Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/de-gruyter/innovation-scaling-up-and-local-development-in-peripheral-regions-do-50di9vQlsG
- Publisher
- de Gruyter
- Copyright
- © 2022 the author(s), published by De Gruyter.
- ISSN
- 0044-3751
- DOI
- 10.1515/zfw-2022-0028
- Publisher site
- See Article on Publisher Site
Abstract
1IntroductionSince the 1980s regional development theories and policy have increasingly focussed upon innovation to explain differences in regional economic trajectories (Landabaso 1997; Sternberg 2000; Cooke et al. 2011; OECD 2011; 2020; Government of Canada 2018). However, a growing number of researchers are questioning whether local innovation is sufficient to bring about local development (Marques and Morgan 2021; Iammarino et al. 2017; Shearmur 2016; Shearmur and Bonnet 2011): whilst innovation is necessary for development to occur, these writers suggest it is not sufficient.The empirical question addressed in this paper is situated within this debate. Whereas the reasons why local innovation is not necessarily connected to local development have been discussed from an institutional (Marques and Morgan 2021; Morgan 2019; Iammarino et al. 2017) and firm-level (Shearmur 2016) perspective, the evidence brought forward has been high-level. Marques and Morgan (2021) illustrate a disconnection between regional productivity and employment growth, Shearmur and Bonnet (2011) reveal that patenting is not directly connected to local employment or income growth, and Iammarino et al. (2017) discuss theoretical limitations of the idea that innovation and knowledge spillovers will necessarily lead to regional development.In this paper we empirically examine a specific mechanism that could contribute to the disconnection between local firm-level innovation and local growth: the mobility of innovators. ‘Mobility’ is understood widely: an innovative establishment can move elsewhere, but it can also expand non-locally or can call upon non-local resources as it innovates (Shearmur 2015). In all these cases the locality benefits less from the establishment’s innovation than if it remained local, expanded locally, or turned to local suppliers.Our questions are as follows: do innovators in peripheral regions tend to use non-local service-providers when they expand capacity? When these innovators increase their capacity beyond the confines of the establishment, does this occur outside the region? Whatever the processes thought to link local innovation with local development, these questions are important: if innovators in peripheral regions tend to collaborate more with external service partners when they are expanding, and if they tend to scale-up outside the region, this would limit their local development impact – in terms of services, jobs, local spending and milieu effects.Our results, which draw upon a survey of 632 manufacturing establishments, reveal a significant reliance by innovators in the periphery on non-local services and a significant, though less robust, tendency for non-local capacity expansion. The nature of our data is important: robustness issues, attributable to the small numbers of establishments which expand capacity (OECD 2021), may explain why this process has been overlooked in a research agenda that consistently seeks out robustness, thereby maybe overlooking more subtle trends.If the processes exemplified by the small number of capacity expansions in this cross-sectional analysis are indicative of persistent long-term processes (as suggested by wider debates and high-level evidence), then our results – consistent with theory – provide empirical clues, at the level of the firm, as to why local development is disconnected from firm-level innovation in peripheral areas. In a context where innovation policy increasingly focuses on the scaling-up and growth of innovators (ISED 2019; OECD 2021) our results suggest that a more concerted effort is warranted to gather longitudinal data to examine the mechanisms linking local innovation, firm expansion, and local development.The paper is structured as follows. We begin by discussing two approaches to understanding the connection between local firm-level innovation and local development. First, the innovation systems approach, which considers local firms as embedded in local systems: these cultural and institutional settings both support innovative firms and anchor them locally. Second, the new economic geography approach, which considers firms as individual and rational actors, choosing where to locate on the basis of market-driven rationales: local anchoring, to the extent it exists in NEG, is the outcome of economic calculation and of local dynamics that accrue from self-reinforcing agglomeration processes. Then, after describing the data, we examine the extent to which innovative establishments are locally embedded by exploring whether they expand locally and use local services when they expand. Whereas metropolitan firms tend to expand locally, within metropolitan regions, and resort to metropolitan service-providers, firms in the periphery expand towards metropolitan areas and turn to non-local service-providers. To conclude, we discuss how development policies premised upon local innovation should carefully consider what happens after firms have innovated and begin scaling-up.2Innovation-led local development: theory or policy objective?2.1On what basis is it expected that firm-level innovation and local development are linked?An assumption that underpins many local development policies (OECD 2011; 2020), especially in North America (Government of Canada 2018; OECD 2019), is that if local firms are innovative, they will capture new markets and grow, leading to development of the locality.However, it is not always clear upon what conceptual basis the link between local firm-level innovation and local development is made. Endogenous growth theory (Romer 1994), which has gained wide acceptance, posits that within closed economic systems innovation – resulting from new collaborations, new combinations of ideas, processes and technologies – drives growth. However, regions are not closed systems: they are made up of multiple (mobile) actors, connections with external actors, as well as embedded institutions and cultures.Two broad bodies of theory bridge the gap between endogenous growth theory and regional development. On the one hand, regions are approached as entities that are culturally and institutionally distinct. Some local cultures and institutions are more open to and supportive of innovation than others. The literature on regional innovation systems (RIS) describes how market-oriented (firm-level) and non-market (community-level) development inter-connect (Marques and Morgan 2018; Bathelt and Henn 2017; Doloreux and Porto, 2017; Cooke et al. 2011): a locality with well-developed institutional and social capacities will not only enable firms