
very important role of the number of patents compared with a more generic indicator of absorptive
capacity such as the province GDP. Still in line with expectations, the coefficient of the distance
variable is significant and negative, thus supporting the choice of the gravity model for our analysis. Insert Table 1 around here Looking at column 2 of Table 1, we notice that the fit of the model increases when including
immigration stocks, as measured by both the R-squared and by the AIC statistics. The coefficient is
significant and positive, supporting our overarching hypothesis H1.a about the role of immigrants as
transnational bridges of innovation-based knowledge. According to our results, increasing the
immigrant population by 10% would increase the count of co-invention ties by approximately 0.3%. Column 3-5 focus on the period affected by the financial and economic crisis (2006-2011). Column
3 reports the results of the base-line specification looks at the effects of immigration stocks;
Column 4 includes emigration stocks, and Column 5 includes both immigrant and emigrant stocks
Again, including migration variables (both immigration or emigration) improves model fit. Emigration, too, is found to promote co-inventions, supporting also H1.b. The elasticity of coinventions to emigrationis even larger than for immigration: a 10% increase in the emigrant
population would increase EU co-invention ties by approximately 0.5%. The two coefficients
remain significant even when both variables are included jointly in column 5. Overall, these results
confirm also the arguments about the role emigrants as transnational knowledge bridgeswith respect
to innovation-based knowledge. Insert Table 2 around here Table 2 reports the results of the estimates obtained by including province-level measures of social
capital (Column 1) as well as their interactions with the Inclusive dummy and with its complement
to 1(Column 2).,. Columns 1-2 reports the results for the full period and immigration only, while
the other columns refer to the 2006-2011 period, as they include emigration too8. Notice that,
comparing Columns 2 and with the other, the effect of immigration on knowledge flows seems to
change significantly before and after 2006. To get clearer insights on this issue, in Appendix Table
A3.3, we split our sample into two and study the effects of immigration on patent co-inventorships
before and after 2006. The results confirm that the pre-2006 effect is quite large (close to 0.06) and
significant, while the post-2006 effect is smaller and, in the cases of EU countries, insignificant. First of all, the results in columns 1, 3, 4 and 5 suggest that social capital per se does not affect
knowledge flows at standard significance levels. This implies that, unlike GDP and patents, social
capital does not directly increase the mass of knowledge to be exchanged transnationally (to the
extent that our dependent variable measures it). . Instead, columns 2 and ,6-9, of Table 2 show that
social capital has a very important indirect effect on knowledge flows, which operates through the
interaction with the migration variables, with similar results arising whether one looks at
immigration or emigration. Indeed, in these specifications, the effect of immigrants and emigrants
is positive and significant in both high and low social capital provinces; however, the effects are
twice as large in provinces with high social capital, pointing to their significantly greater inclusive
capacity9. 8
Notice that the interpretation of the results is not symmetric for immigrants and emigrants. In the
latter case, the interaction of the log of emigrants with Inclusive is illustrative of the cases where the moving
community is characterized by comparatively higher social capital and possibly better ability to integrate,
similarly to the interpretation of the estimates in Table 3. The results are still in line with our expectations. 9
The result that the main effect of social capital in these specifications is negative and
significant could be interpreted in two opposing ways: as a substitution of transnational
coinventorships with domestic coinventorships (which would be in line with the prevalence of a
more bonding kind of social capital), or as a reduction of coinventorships in absolute terms.
