Knowledge, learning, and innovation are vital elements in facilitating economic development and growth. Technological change, which is a synonym for generating knowledge, the diffusion thereof, and subsequent application in the marketplace in the form of novel products and processes, i.e. innovations, has a strong effect on the collective wealth of regions and nations. Knowledge spillovers, which are unintended knowledge flows that take place among spatial (geography) and sectoral (industry) units of observation, provide a rationale for diverging growth rates among spatial units, well beyond what might be explained by variations in jurisdictional factor endowments, and thus are of particular interest in this context. Measuring and quantifying the creation and diffusion of knowledge has proven to be a challenging endeavor. One way to capture technical and economically valuable knowledge is by means of patent and patent citation analysis. Following this approach, and utilizing a novel patent database that has been specifically developed for this purpose, the present dissertation investigates the spatio-sectoral patterns of knowledge spillovers in the Canadian economy over the time period 1983 to 2007. The employed research methodology addresses existing limitations in this stream of research, and contributes to the continuing debate regarding the significance of sectoral specialization versus diversity, and local versus non-local knowledge spillovers as the main driver of knowledge formation processes leading to innovation at the sub-regional scale. The findings indicate that knowledge spillovers are localized, and furthermore, that this localization effect has increased over time for both spillovers within a particular industry, as well as between industry sectors. The analysis of micro-geographic industry specific spatio-sectoral knowledge formation processes, and the inquiry into local sectoral knowledge spillover patterns, outlines how regional evolutionary technology trajectories potentially shape the rate and direction of technological change, and consequently influence economic growth, at a particular place.

Knowledge, learning, and innovation are vital elements in facilitating economic development and growth. Technological change, which is a synonym for generating knowledge, the diffusion thereof, and subsequent application in the marketplace in the form of novel products and processes, i.e. innovations, has a strong effect on the collective wealth of regions and nations. Knowledge spillovers, which are unintended knowledge flows that take place among spatial (geography) and sectoral (industry) units of observation, provide a rationale for diverging growth rates among spatial units, well beyond what might be explained by variations in jurisdictional factor endowments, and thus are of particular interest in this context. Measuring and quantifying the creation and diffusion of knowledge has proven to be a challenging endeavor. One way to capture technical and economically valuable knowledge is by means of patent and patent citation analysis. Following this approach, and utilizing a novel patent database that has been specifically developed for this purpose, the present dissertation investigates the spatio-sectoral patterns of knowledge spillovers in the Canadian economy over the time period 1983 to 2007. The employed research methodology addresses existing limitations in this stream of research, and contributes to the continuing debate regarding the significance of sectoral specialization versus diversity, and local versus non-local knowledge spillovers as the main driver of knowledge formation processes leading to innovation at the sub-regional scale. The findings indicate that knowledge spillovers are localized, and furthermore, that this localization effect has increased over time for both spillovers within a particular industry, as well as between industry sectors. The analysis of micro-geographic industry specific spatio-sectoral knowledge formation processes, and the inquiry into local sectoral knowledge spillover patterns, outlines how regional evolutionary technology trajectories potentially shape the rate and direction of technological change, and consequently influence economic growth, at a particular place.