to innovate, but will also prime the region for capturing the benefits of this innovation. Such internal processes can also prime regions to harness external resources potentially allowing regions to incorporate such resources into new growth paths (Trippl et al, 2018). From this perspective local firms cannot be separated from the local system: they are constituent elements of it, even though certain firms may be more embedded than others.RIS focuses on culture, actors, and institutions within the region. A well-functioning system, which can be stimulated by policy, will lead to endogenous development: it is local dynamics and actors (including their capacity to marshal non-local networks and resources) that lead to local innovation, which then reinforces these local dynamics (Marques and Morgan 2018; Cooke et al. 2011): in theory at least, the system should both enhance local innovation and anchor innovators to the locality, allowing them to expand locally. From this perspective, the absence of a generalised connection between local firm-level innovation and local development can be attributable to the fact that some regions, even if they house innovators, are not institutionally thick enough to anchor them. Peripheral and lagging regions – which are not to be confused, but which overlap and face similar challenges – often lack the institutional thickness to support innovators as they scale-up (Glückler et al, 2022; Marques & Morgan, 2021)On the other hand, local development can be understood as the outcome of market-driven decisions taken by individual actors. Myrdal (1957) describes the link between regional growth and the location choices of mobile factors of production. As a region grows, agglomeration economies develop: these economies can be static (associated with the division of labour and economies of scale) or dynamic (associated with knowledge exchange, diversity and innovation – often referred to as Marshall-Arrow-Romer economies). Workers and firms are attracted to more prosperous and innovative regions, which continue to grow, further strenghtening agglomeration economies. More recently, Krugman (1991) formalised this process within the context of new economic geography (NEG).NEG, in contrast to RIS, treats actors and resources as essentially mobile and disembedded: firms and workers arbitrate between regions based on opportunities, which grow in a self-reinforcing way within agglomerations (Behrens and Nicoud 2011). For NEG, regional dynamics are the outcome of micro-level arbitrations rather than of cultural and institutional processes.Neither of these approaches has much to say about innovation in peripheral or laggard regions: it is only certain regions – either those institutionally primed (Coenen and Morgan 2020; Morgan 2019) or those where dynamic agglomeration economies have emerged (Krugman 1991; Behrens and Nicoud 2011) – that will benefit from the innovativeness of local firms. Even though firms innovate in the periphery (Glückler et al, 2022), it is unclear how they can prosper there, scale-up, or generate local development dynamics.Indeed, even at the national level the connection between economic innovation and wealth or job creation is tenuous. A panel that recently explored innovation and R&D in Canada concludes that“Canada is a highly innovative nation, but significant barriers prevent the translation of innovation into wealth creation […] Canada’s promising start-ups are often acquired and developed in other countries, leading to a loss of economic and commercial benefits” (Council of Canadian Academies 2018: xxiii-xxiv).In short, the idea that local firm-level innovation is connected to local development may be valid in certain contexts. However, peripheral or laggard regions are less likely to develop, even if they house innovative firms, than larger, more central and more institutionally thick regions Whilst it is possible that institutionally thin regions can leverage external resources in such a way that local development processes emerge (Morgan 2019; Trippl et al, 2018), such regions are often unable to do so (Marques & Morgan, 2021). Innovators in institutionally thin regions, especially if they are geographically distant from major markets, are more likely to behave as NEG predicts, seeking out conducive environments within which to expand (Shearmur 2015).2.2The rise of local development strategies premised upon firm-level innovationSince there is no compelling theoretical reason to assume that local firm-level innovation necessarily leads to local development, the question remains as to why this assumption is so often implicit in local development policy. This can be explained partly by the policy objective of linking local firm-level innovation with local development – an objective that has morphed into an expectation – and partly by the misapplication of endogenous growth theory to small open regional economies. We briefly trace these two intertwined possibilities below.In the 1980s researchers such as Massey (1985) and Storper and Walker (1989) described the effects of growing globalisation and nascent financialization. Multi-establishment corporations were increasingly able to close specific establishments in a strategic fashion, effectively moving capital from region to region, or across national borders. This geographic arbitration was in part motivated by economic factors (lower costs, more collaborators, proximity to markets …) of the type outlined by NEG, but was also motivated by a desire to discipline workers. Certain types of local culture, notably workplace organisation, were perceived by owners of capital as an impediment to profit maximisation. Capital mobility, or sometimes just the threat of mobility, was a way to lower overall production costs and reduce disruption (Massey 1985).In reaction to this, the embedding of economic activity, aligning productivity and innovation with local culture and institutions, was seen as a way to counter socially destructive capital flows (Cox 1995; Storper and Walker 1989). Thus, local endogenous development strategies should be distinguished from endogenous growth theory: endogenous strategies make political and policy sense because local growth is not predicted by endogenous growth theory (Margarian 2011; Shearmur 2016): local endogenous development strategies – such as those suggested in the RIS literature – are required because local development is not a necessary outcome of firm-level decision making (Morgan 2019).Local policies focussing on entrepreneurship and firm-level innovation remain popular, particularly in North America, because they are attractive to researchers, policy makers and politicians from across the political spectrum. From a more left-wing or communitarian perspective, encouraging local firms to innovate can be seen as an act of resistance vis-à-vis globalisation and capital mobility (Massey 1985), and also harks