Insert Table 3 here As regards the country-level social capital, which in interaction with immigration is a feature of the
migrating community and in interaction with emigration is a feature of the destination country, the
results provide strong support to the argument that social capital, especially if bonding, magnifies
the knowledge bridging role of migrants. In Column 1 of Table 3, we interact the immigration
variables with Bridging: the results show that the knowledge bridging role of immigrants is entirely
to be attributed to migrants from countries marked by high levels of bridging social capital. The
same result is obtained for Bonding (Column 2): when this is the only social capital variable
included in the model, high levels of bonding social capital are found to promote coinventorship,
presumably because the bonding and bridging variables are correlated. Indeed, when including both
of them, the effect of the bonding variable disappears and the entire knowledge bridging effect is to
be attributed to bridging types of social capital10. In the later sub-period, bonding types of social
capital result insignificant, and the lack of bridging social capital becomes more binding in a
negative sense. It could be argued that openness to interactions with foreign migrants benefits the
local systems not so much in terms of fostering knowledge flows as in protecting them from lock-in. Overall, an important result emerges. The positive effect of immigration and emigration on
knowledge flows is definitely activated by higher levels of social capital, pointing to a strong
economic case for promoting the integration of the immigrant communities.11 In this sense, our
second hypothesis H.2 gets confirmed. Insert Table 4 around here Turning to our third hypothesis on the language commonality, the results of the interaction term
between the Spanish variable and the immigrant and emigrant stocks contradict H3.a and H3.b
respectively. Table 4 actually shows a result that is commonly observed in the literature on the protrade effects of immigration (Girma and Yu, 2002; Dunlevy, 2006): the positive effect of migration
stocks is mainly observed for immigrants from countries speaking languages other than Spanish.
Unfortunately, with the current structure of our data, we cannot test which interpretation is more
suitable. 10
Obviously, we cannot include a measure of the country-level social capital among the attractor
variables due to collinearity with the country-time effects. 11
The case where the log of social capital equals zero and the effect of immigration or emigration is strictly
negative is actually theoretical, considering that, by the way the social capital variable is constructed, this would imply
that the level of social capital in that year is exactly equal to its level in 1983, a case that is not realized in our data. The
cases where the volume of social capital is less than in 1983 amount to a 1.15% of the total. They refer to the first two
years of our panel, 1998 and 1999 in five provinces: Ourense, Pontevedra, Zamora, Albacete, Caceres and Cordoba. Notice that this does not rule out that knowledge flows are promoted by language commonality: this
effect is absorbed by the bilateral effects. Hence, the coefficient at stake is the specific coeteris
paribus effect of immigration from Spanish-speaking countries on knowledge flows. Because of
language commonality, the barriers to knowledge flows are lower and immigrants are not necessary
to bridge them. Instead, their contribution is positive when it comes to facilitating knowledge flows
with non-Spanish speaking countries, as in this case it is actually bridging (language) barriers to
exchanging information with their home countries. Insert Table 5 around here The results of our last hypothesis (H.4a) are finally reported by Table 5, which looks at the
moderating role of human capital. As expected, higher education
reinforces the impact of
immigrants on innovation-based knowledge flows. This result is not surprising, considering that
patent co-invention obviously requires high qualifications to occur. The results are very similar
when comparing the whole sample with the EU subsample, and when comparing inward and
outward citations. 5. Conclusions This paper has adopted a gravity model to study whether migrants promote knowledge flows
(measured by patent co-inventorships and citations) between host countries (e.g. i. Spanish
provinces; ii. EU countries) and the immigrants/emigrants’ countries of origin (e.g. i. Northern
Africa or Latin America; ii. Spain) Our results, presented along with a large set of robustness
checks and heterogeneity analyses, robustly show that both immigrants and emigrants play an
important role in opening up the host local system to new knowledge flows. In most cases, the
quantitative dimension of the effect is non-negligible and is fully coherent with the findings of a
branch of the international trade literature that assumes that such an information effect exists and
promotes trade and FDI. Even if the outcome is different, the literature assumes the underlying
mechanism to be exactly the one that we propose, i.e. an information bridging effect. Considering
that our coinventorship measure reflects a rather codified kind of knowledge flows and that we
showed the importance of social and linguistic factors besides education, we are inclined to
conclude that the reported results represent lower bounds. Our results also support a strong economic case for promoting the social integration of immigrants.