En décembre 2004, les autorités publiques françaises lancent le premier appel à projet donnant le coup d'envoi de la politique des pôles de compétitivité dont la phase 3, lancé par le gouvernement le 6 janvier 2013, met particulièrement l'accent sur les retombées économiques issues des pôles de compétitivité qui devront être amplifiées. L'objectif de cette thèse est de vérifier l'existence de telles retombées (spillovers) issues des investissements directs étrangers (IDE) dans le cas français. Plus précisément nous analysons les conséquences de spillovers sur la productivité des firmes en France et nous cherchons à savoir si les décisions d'investir en Recherche et Développement (R&D) sont étroitement liées au niveau des spillovers circulant entre les firmes locales et les firmes étrangères implantées en France. Dans un premier chapitre, nous vérifions l'existence de spillovers de connaissance à travers deux canaux de transmission : horizontal et vertical. Nous mettons en évidence les spillovers qui se diffusent d'une part entre clients étrangers et fournisseurs locaux dans les secteurs en amont (backward linkages) et d'autre part, entre fournisseurs étrangers et clients locaux dans les secteurs en aval (forward linkages). Nous suggérons que dans le cas de la France, les spillovers de connaissance se manifestent principalement via les backward spillovers, alors que les spillovers au sein d'un même secteur (horizontal spillovers) et les spillovers diffusés par les fournisseurs étrangers vers les clients locaux agissent plutôt comme un frein à la productivité des firmes en France. Dans le second chapitre, nous portons notre attention à l'intérêt des firmes d'investir davantage en R&D en fonction de l'intensité des retombées issus de ces activités. Nous construisons, pour ce faire, un modèle théorique permettant de modéliser les stratégies des firmes en termes de R&D en fonction du savoir-faire diffusé par une filiale étrangère dans le pays d'accueil (international R&D spillovers), du savoir-faire diffusé par une firme locale (reverse R&D spillovers) et de l'échange du savoir-faire entre une filiale et sa maison-mère située dans son pays d'origine (internal technological transfer). Nous montrons que la prise en compte de ces différents canaux de spillovers exogènes dans le modèle renvoie à un cadre du dilemme du prisonnier où la diffusion d'un niveau élevé et comparable du savoir-faire encourage les firmes à investir en R&D et qu'une diffusion faible par rapport à un seuil (établi en fonction de l'intensité de la concurrence) les incite à réduire leurs activités de R&D. Dans un cadre de extit{spillovers} asymétrique, la firme qui absorbe le plus la technologie de sa concurrente a intérêt à investir davantage en R&D. Dans le dernier chapitre, nous nous intéressons aux international spillovers et reverse spillovers issus non seulement des activités de R&D mais aussi des activités d'outsourcing. Nous suggérons que les effets des spillovers internationaux sur la productivité des firmes en France sont plus importants que ceux des reverse spillovers. Lorsqu'il s'agit de spillovers issus des activités de R&D, l'effet est positif et significatif en amont et en aval mais plus important pour les liens en backward. Concernant les spillovers issus des activités d'outsourcing, l'effet est uniquement en faveur des spillovers diffusés par les donneurs d'ordre. Ces effets sont mitigés selon le niveau technologique du secteur concerné. L'effet des spillovers sur la productivité des firmes appartenant à un secteur de niveau technologique élevé semble être plus important que pour celles appartenant à des secteurs de bas et/ou moyen niveau technologique. / In December 2004, the French public authorities launched the first call for projects kicking off the policy clusters, including Phase 3 launched by the government on 9 January 2013, with particular emphasis on issues of economic spinoffs clusters that should be amplified. The objective of this thesis is to verify the existence of such spinoffs (spillovers) from foreign direct investment (FDI) in the French case. Specifically, we analyze the impact of spillovers on the productivity of firms located in France and we want to know if the decisions to invest in R&D are closely related to the level of spillovers flowing between local firms and foreign firms operating in France. In the first chapter, we aim at verifying the existence of knowledge spilovers through two transmission channels: horizontal and vertical ones. We highlight that the know-how spreads on the one hand, between foreign customers and local supplier in the upstream sector (backward linkages) and on the other hand, between local suppliers and foreign customers in the downstream sector (forward linkages). We suggest that, in the French case, the knowledge spillovers occur primarily through backward spillovers, while spillovers within the same sector (horizontal spillovers) and spillovers diffused from the foreign suppliers to the local customers rather act as a brake on the productivity of firms located in France. In the second chapter, we focus on the incentive of firms to further invest in R&D with respect to the intensity of spillovers resulting from these activities. For this purpose, we build a theoretical model to explain the strategies of firms in terms of R&D based on the technological know-how disseminated by a foreign subsidiary in the host country (international R&D spillovers), know-how issued by a local firm (reverse R&D spillovers) and the know-how that being exchanged between the subsidiary firm and its parent company located in its home country (internal technological transfer). We show that taking into account these different channels of exogenous spillovers refers to the prisoner's dilemma where the diffusion of a high and comparable level of know-how encourages firms to invest in R&D and a low level of spillovers compared to a threshold value (determined based on the intensity of competition) forces them to reduce their R&D investment. In a context of asymmetric spillovers, the firm which more absorbs the technology of its competitor increases its investment in R&D. In the last chapter, we focus on international spillovers and reverse spillovers not only from R&D activities but also from outsourcing activities. We suggest that the effects of international spillovers on the productivity of firms in France are more important than those of reverse spillovers. As far as R&D spillovers are concerned, their effect on a firms' productivity is positive and significant in the upstream and downstream sector, but is more important for backward linkages. Concerning spillovers from outsourcing activities, the effect is only in favor of spillovers diffused by the contractor. These effects depend on the technological level of the sector concerned. The effect of spillovers on the productivity seems to be more important for firms belonging to high-tech sectors than for firms belonging to medium- and low-tech sectors.