to social-anarchist ideas of localised decision-making and participation (Marshall 1991). From a more right-wing or neo-liberal perspective, encouraging localities and regions to support local entrepreneurs absolves central government from intervention, is market-oriented, and legitimises the idea of competitiveness, i. e., that regions are competing against one another (Eisinger 1988; Stimson et al. 2011) as opposed to being part of an interdependent geographic system of production (Coffey 1998).Whereas the European Union (EU) proposes national and continental approaches to regional development that are not solely focussed on firm-level innovation (Marques and Morgan 2018), the absence of regional or local development policy frameworks in North America has encouraged a more fragmented, locally-driven and firm-focussed, approach (OECD 2019; Vodden et al. 2019). Entrepreneurship – the cultivation of innovative local small firms – is a keystone of many locally devised and implemented approaches to development both in the North America (Vodden et al. 2019; Malecki 2017) and Europe (Spigel 2016).2.3Firm-level innovation and local development: processes, evidence and (some) anecdotesGenerating RIS rich enough to embed actors is challenging, particularly in low-density and remote regions (Glückler et al. 2022; Morgan 2019). Shearmur (2015) argues that innovators in such places will not necessarily remain local once they have innovated. He suggests that, once an innovation is introduced, innovators who wish to scale-up will tend to expand capacity (either by moving or by opening subsidiaries) into regions more conducive to the development and marketing of their innovation (a process similar to that identified by the Council of Canadian Academics (2018) at the international scale).Another way the impact of innovation would not be felt locally is if innovators turn to external services and suppliers when they expand (Shearmur 2015): indeed, Pritchard et al (2012) document “the flight of high-order functions to larger regional centres [which] over time truncates the functional capacities of small-town economies” (p554). This occurs partly as a consequence of falling local demand, but is also attributable to the higher variety and specialisation of urban services relative to those situated in smaller markets: urban services are increasingly accessible to firms in the periphery thanks to internet (MacPherson, 2008). A past generation of regional analysts would have said there are fewer, and diminishing, local multiplier effects in sparse and peripheral regions (Archibald 1967).There exists some correlation-based evidence of these processes, resting on the analysis of local patents (Shearmur and Bonnet 2011) or local productivity (Marques and Morgan 2021) and their (dis)connection with local income and employment growth. However, there is so far no direct and systematic evidence of the firm-level processes suggested by Shearmur (2015). Rather, there is anecdotal evidence: but anecdotes work both ways and do not allow weak but persistent long-term trends to be discerned.On the one hand, there are innovators such as Emil Gulbranson in Vanderhoof, British Columbia, a village of 4400 people 100 km west of Prince George, a town of 75,000, itself 800 km north of Vancouver. Vanderhoof and its region had a hay surplus: in the early 2010s the BC Forage council explored export markets, such as China (Dobb 2014). Gulbranson founded Agri Green Enterprises Inc. in 2017 to produce and market an innovative hay dryer (Jaynes 2018), which has found buyers in Europe, Africa and across the USAhttps://www.facebook.com/pg/haydryer/posts/. There is no evidence that this firm has expanded beyond the region: on the contrary, Vanderhoof seems to be developing endogenously as the agricultural sector innovates both in terms of technology and in terms of marketing. Agri Green still operates at a small scale: should it grow, it will be interesting to see if it can continue to operate locally.On the other hand, there are innovators such as Suzanne Belisle and her colleagues in Chibougamau, a town of 7500 people, 700 km to the north of Montreal and 450 km from the closest larger town. These nurses observed that many children, particularly infants, were in considerable pain when intravenous catheters are inserted. Mme Belisle designed and developed a lightweight and flexible splint that holds the catheter in place whilst allowing babies to move freely and without pain: it has now been patented (Larochelle 2021). However, to market the product within the hospital network, and to obtain regulatory approval, the Chibougamau-based firm (Equipment FIX I/V) reached beyond Chibougamau, establishing a presence in Montreal via a specialised distributor, Dufort et Lavignehttps://www.dufortlavigne.com/pages/qui; https://www.fix-iv.com/en/place-an-order/. There is a sizeable market for such splints, since each infant on intravenous drip or medication requires one.Both these innovators, whose markets extend well beyond their local communities, contribute to local development in certain ways (such as by example to other entrepreneurs): however, their impact in terms of jobs and income is not necessarily local. Agri Green Enterprises, seems, for the time being, to be locally embedded by virtue of its raw material (hay), knowledge-base (agriculture) and small but active local milieu: a local innovation system seems to have emerged. Equipment FIX I/V, on the other hand, has no particular reason to remain rooted in Chibougamau: already its regulatory and distributional activities are feeding employment and income in Montreal because the city provides proximity to Québec markets, access to international markets, and access to regulators.In this paper we explore whether innovators in peripheral regions, which face many of the same challenges as ‘lagging’ regions (Glückler et al, 2022; Morgan, 2019), have a stronger tendency than establishments located closer to (or within) metropolitan areas to scale-up their operations beyond their local area, and/or to use service-providers outside their local area when they expand. Such geographies are difficult to track: unlike patent applicants – for which formal databases exist that geo-localise applicant firms as well as inventors – the locations of innovators’ service-providers, new facilities, new subsidiaries, or new offices are not recorded. These data therefore need to be gathered by way of a survey: the data and method will be described in the next section.3Data and method3.1DataThe data analysed in this paper are from an original survey of 632634 responses were obtained. Two are dropped because no information is provided on expansion and service-use. Furthermore, not all establishments provide detailed enough geographic information for each dimension we analyse: the analyses run on 628 to 631 observations. small and medium sized manufacturing