Indeed, both from the side of the host economy and from the side of the migrating communities,
social capital is found to magnify the knowledge bridging effect of immigrants, suggesting that, in
connection with the mobility of people, it increases the local system's absorptive capacity. Secondly, we highlighted that the information-facilitating role is mainly to be attributed to nonSpanish speaking immigrants, who apparently bear non-redundant knowledge that can otherwise
not be accessed by the local system. This result reveals that, even when focusing on high tech
sectors such as those which lead to filing patents, where the role of vehicular languages such as
English can be expected to be widespread, barriers that can be attributed to language differences
resist and are effectively bridged by bilingual migrants. These could for instance relate to
difficulties in the translation of specific terms of the technical language. Finally, our results confirmed the important role of human capital and role of global
macroeconomic conditions affecting not only the magnitude, but also the composition of the
immigrant population, leading to a substantial reduction in the magnitude of the effect from the preto the post-2006 period. The implications of this study are straightforward. Social and innovation
policies should be integrated as social objectives seems to reinforce innovation goals. Support to
bridging types of associations and to language training both of immigrants and of natives to
improve a local system competitiveness on the global innovation scene is according to our results
not only going to promote social goals but also the capacity of the local system to innovate. These
recommendations accompany a more standard call to promote countercyclical policies attracting
skilled immigration. Indeed, the effect for highly qualified workers does not change significantly
with the crisis years, while the effect of low skilled immigration in the crisis years is more strongly
negative, suggesting that the latter categories are less resilient to the recession, at least with respect
to our outcomes of interests. The determinants and the implications of these results should be
further investigated in future extensions of this work. References Adelr P. and Kwon S. (2002), Social capital: prospects for a new concept, Academy of Management Review,
Vol.27, pp17-40. Alberdi X., Gibaja J. and Parrilli M.D. (2014), Evaluacion de la fragmentacion de los sistems regionales de
innovacion: una tipologia para Espana, Investigaciones Regionales, Vol. 28, pp.7-35. Alesina, A. and E. L. Ferrara (2000, August). Participation in heterogeneous communities.The Quarterly
Journal of Economics 115 (3), 847–904. Alesina, A. and E. La Ferrara (2002, August). Who trusts others? Journal of Public Economics 85 (2), 207–
234. Alesina, A., and La Ferrara, E. (2005). Ethnic diversity and economic performance. Journal of Economic
Literature 43(3): 762- 800. Almeida, P., Kogut, B. (1999) Localization of knowledge and the mobility of engineers in regional networks.
Management Science, 45: 905–917. Agrawal, A., I. Cockburn, and J. McHale (2006), “Gone but Not Forgotten: Labor Flows, Knowledge
Spillovers, and Enduring Social Capital,” Journal of Economic Geography, 6(5): 571-591. Agrawal, A., D. Kapur, and J. McHale (2008) “How Do Spatial and Social Proximity Influence Knowledge
Flows? Evidence from Patent Data” Journal of Urban Economics, 64, 258–269. Agrawal, Ajay, Devesh Kapur, John McHale, and Alex Oettl (2011) “Brain Drain or Brain Bank? The
Impact of Skilled Emigration on Poor-Country Innovation,” Journal of Urban Economics, 69(1), 43-55. Anderson, J. and E. van Wincoop (2003). Gravity with gravitas: a solution to the border puzzle. American
Economic Review 93, 170-92. Arrow, K. (1962) "Economic Welfare and the Allocation of Resources for Invention," NBER Chapters, in:
The Rate and Direction of Inventive Activity: Economic and Social Factors, pages 609-626 National Bureau
of Economic Research, Inc. Autio, E. (1998) Evaluation of RTD in regional systems of innovation, European Planning Studies, 6 (1998),
pp. 131–140 Baldwin, R. and D. Taglioni (2007). Gravity for dummies and dummies for gravity equations. NBER
working paper series, Working Paper 12516, http://www.nber.org/papers/w12516. Balland, P.A., R. Boschma and K. Frenken (2015) Proximity and innovation: From statics to dynamics,
Regional Studies 49 (6), 907-920. Bathelt, H., A. Malmberg, and P. Maskell (2004). Clusters and knowledge: Local buzz, global pipelines and
the process of knowledge creation. Progress in Human Geography 28 (1), 31–56. Belenzon S. and Schankerman M. (2013), Spreading the word: Geography, Policy and Knowledge Spillovers,
The Review of Economics and Statistics, July 2013, 95(3): 884–903 Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional Studies 39 (1), 61–74. Braczyk, H.-J., Cooke, P. and Heidenreich M. (Eds.), Regional Innovation Systems, UCL Press, London
(1998) Breschi, S. and Lissoni, F. (2001) Knowledge Spillovers and Local Innovation Systems: A Critical Survey.