Partant d’un manque de consensus académique quant à l’efficacité des différents dispositifs de soutien à l’innovation, cette thèse s’intéresse aux politiques de financement de la R&D dans un contexte où les entreprises peuvent coordonner leurs effortsde R&D et se rapprocher d’un secteur public de la recherche. Dans une première partie, nous présentons des éléments de la littérature économique sur les mesures incitatives de promotion de la R&D. Nous choisissons par ailleurs de contextualiser notre approche autour de la politique des pôles de compétitivité menée en France. Dans une seconde partie, nous proposons une modélisation théorique originale, à même d’étudier conjointement trois instruments publics de soutien à la R&D : encouragement aux stratégies coopératives de R&D, soutien financier à la R&D privée et soutien financier au secteur public de la recherche. Nos résultats théoriques nous permettent ainsi de proposer plusieurs recommandations de politique publique. Premièrement, nous mettons en avant le rejet de tout effet d’éviction de l’intervention publique sur les activités privées de recherche. Nous montrons également qu’une politique de financement d’un secteur public de la recherche peut représenter unealternative efficace à la politique visant à soutenir la R&D privée, sous condition que la recherche publique génère d’importants effets de débordement. Enfin, la répartition optimale du budget public destiné au soutien à l’innovation s’avère êtreplus généreuse envers le secteur public à mesure que les entreprises privées se situent à proximité de l’acteur public de la recherche. A l’inverse, le soutien public à l’innovation doit se faire davantage en faveur des entreprises privées à mesure que ces dernières sont soumises à une concurrence forte et à des difficultés d’appropriation de leur recherche. / Observing a lack of academic consensus about R&D public supports efficaciency, we focus on public R&D funding policies by considering R&D cooperation and proximity between firms and a public research sector. In a first way, we introduce how economic literature studies incentives to promote R&D efforts. In addition, we decide to illustrate our theorotical approach through french cluster policy called « politique des pôles de compétitivité ». In a second way, we develop an original theorotical modelisation able to evaluate the efficacity of three public instruments that promote R&D efforts and innovation : promote R&D cooperation, subsidizing private R&D sector and funding to public research sector. Our theorotical results lead us to make some recommandations to the policy makers. First, we conclude to an additionality effect of public policies on R&D efforts, that means we reject all crowding-out effect of public intervention. Moreover, we show that funding a public research sector (SPU policy) leads to better performance than the policy consisting to subsidize private R&D efforts (SPR policy) but only if the level of public spillovers is strong enough. Then, about the distribution of public fundings between public and private sector (SPM policy), we conclude that the proportion allocated to private sector always increases with the level of inter-firm spillovers and with the concentration of the industry if and only if the level of inter-firm spillovers is high enough. In the opposite, this proportion allocated to private sector decreases with the level of public knowledge externalities to the private sector, due to a closer proximity between public and private bodies, regardless of whether firms cooperate or not in R&D.