establishments (SMEs) in the province of Québec, drawn from the population of Québec manufacturing establishments held, and updated on a rolling basis, by the CRIQ (Centre de Recherche Industrielle du Québec). These are establishment not firm level data. The 6797 establishments identified by CRIQ are first divided by location into metropolitan (establishments located within the Montreal and Québec urban agglomerations), central (those located within about 100km of Montreal and Québec cities) and peripheral. The sample is a stratified random sample: approximately equal representation of metropolitan, central and peripheral establishments was sought, but within these geographic categories sampling is random (see Table 1 for descriptive statistics).Table 1: Descriptive statistics Location of establishmentsMetropolitan* regionCentral regionPeripheral regionTotalNumber of observations145135101632Internal capacity expansion72 %74 %76 %74 %External capacity expansion 14 %11 %7 %11 %Service-use for expansion40 %45 %46 %43 %New-to-market innovation Any new-to-market innovation**59 %64 %58 %60 %Product innovation47 %45 %45 %46 %Process innovation32 %34 %31 %33 %Organizational innovation25 %20 %26 %23 %Marketing innovation 19 %22 %19 %20 %Pavitt sectors*** Resource based 23 %40 %37 %29 %Labour intensive43 %33 %23 %39 %Scale Intensive49 %29 %23 %22 %Specialised & Science43 %32 %25 %11 %Size Mean size (employees)293433321–1034 %31 %35 %33 %11–2023 %24 %23 %23 %21–5029 %27 %25 %27 %51 and more 14 %19 %17 %16 %* Metropolitan areas consist of Montréal and Québec census metropolitan areas. The next largest cities in the province are Saguenay and Sherbrooke census agglomerations (CAs), both with about 150,000 people, followed by CAs such as Rimouski and Rouyn-Noranda, 50,000 and 30,000 respectively. We have deliberately not sampled establishments in the vicinity of Ottawa, and have ensured that sampled non-metropolitan establishments come from a variety of non-metropolitan regions in Québec.** The propensity to introduce a new-to-market innovation does not differ by type of region (chi2=0.88, p(chi2=0)=0.675)*** The distribution of resource-based establishments across locations differs significantly from the distributions of the three other sectors (chi2=25.8, p(chi2=0)<0.0001, DF=3), which do not differ significantly between each other (chi2=1.3, p(chi2=0)=0.87, DF=2).Only establishments listed between 5 and 250 employees in the CRIQ database were approached, but because size fluctuates some respondents declare more than 250 employees, some fewer than 5. The survey was conducted from January to April 2020: 285 were contacted pre-Covid, 347 after. There are no discernable differences in response profile between pre- and post- Covid respondents, as expected because questions focus upon innovation during the three years preceding the survey and on concrete plans to augment capacity, i.e concrete plans for capacity increase in the twelve months following the survey. In April 2020 Covid was not expected to last much more than a few months, so the latter question provides valid information about expansion plans, even though, given the circumstances, planned expansions may not have been realised.Following Oslo Manual guidelines (OECD 2018), establishments are asked whether they have introduced four types of innovation (i. product, ii. process, iii. organizational, iv. marketing), over the three preceding years (2017–2019) and whether the innovation is new-to-market (radical) or new-to-establishment (incremental). In the analysis we only focus upon new-to-market innovation: an innovator is considered to be any establishment that has introduced at least one new-to-market innovation.Establishments are also asked specific questions about five types of capacity expansion which broadly correspond to stages in the value-chain (i. R&D, ii. production, iii. logistics and distribution, iv. marketing and client relations, v. after-sales service). Finally, respondents are asked whether their establishment used external consultants (i. e., management, engineering, accounting, legal or other consultants) in connection with their capacity increases.For each of these six items (five types of capacity increase as well as use of consultants) respondents are asked to specify where the increase in capacity occurred (or will occur) or, for consultants, where the principal service-provider is located. To cross-validate responses, this question is asked in two ways: the name of the municipality is requested as well as the approximate distance of this municipality from the responding establishment’s location. Using the postal code of town-halls as the point of measurement, distances are calculated using GIS software and checked against estimates made by respondents. These estimates correlate strongly with calculated distances. Respondent estimates serve to disambiguate municipalities (for instance Sainte-Agathe-de-Lotbinière is a small municipality South-West of Québec-City, and Sainte-Agathe-des-Monts is north of Montréal: both can be referred to as Sainte-Agathe). On occasion respondents only provide the estimated distance, not the municipality. Ten respondents declare using out-of-province service-providers, and eighteen declare that they are expanding outside the province: when a specific out-of-province location is mentioned it is a city, typically in the US or Canada, so these responses are classified as metropolitan.Of the 533 establishments which expanded capacity (whether within or beyond their establishment), 266 (42 %) used services in the process. However, only 68 establishments (i. e. 11 %), whether innovators or not, have expandedTo lighten the text the term ‘have expanded’ will be used: this also includes establishments with concrete expansion plans over the 12 months following the survey. capacity beyond the confines of the current establishment (of which 54 report one type of expansion, 9 two, 4 three and 1 five); this is commensurate with low levels of SME scaling-up reported by the OECD (2021). For establishments which report more than one type of expansion, it is the geographically most distant espansion that is used to determine the location of its expansion.3.2MethodTo understand the geography of extra-establishment expansions (and of service-use), observations need to be separated across innovators and non-innovators, across location of the innovator (metropolitan, central, peripheral) and across location of the capacity expansion (local and non-local). This poses no particular problem for service-use: but analysing for extra-establishment expansion, i. e. analysing 68 observations across 12 categories, is problematic because the data are sparse.Our approach has been to analyse contingency tables, distinguishing innovators from non-innovators, crossing the location of each respondents’ expansion (or principal service-provider) with their location. To determine