Industrial and Corporate Change 10 (4): 975-1005 doi:10.1093/icc/10.4.975 Breschi, S. and Lissoni,F. (2006a) "Mobility of inventors and the geography of knowledge spillovers. New
evidence on US data". KITeS Working Papers 184, KITeS, Centre for Knowledge, Internationalization and
Technology Studies, Universita' Bocconi, Milano, Italy, revised Oct 2006. Breschi S., Lissoni F. (2006b) Knowledge networks from Patent data methodological issues and research
targets. In: Henk F.M., Glänzel W., Schmoch U. (eds) Handbook of quantitative science and technology
research. Springer, The Netherlands.Chesbrough, H. (2003b). The era of open innovation. Sloan
Management Review, Summer, 35–41. Breschi, S. and Francesco L. (2009) “Mobility of skilled workers and coinvention networks: an anatomy of
localized knowledge flows,” Journal of Economic Geography, 9, 439-468. Brubaker, R. (2005) The “diaspora” diaspora, Ethnic and Racial Studies, 28(1) Castles, S. (2002) "Migration and community formation under conditions of globalization." International
migration review 36.4: 1143-1168. Chesbrough, H. W., Vanhaverbeke, W., & West, J. (2006). Open innovation. Researching a new paradigm.
New York: Oxford University Press. Coe, N. and T. Bunnell (2003). Spatialising knowledge communities: Towards a conceptualisation of
transnational innovation networks. Global Networks 3, 437–456. Cohen, W. and D. Levinthal (1990). Absorptive capacity: A new perspective on learning and innovation.
Administrative Science Quarterly 35, 128–152. Cooke, P. , Boekholt, P. , Tödtling, F. (2000) The Governance of Innovation in Europe, Pinter, London. Cortinovis, N., Xiao, J., Boschma, R. and van Oort, F. (2016). Quality of government and social capital as
drivers of regional diversification in Europe. Papers in Evolutionary Economic Geography # 16.10.
http://econ.geog.uu.nl/peeg/peeg.html Criscuolo, P. and Verspagen, B., 2008. "Does it matter where patent citations come from? Inventor vs.
examiner citations in European patents," Research Policy, 37(10), 1892-1908. Bairoch, P. (1997) Autour de l'histoire sociale du temps. Zurich : Chronos Verlag. Bratti, M., De Benedictis, L. and Santoni, G. (2014). "On the pro-trade effects of immigrants," Review of
World Economics (Weltwirtschaftliches Archiv), Springer; Institut für Weltwirtschaft (Kiel Institute for the
World Economy), vol. 150(3), pages 557-594, August. Cappelli, R. and F. Montobbio (2013). European Integration and Knowledge Flows across European Regions.
Technical report, Department Cognetti de Martiis Working Papers 201322. Combes P., Mayer, T. and Thisse, J.F. (2008) "Economic Geography: The Integration of Regions and
Nations," Princeton University Press Cooke, P. (2005) Regionally asymmetric knowledge capabilities and open innovation. Exploring
‘Globalisation 2’—A new model of industry organisation. Research Policy 34 1128–1149 Cortinovis, N., Xiao, J., Boschma, R. and van Oort, F. (2016) Quality of government and social capital as
drivers of regional diversification in Europe, Papers in Evolutionary Economic Geography # 16.10, Utrecht
University Dasgupta, P., David, P. (1987) Information disclosure and the economics of science and technology. In G.
Feiwel (ed.) Arrow and the Ascent of Modern Economic Theory. New York, NY: New York University
Press. D'Ambrosio, A. (2015) Migration flows and local systems of production: New comparative evidence on Italy
and Spain. Doctoral thesis, University of Trento. D'Ambrosio, A. and Montresor, S. (2016). Migration and trade flows: new evidence from Spanish provinces.
mimeo. De Arcangelis, G., Di Porto, E. and Santoni, G. (2013), Migration, Production Structure and Exports.
http://dx.doi.org/10.2139/ssrn.2249943 De Benedictis, L. and Taglioni, D. (2011) The Gravity Model in International Trade, in De Benedictis, L.
and Salvatici, L. (eds.) “The trade impact of European Union preferential policies: An Analysis through
gravity models”, Springer, pp.55-89, 2011. Dekker, P. and Uslaner, E.M. (2001). ‘Introduction’ in “Social Capital and Participation in Everyday Life”,
Uslaner, E.M., London: Routledge, 1-8. Dunlevy, J. (2006). The impact of corruption and language on the pro-trade effect of immigrants: Evidence
from the American States. Review of Economics and Statistics 88 (1), 182-186. Faggian, A. and McCann, P. (2006) Human capital flows and regional knowledge assets: a simultaneous
equation approach Oxf. Econ. Pap. (2006) 58 (3): 475-500. doi: 10.1093/oep/gpl010) Felbermayr, G., V. Grossmann, and W. Kohler (2012, October). Migration, International Trade and Capital
Formation: Cause or Effect? IZA Discussion Papers 6975, Institute for the Study of Labor (IZA). Fernandez, J., Pérez, F., and Serrano, L. (2015) Crisis Económica, Confianza y capital social. Bilbao,
Fundación BBVA. http://www.fbbva.es/TLFU/dat/DE_2015_IVIE_crisis_economica.pdf Fujita M, Krugman P, Venables A J, (1999) The Spatial Economy: Cites, Regions and International Trade
(MIT Press, Cambridge, MA) Gagliardi, L. (2015) Does skilled migration foster innovative performance? Evidence from British local areas.
Papers in Regional Science, 94: 773–794. doi: 10.1111/pirs.12095. Girma, S. and Z. Yu (2002). The link between immigration and trade: Evidence from the United Kingdom.
Weltwirtschaftliches Archiv 138, 115–30. Gould, D. M. (1994). Immigrant links to the home country: Empirical implications for U.S. bilateral trade
flows. The Review of Economics and Statistics 76 (2), 302-316. Granato, J., Inglehart, R., and Leblang, D. (1996b). The effect of cultural values on economic development:
theory, hypotheses, and some empirical tests. American Journal of Political Science, 40(3):607–631. Granovetter, M. (1973) The strength of weak ties. American Journal of Sociology. LXXIII: 1360–1380. Greene, William, (2004), Fixed Effects and Bias Due to the Incidental Parameters Problem in the Tobit
Model, Econometric Reviews, 23, issue 2, p. 125-147 Head, K. and T. Mayer (2014). Gravity equations: Workhorse, toolkit, and cookbook. In G. Gopinath, E.
Helpman, and K. Rogo (Eds.), Handbook of International Economics. Elsevier. Helpman, E., and Krugman, P. (1985) Market Structure and Foreign Trade. Cambridge, MA:MIT Press. Jaffe, A., Trajtenberg, M., Henderson, R. (1993) Geographic localization of knowledge flows as evidenced
by patent citations. Quarterly Journal of Economics, CVIII: 577–598. Jaffe, A.B, Fogarty M.S and Banks B.A., (1998) "Evidence from Patents and Patent Citations on the Impact
of NASA and Other Federal Labs on Commercial Innovation," Journal of Industrial Economics, 46(2), 183205,. Lewis, E. “Immigration and Production Technology,” Annual Review of Economics 5, August 2013. Knack, S. and Keefer (1997). Does social capital have an economic payoff? A cross country investigation.