whether patterns revealed in these tables are significant, a chi2 test is used in cases where all assumptions are met (i. e. when we analyze service-use). In cases where these assumptions are not met, Fisher’s exact test is used: this is an iterative method that calculates all possible permutations of the contigency table, and that determines the probability of the observed permutation relative to the null hypothesis of no association between the tables’ two dimensions.The question these tests are designed to answer is the following: is there a statistically significant difference between the location of capacity expansion (or of principal service-provider) for establishments located in metropolitan, central and peripheral regions. Can this difference be observed when the location of capacity expansion (or of principal service-provider) is classified across the metropolitan/non-metropolitan dimension, and can it be observed across the local/non-local dimension? The nature of these differences is identified by examining the contingency tables.Our approach is essentially descriptive: the research strategy is to first identify whether our data are compatible with the idea that innovative establishments in peripheral regions have a higher tendency than non-innovative establishments to use non-local services or to expand beyond the local region, and to then examine whether this tendency differs across peripheral, central and metropolitan regions. We push the data further in the Appendix by isolating external expansions for analysis (A1) and by introducing some control variables into exact logistic regressions (A2), but the data’s sparseness means that conclusions are necessarily tentative. Explanations for the patterns that emerge from our data cannot be empirically explored in this paper, but possible explanations compatible with the literature are discussed in the final section.After presenting results, we return to the issue of data sparseness, which is relevant for understanding why the connection between local innovation and local development is only now being questioned.4ResultsResults are presented in three stages. First, the basic premise of this research, that the propensity of SMEs to innovate does not differ across the metropolitan-peripheral spectrum is verified. Second, the geography of capacity expansion is explored. Third, the geography of service-use related to these capacity increases is also explored.4.1Geography, innovation and capacity expansionThere is no difference in the propensity to introduce new-to-market innovations across the metropolitan-periphery spectrum (Table 1): in keeping with other studies from Québec and elsewhere (as reviewed by Eder 2019), when survey data at the establishment level are used, and when a broader view of innovation than patents is adopted, differences between cities and peripheral areas are not evident. This does not mean that innovation processes are identical across different types of regions but shows that establishments across Québec are similarly intent on updating their products and processes, and on doing so before competitors.Although there is a strong connection between new-to-market innovation and capacity increase (which is to be expected), there is no indication that new-to-market innovators are more likely to expand externally (Table 2).Table 2: Innovation and capacity increase Capacity increaseInnovation No new-to-market innovationNew-to-market innovationChi2No capacity increase 77 20chi2=75.3Capacity increase174361p(chi2=0) <0.0001 No external capacity increase229335chi2=1.73External capacity increase 22 46p(chi2=0) = 0.1894.2The geography of external expansionAlthough respondents are asked about five specific types of external expansion, given the small numbers these are aggregated. Each type of external expansion implies either setting-up a new facility or working with a sub-contractor: hence all have implications in terms of job creation, investment and income generation.A large proportion of establishments declare expansion internal to the establishment (Table 1). We do not have specific information about the nature of such internal expansion: whilst it may involve investment in machinery, the conversion of spaces into R&D labs, investment in delivery trucks, and whilst it may also involve some hiring, we interpret internal expansion as being more modest and as generating fewer jobs and less income than external capacity increase. This interpretation rests on the assumption that internal capacity expansion is simpler, can be very minor, and is the first type of expansion that most expanding firms will consider: it is only if this is not adequate that establishments will turn towards external expansion. External capacity expansion requires considerable coordination, contractual arrangements, organisation, and resources: it will rarely be minor, and typically will be undertaken by firms which envisage sizeable growth in parts of their business.The results are divided into two parts (Table 3): part A examines the location of capacity expansion across the local/non-local dimension. A local expansion is one that occurs within 50km of the establishment: the results are almost identical if 100km is used. Part B examines the location of capacity expansion across the metropolitan/non-metropolitan dimension. For each part of the table, a separate Fisher exact test is presented for non-innovators (Table 3, A1 and B1) and innovators (Table 3, A2 and B2).There is stark contrast between innovators and non-innovators. For non-innovators, the geographic profile of capacity expansion does not differ significantly whatever the location of the expanding establishment. In all types of location about 30 % of establishments are not expanding, about 60 % are expanding internally, and 6 to 10 % are expanding externally. In contrast, for innovators the geographic profile of capacity expansion differs significantly depending on where the establishment is located. Fewer innovators in metropolitan areas don’t expand at all, and innovators in peripheral regions have a slighty higher tendency to expand internally than those in metropolitan areas. Turning to external expansion, not a single establishment in peripheral areas is expanding locally: all external expansion is non-local. In contrast 6 to 8 % of metropolitan and central innovators are expanding locally. It should be noted, however, that as many metropolitan innovators expand non-locally as locally, whereas few central ones expand non-locally. If we focus only upon external expansions (Appendix 1), a similar statistically significant pattern emerges: all external expansion by peripheral innovators is non-local, whereas half is non-local for metropolitan and only a fifth for central innovators; no differences are evident for non-innovators.These first results do not rule out the possibility that some metropolitan and some central