The Quarterly Journal of Economics, 112:1251–1288. Krugman, P. (1991)"Increasing Returns and Economic Geography", Journal of Political Economy,
University of Chicago Press, vol. 99(3), pages 483-99. Maggioni, M., T. Uberti, and S. Usai (2011). Treating patents as relational data: knowledge transfers and
spillovers across Italian provinces. Industry & Innovation 18, 39–67. Manacorda, M., Manning A., and Wadsworth, J.. "The impact of immigration on the structure of wages:
theory and evidence from Britain." Journal of the European Economic Association 10.1 (2012): 120-151. Maurseth, P. and B. Verspagen (2002). Knowledge Spillovers in Europe: A Patent Citations Anal- ysis.
Scandinavian Journal of Economics 104(4), 531–45. Miguélez, E. (2016) Inventor Diasporas and the Internationalization of Technology, The World Bank
Economic Review, doi 10.1093/wber/lhw013. Montobbio, F. and Sterzi, V.A., 2013. "The Globalization of Technology in Emerging Markets: A Gravity
Model on the Determinants of International Patent Collaborations," World Development, Elsevier, vol. 44(C),
pages 281-299. Murat, M. and B. Pistoiesi (2009). Migrant networks: Empirical implications for the Italian bilateral trade.
International Economic Journal 23 (3), 371–390. Nelson, R. (1959) "The Simple Economics of Basic Scientific Research," Journal of Political Economy,
University of Chicago Press, vol. 67, pages 297. Niebuhr, A. (2010). Migration and innovation: Does cultural diversity matter for regional R&D activity?
Papers in Regional Science 89, 563–585. Ottaviano, G.I.P, and Peri, G.. "Rethinking the effect of immigration on wages." Journal of the European
economic association 10.1 (2012): 152-197. Paci, R. and S. Usai (2009). Knowledge flows across the European regions. Annals of Regional Science 43,
669–690. Paci, R. and Usai, S. (2000) "Technological Enclaves and Industrial Districts: An Analysis of the Regional
Distribution of Innovative Activity in Europe," Regional Studies, 34(2), 97-114. Parrilli M.D. (2012), Heterogeneous social capital: a window of opportunity for local economies, in Cooke
P., Parrilli M.D. and Curbelo J.L., Innovation, global change and territorial resilience, Edward Elgar. Putnam R. (2000), Bowling alone: the collapse of the American community, Simon and Schuster, New York. Peri, G. and F. Requena-Silvente (2010). The trade creation effect of immigrants: Evidence from the
remarkable case of Spain. Canadian Journal of Economics 43 (4), 1433-1459. Peri, G. and Sparber, C.. 2009. "Task Specialization, Immigration, and Wages." American Economic Journal:
Applied Economics, 1(3): 135-69. Pérez, F., Montesinos, V., Serrano, L., and Fernández de Guevara, J. (2005). La Medición del Capital Social:
Una Aproximación Económica. Bilbao, Fundación BBVA. Picci, L. (2010). The internationalization of inventive activity: A gravity model using patent data. Research
Policy 39, 1070–1081. Ponds R., Van Oort F., Frenken K. (2007) The geographical and institutional proximity of research
collaboration. Papers in Regional Science 86: 423–443 Portes, A. (2000). Globalization from Below: The Rise of Transnational Communities. In Kalb, D. van der
Land, Starting, R., van Steenbergen, B. and Wilterdink, N., “The Ends of Globalization: Bringing Society
Back in”, Rowman & Littlefield, Lanham. Portes, A. and J. Sensenbrenner (1993). Embeddedness and immigration: Notes on the social determinants of
economic action. American Journal of Sociology 98, 1320–50. Quatraro, F. and S. Usai (2016). Are knowledge flows all alike? Evidence from European regions.