establishments are expanding into peripheral locations: this would compensate for the lack of local expansion of innovators in the periphery.Part B of Table 3 examines whether expansion is occurring towards metropolitan or towards non-metropolitan regions. The data are too sparse to subdivide non-metropolitan regions into centre and periphery. Almost all metropolitan innovators that are externally expanding do so towards metropolitan areas, and very few do so towards non-metropolitan locations. Almost all peripheral innovators which are externally expanding also do so towards metropolitan areas. Innovators in central locations are evenly balanced between external expansion towards metropolitan and non-metropolitan areas. Metropolitan areas therefore benefit from expansion of their own innovators as well as from expansion of peripheral innovators (Appendix 1). For innovators in central areas, external expansions are evenly balanced between local and metroplitan areas: this still means that half of their firms’ expansion benefits metro areas.When parts A and B of Table 3 are considered jointly, two important patterns are revealed. First, even though innovative metropolitan establishments do not necessarily externally expand locally, they almost always expand into metropolitan environments. Although we have not measured this, it is feasible that Québec’s metropolitan areas receive expansion from metropolitan areas outside the province, in the same way that they receive expansion from central and peripheral areas. There is no evidence of a drain from metropolitan areas writ large, though specific metro areas may be losing out to others (Council of Canadian Academics 2018).Second, and conversely, innovative peripheral establishments have a slightly higher tendency to expand internally, which we interpret as more modest and less intensive expansion: when they do expand externally, it is invariably non-local and almost always into a metropolitan area. Thus, spill-overs – in terms of activity generated by innovation-related expansion of establishments in the periphery – converge upon metropolitan areas.Though these results are based on a small number of observations, the pattern they reveal is clear and is consistent with the wider literature on rural to urban migration, on static agglomeration economies, and on the mobility of innovators as the innovation process moves from introduction to development and marketing (Shearmur 2015). To the extent that the data allow us to verify, the key results of Table 3 are robust to controls for sector and establishment size (Appendix 2).Table 3: The location of SME capacity increases in QuébecLocal/non-local capacity increaseNoneInternalLocal(within 50 km)Non-local (over 50 km)nFisher’s exact testA1. Non-innovators p=0.697Metropolitan32.3 %58.6 %5.1 % 4.0 % 99Central28.8 %60.0 %2.5 % 8.8 % 80Peripheral30.6 %63.9 %2.8 % 2.8 % 72 A2. Innovators p=0.018Metropolitan 2.8 %80.7 %8.3 % 8.3 %145Central 6.7 %83.0 %5.9 % 4.4 %135Peripheral 6.9 %85.1 %0.0 % 7.9 %101 Metropolitan/non-metropolitan capacity increaseNoneInternalExternal non-metropolitanExternalMetropolitan+nFisher’s exact testB1. Non-innovators p=0.825Metropolitan32.3 %58.6 %2.0 % 7.1 % 99Central28.8 %60.0 %2.5 % 8.8 % 80Peripheral30.6 %63.9 %2.8 % 2.8 % 72 B2. Innovators p=0.044Metropolitan 2.8 %80.7 %2.1 %14.5 %145Central 6.7 %83.6 %4.5 % 5.2 %134Peripheral 6.9 %85.1 %1.0 % 6.9 %101Note: Fisher’s exact test is an iterative method which computes all possible permutations of the table and provides the exact probability that the measured data supports the null-hypothesis of independence of the two dimensions. It is used because the sparse data do not meet the assumptions necessary for a chi2 test. Panel A shows that the geography of expansion differs for innovators, but not for non-innovators: in particular, peripheral innovators only expand non-locally, whereas central and metropolitan innovators expand locally as well as non-locally. This result is confirmed, focusing only on external expansions, in Appendix 1. Panel B shows what type of region establishments expand towards. For peripheral innovators “non-local” expansion is principally “metropolitan”. There is virtually no expansion of metropolitan establishments into non-metropolitan areas.4.3The geography of service-use connected to capacity expansionWhen analysing the location of service-provision (as it relates to capacity expansion) we do not distinguish between internal and external: all service-providers are by definition external to the establishment (Table 4).Table 4: The location of service-use related to capacity increases in QuébecNoneLocal(within 50 km)Non-local(over 50 km)n.Chi-square testLocal/non-local service useA1. Non-innovators Metropolitan71.7 %24.2 % 4.0 % 99chi2=15.6p(chi2=0) = 0.004 Central67.5 %18.8 %13.8 % 80Peripheral75.0 % 6.9 %18.1 % 72 A2. Innovators Metropolitan53.8 %37.2 % 9.0 %145chi2=33.34p(chi2=0) <0.0001 Central48.9 %21.5 %29.6 %135Peripheral37.6 %23.8 %38.5 %101 Metropolitan/non-metropolitan service useNoneNon-metropolitanMetropolitan+ n.Chi-square testB1. Non-innovators Metropolitan72.4 % 0.0 %27.6 % 98chi2=32.4p(chi2=0) <0.0001 Central67.5 %22.5 %10.0 % 80Peripheral76.1 % 7.0 %16.9 % 71 B2. Innovators Metropolitan54.2 % 4.2 %41.7 %144chi2=40.6p(chi2=0) <0.0001Central48.9 %27.4 %23.7 %135Peripheral39.0 %33.0 %28.0 %100Note: Panel A shows how the geography of expansion-related services differs between innovators and non-innovators, and across metropolitan, central and peripheral regions: peripheral establishments tend to use non-local services, but peripheral innovators use more services than non-innovators, and also use more non-local services. Panel B shows from what type of regions establishments source their services: peripheral innovators are evenly split between using metropolitan and non-metropolitan services. Metropolitan innovators use almost no non-metropolitan services. Many establishments that are expanding – whether internally or externally – use external service-providers (Table 4). Whilst both innovators and non-innovators use external services for capacity expansion, only 30 % of non-innovators do so compared to 50 % of innovators. The propensity to use external services amongst non-innovators is similar wherever establishments are located (Table 4, part A). In contrast, the propensity for peripheral innovators to use external services is higher than for central or metropolitan innovators: this may partly reflect the greater challenges faced by peripheral establishments as they envisage non-local expansion, presumably over long distances.In all locations the tendency to use non-local services is higher for innovators. Metropolitan establishments have a far higher propensity to use local services than either central or peripheral establishments. Leakages associated with service-provision (in terms of fees, of service employment, and of strengthening the service sector …) are higher for peripheral than for central or metropolitan areas: this is driven by expanding innovators who not only use more services but use more non-local services.Looking at these numbers a different way, the dominance of metropolitan areas in terms of service-provision becomes evident (Table 4, Part B): almost half of metropolitan innovators turn to metropolitan service-providers for advice on expansion, and about a quarter of central and peripheral ones do. Conversely, only 4.2 % of metropolitan innovators turn to service-providers outside metropolitan areas, and about a third of central and peripheral innovators do so. Only 7 % of peripheral non-innovators turn to non-metropolitan service-providers, whereas 22.5 % of central non-innovators do. A higher proportion of innovators in metropolitan areas consult no service-providers for expansion purposes than in the periphery (54 % v. 39 %), maybe because local expansion – more prevalent amongst metropolitan innovators – can be more readily envisaged without external advice.Metropolitan regions benefit from service-demand generated by central and peripheral establishments that are expanding, but there is no reciprocal benefit for non-metropolitan regions: they do not benefit from service-demand from expanding metropolitan establishments. This is not a surprise: business-service providers tend to locate in metropolitan areas. Yet, considering that many policies encourage local innovation activities in view of generating local economic development, this result – unsurprising as it is – should give pause. It reveals that consulting and business services – major industries in metropolitan areas that are high-wage and fast-growing – are partly fueled by the expansion of establishments, innovative ones in particular, located in non-metropolitan areas. Resources and knowledge generated by expanding establishments in the periphery are transferred towards metropolitan areas.5DiscussionThe purpose of this paper is not to argue that value always flows from the periphery to the centre. Its focus is more specific: to the extent that many local development policies aim to stimulate firm-level innovation, the question addressed is whether, when innovators scale-up, they do so locally (thereby benefitting the local economy) or non-locally (thereby benefitting other localities and regions).This line of questioning is motivated by some contradictions inherent in the ‘innovation’ approach to local development (Marques and Morgan 2021; Iammarino et al. 2017), and by recent policy focus on the scaling-up of innovative SMEs (OECD 2021; ISED 2019). Our questions empirically probe Shearmur’s (2015) suggestion that in sparse and peripheral regions establishments innovate but turn towards metropolitan areas when they scale-up, expand capacity, and market their innovations. They also explore Glückler et al.’s (2022) more general suggestion that innovative agents can strategically move between more central and more peripheral positions (and vice-versa) to leverage the advantages of both at different points along the innovation process.Our data reveal that when innovators expand their capacity externally the geography of this expansion differs across the metropolitan to periphery spectrum for innovators, but not for non-innovators. Metropolitan innovators either expand locally or expand into other metropolitan areas: they are embedded in metropolitan milieux. Peripheral innovators expand almost exclusively non-locally, into metropolitan areas: they are not embedded in peripheral milieux. The results for service-use (as it relates to capacity expansion) are similar: although peripheral establishments use some locally-sourced services, they are more likely to use non-local services, with a high propensity to turn towards metropolitan-based service-providers. This result, which holds for innovators and non-innovators alike, is stronger for innovators and concerns a higher proportion of them.Results for establishments in central locations – for both expansion and service-use – lie between those for metropolitan establishments and those for peripheral establishments.To the extent that local expansion and local expansion-related service-use contribute to endogenous development processes, our results confirm that endogenous processes occur within metropolitan areas, which is as expected since these regions generate agglomeration economies and tend to be institutionally thick. Metropolitan areas benefit from the dynamism of their own innovative establishments and from the dynamism of each others’ innovative establishments: our results are compatible with the local buzz of large cities, and with connexions (pipelines?) between these cities (Bathelt et al. 2004).In central regions there is evidence that innovators expand capacity into both non-metropolitan and metropolitan areas, with non-metropolitan expansion tending to be local. There is also a slightly lower tendency for central than for peripheral establishments to use expansion-related services based in metropolitan areas. Thus, central regions occupy an intermediate position, with some evidence of local expansion by innovators, as well as some evidence of establishments turning towards metropolitan areas.In peripheral regions, the evidence is overwhelming that external capacity expansion by innovators (to the extent it occurs), as well as service-use in aid of this expansion, are non-local. It is thus difficult to argue, at least based on these two indicators, that local innovators in peripheral areas set up virtuous local multiplier and spillover dynamics once they have innovated. Conversely, peripheral regions do not seem able to generate dynamics that would enable innovators to scale-up locally (Morgan 2019).Whilst our results pertaining to service-use are not surprising – they confirm that this holds true for service-use related to capacity expansion – those relating to the location of external capacity expansion are novel. However, the external expansion results, although statisically significant (Table 3, A2; Appendix A2), rest upon sparse cross-sectional data. The data are sparse because, during the four-year period about which respondents are questioned, most capacity expansion happens within the confines of establishments. This type of minor expansion is common across establishments in all locations, and is more common for innovators.When it comes to major expansion, involving extra capacity beyond the establishment, numbers are small – in keeping with recent OECD (2021) figures. Over a four-year period only about 11 % of establishments engage in it, 14 % in cities and 7 % in peripheral regions (Table 1). This is commensurate with establishment owners in peripheral areas being constrained by lack of local resources, or choosing – for lifestyle reasons – to limit growth (Dominici et al. 2019). It is also commensurate with