Forthcoming in Regional Studies. Rodriguez-Pose, A. (2014), Leveraging research, science and innovation to strengthen social and regional
cohesion. Policy Paper by the Research, Innovation, and Science Policy Experts (RISE). EUR 27364 EN. Rauch, J. and V. Trinidade (2002). Ethnic Chinese networks in international trade. Review of Economics and
Statistics 84 (1), 116-30. Saxenian, A. (2006). The new Argonauts: Regional advantage in a global economy. Cambridge, MA:
Harvard University Press. Singh, J. (2005) Collaborative networks as determinants of knowledge diffusion patterns. Management
Science, 51: 756–770. Sonderegger P. and Taube F. (2010), Clster lifecycle and diaspora effects: evidence from the Indian IT
cluster in Bangalore, Journal of International Management, Vol.16, pp.383-397. Tödtling, F. and Trippl, M. (2005) One size fits all?: Towards a differentiated regional innovation policy
approach, Research Policy, 34(8), 1203-1219. Thompson P., (2006). "Patent Citations and the Geography of Knowledge Spillovers: Evidence from
Inventor- and Examiner-added Citations," The Review of Economics and Statistics, 88(2), 383-388, Thompson P. and Fox-Kean M., (2005). "Patent Citations and the Geography of Knowledge Spillovers: A
Reassessment," American Economic Review 95(1), 450-460, Tortosa-Ausina, E. and Peiró, P. (2012). "Social Capital, Investment and Economic Growth: Evidence for
Spanish Provinces," Working Papers 2012122, Fundacion BBVA / BBVA Foundation. Vertovec,S. (1999) "Conceiving and researching transnationalism." Ethnic and racial studies 22.2: 447-462. Wagner, D., K. Head, and J. Ries (2002). Immigration and the trade of provinces. Scottish Journal of
Political Economy 49 (5), 507-525. Williams, A. (2007). Listen to me, learn with me: International migration and knowledge transfer. British
Journal of Industrial Relations 45, 361–382. Zak, P. and Knack, S. (2001). Trust and growth. The Economic Journal, 111(470):295–321. Appendix 1. Data sources Innovation-based knowledge flows involving Spanish provinces and foreign countries have been
investigated by using data on patent co-inventions and citations from the OECD REGPAT and the
OECD Citations database, covering patent applications filed to the European Patent Office (EPO)
and under the Patent Cooperation Treaty (PCT). As far as the analysis of migration is concerned, provincial data on residents with a foreign
citizenship in Spain have been taken from the official Spanish register of residents (the “Padron
Municipal”), publicly available from the Spanish Institute of Statistics (“Instituto Nacional de
Estadistica, INE”). The records include data on residents with foreign nationality and are not
matched with the register on the stay permit, hence they may picture also part of the irregularly
residing foreigners. Provincial data on emigrants are drawn from the “Censo Electoral de
Residentes Ausentes”, CERA, as in Murat and Pistoiesi (2009), i.e. from electoral registers of
Spanish nationals who reside abroad but maintain their voting rights in Spain. As regards the social capital of the migrating communities, we integrated two waves of the
European Value Survey with four waves of the World Value Survey (Cortinovis et al., 2016;
Granato et al., 1996a; Knack and Keefer, 1997; Zak and Knack, 2001) to be able to cover all years
and countries under consideration. Reliance on these data implies that we assume that the countrylevel social capital is a good proxy for the social capital of the migrating communities, and that the
country's level of social capital does not change significantly within the waves. These data cannot
be employed to measure province-level social capital, as they available at this level of
disaggregation. Hence, to estimate the social capital of the hosting communities we used the
measures of social capital volume provided by the Ivie12 study (Fernandez et al 2015; Pérez et al.,
2005; Tortosa-Ausina and Peiro, 2012). In this study, social capital is mainly construed as trust and
is modelled as an asset in which people can deliberately invest; this investment is costly and entails
a cooperation effort; an individual decides whether to undertake the investment considering the
expected duration of this effort as well as the present value of future gains. The gains from investing
in social capital, in turn, are assumed to vary by regions (province in our database) with the degree
of connectivity of social networks, the inequality in society, and with the extent to which an
individual's social capital decisions affect the behaviour of others (see Fernandez et al. 2015, for
details). The resulting estimate of social capital volume provides an index (base year is 1983) of the
within-province variation of social capital over time, which suits our empirical analysis.
12
Instituto Valenciano de Investigaciones Económicas (www.ivie.es), in collaboration with the Fundacion
Banco Bilbao Vizcaya (FBBVA, www.fbbva.es). Considering that that overall levels of trust are likely to be higher in more integrated and less
polarized societies (Alesina and La Ferrara, 2000, 2002), it seems safe to conclude that this measure
pictures a more “bridging”, rather than “bonding”, kind of social capital. The data on the human capital of migrants are drawn from the OECD-DIOC database, which
provides aggregate (and cross-sectional, referring exclusively to 2007) information on the
immigrants’ qualification by country of origin, by referring to the International Standard
Classification of Education (ISCED) codes developed by UNESCO to evaluate the quality of the
world countries’ education systems. Unfortunately, data availability prevents us from building up a
reliable measurement of the human capital qualifications of the Spanish emigrants by destination
countries, so that our last hypothesis (H.4) will be tested only with respect to the immigrants’
qualification by country of destination (H.4a). Finally, all the other data required by the selected econometric strategy (see Section 3.2) – like
those on the GDP of Spanish provinces - come from publicly available data on the INE website. Appendix 2. Descriptives Insert Tables A2.1, A2.2, and A2.3 here Appendix 3. Robustness checks In Table A3.1 we report our robustness checks for the test of H1.a, with respect to immigrants over
the period 1998-2011, by restricting the analysis to EU countries and using patent citations instead
of coinventorships as a dependent variable. The accuracy of the gravity model in accounting for the
knowledge flows at stake is confirmed, as well as the positive impact of immigrants on patent
citations, but with a smaller elasticity: a 10% increase in immigration stocks would increase the
number of citations by slightly less than 0.1%, about a half the effect found for coinventorships.
This is quite interesting and actually confirms our choice of focusing on coinventorships, as
expression of thicker knowledge flows, also intensive of tacit knowledge and thus more
substantially affected by immigration than patent citations. It is worth noticing that the coefficient
of immigration stocks on inward and outward citations is very similar, although it is only significant
at a 10% level for outward flows. Insert Table A3.1 around here Still with respect to patent citations (and with EU-level coinventorships for the sake of
comparability), Table A3.2 reports the estimates that include both immigrants and emigrants, on the
sub-period 2006-2011. Both H1.a and H1.b get supported. Furthermore, the results confirm a much
larger role of emigrants than immigrants in promoting inward and outward citation flows between
the origin province and the destination country. The impact on citations (either inward or outward)
is, again, much smaller than the one found for coinventorships, still confirming a larger effect of
emigration on more tacit kinds of knowledge flows. Insert Table A3.2 around here Table A3.3 shows more explicitly that the effects of immigration and emigration are affected by
macroeconomic conditions and change before and after 2006, using the log of coinventorships
along with inward and outward citations as dependent variables. The results are qualitatively similar
to those obtained for coinventorships, but the effects are still generally smaller. Once more, the
effect of immigration on all types of knowledge flows is positive and significant before 2006, and
much smaller and often statistically insignificant after 2006. Tables A3.4 and A3.5 show that the
results obtained, using coinventorships, about the moderating role of social and human capital,
respectively, are robust with respect to the analysis of patent citations.13 H2.a and H2.b, as well as
H.4a are still supported. To be sure, in the latter case an exception is found as, while low-qualified
workers do not significantly affect coinventorships, they are found to decrease patent citations. A
tentative explanation might be that regions employing larger amounts of non-qualified workers are
more likely to specialize in sectors that do not lead to patents. Insert Table A3.4 around here Insert Table A3.5 around here 13
As we said, the robustness check with respect to language commonality is not feasible with respect
to patent citations.