process innovation – a type of innovation more common in peripheral areas – that expands production capacity in view of reducing costs and increasing productivity: this type of innovation, which is often labour saving, can lead to local economic decline if localities do not have the size and diversity to generate alternative uses for freed-up resources (Shearmur 2015). Creative destruction, occurring at an aggregate scale across the economy, may destroy in the periphery whilst creating towards the centre.The small number of external expansions by peripheral establishments coupled with their geographic dispersal makes it difficult to systematically determine where such expansion occurs. Frenken and Boschma (2007), in their discussion of branching processes in Europe, observe that most spin-offs are local: however, unless spin-offs can be subdivided between those originating from metropolitan establishments and those originating from peripheral ones, then the small number of observations drawn from peripheral areas in national databases will be overwhelmed by metropolitan observations, which, our data confirm, tend to expand locally. It is for this reason that establishments in peripheral regions are oversampled in our study: this allows us – albeit tentatively – to identify differences in the geography of capacity expansion according to the location of innovating establishments.It is also important to bear in mind that Québec’s periphery differs from peripheries in continental Europe: it is similar – from a geographic perspective – to that of Nordic countries and places like Australia. Living standards are relatively high, but peripheral areas are characterised by small towns, low density, and physical isolation from metropolitan centres. Much of non-metropolitan continental Europe resembles what we have termed ‘central’ regions, at least with respect to access to metropolitan areas. Conversely, given Europe’s history, institutional and cultural distances between physically proximate regions may be higher than those between physically distant regions in QuébecThis does not hold true for institutional and cultural distances between areas dominated by Indigenous peoples and those dominated by more recent settlers. This proviso does not impact our analysis, which essentially covers the mainstream settler economy.. In this paper our understanding of centre and periphery is distance-based: other types of distance may be more relevant in different contexts (Boschma, 2005).Finally, our results only cover a three-year window. They reveal that innovative peripheral establishments that expand beyond their confines do so into metropolitan areas using non-local (often metropolitan) services. Regional development, however, is a long-term process. Peripheral regions – except a few that benefit from specific resources or amenities – have been declining for at least 40 years in Canada (Polèse and Shearmur 2006). A weak or ambivalent process, if it endures over time, can lead to large disequilibria, in the manner of compound interest.However, small but persistent long-term processes are not readily apparent, and data are not gathered to measure them: thus, many local development policies have overlooked the mechanisms we highlight in this paper. Furthermore, even if such processes are suspected, they are overwhelmed by the general consensus in favour of innovation-led local development and endogenous growth. This is the most compelling reason to carefully consider the results presented here: the paper presents vital clues to a potentially important process, and corroborates recent theoretical discussions (Margarian 2011; Iammarino et al. 2017; Shearmur 2015; Morgan 2019; Glückler et al. 2022) and high-level data analysis (Marques and Morgan 2021; Shearmur and Bonnet 2011).In a legal setting, our evidence would be qualified as circumstantial rather than direct. Without (yet) providing enough evidence to overturn the consensus, our results should at least draw researchers’ and policymakers’ attention to some fundamental questions about the connection (or lack of it, in peripheral regions) between local firm-level innovation and local economic development.These points mirror and extend the conclusions of the OECD’s (2021) pilot project on the scaling-up of SMEs: the OECD argues for better use of existing micro-data to explore scaling-up processes. We suggest that existing micro-data derived from innovation surveys, in-so-far as it is used to inform local development and innovation policy, is not sufficient: it should be complemented by data that probes the geography of scaling-up over time.6Conclusion: the conundrum of policymaking without adequate dataOur results are clear and consistent with theory. However, the data we rely on is sparse and cross-sectional. We suggest a number of reasons why these results should nevertheless be carefully considered: in particular, if the tendency for peripheral establishments to expand into metropolitan areas is indicative of a long-term process, then the data’s sparseness merely reflects the lack of time-series data and/or larger scale studies.From both research and policy perspectives this presents a conundrum: it is difficult to obtain funding for data collection that explores a process that is not recognised, but the process will only be recognised if good data are available. Such data can only be gathered by a large statistical agency, or by a well-funded research team. It is not collected in current innovation surveys, and gathering such data over the long-term – say over a ten-year period – may be prohibitively expensive.Without data to more conclusively establish whether innovative firms in the periphery tend to expand towards metropolitan areas, policymakers and analysts are left with two options. Either they accept the convention that local development is driven by local firm-level innovation – a convention backed by metropolitan-biased empirical evidence and which appears to be consistent with endogenous growth theory – or they consider some of the fundamental questions that have been raised about endogenous regional processes (Margarian 2011; Shearmur 2015; Iammarino et al. 2017; Marques and Morgan 2021; Glückler et al. 2022), as well as pay attention to suggestive evidence as presented in this paper.Knowing more about these fundamental questions would help policy-makers better evaluate the regional processes at play and how they affect local development in peripheral regions. Indeed, if there is a link between firm-level innovation and expansion towards metropolitan regions, as our results suggest, then many policies – such as recent ones that focus upon the scaling-up of innovation (OECD 2021; ISED 2019) – should be spatially modulated to address the movement of expanding innovators away from peripheral and laggard regions.
Journal
Zeitschrift für Wirtschaftsgeographie – de Gruyter
Published: Nov 1, 2022