Global Value Chains Participation and Knowledge Spillovers in Developed and Developing Countries: An Empirical Investigation

Abstract

In this paper we investigate the relationship between participation in global value chains (GVCs) and countries’ innovation performance. We use the recently released world input–output database to build spillover variables by weighting the aggregate R&D stock of partners supplying inputs with two alternative measures of participation in GVCs: the share of foreign value added in a country’s gross exports and the offshoring index. We use these indicators to test empirically the relationship between a country’s innovation performance, proxied by patent per capita, and spillovers generated by GVCs. Our results show that the involvement in GVCs, in particular for developing countries importing inputs from advanced economies, is positively related with a country’s innovation outcome suggesting that international fragmentation of production can indeed be a channel allowing for international technology transfer from developed to developing countries.

Nous étudions la relation entre la participation aux chaînes de valeur mondiales (CVM) et la performance de l'innovation au niveau des pays. Nous utilisons la base de données mondiale des entrées-sorties (WIOD) récemment lancée pour établir les variables de retombées en pondérant le stock total de RD des partenaires fournissant des biens avec deux autres mesures de participation aux CVM: la part de la valeur ajoutée étrangère dans les exportations brutes et l’indice de délocalisation. Nous utilisons ces indicateurs pour tester empiriquement la relation entre le degré d'innovation d'un pays, représenté par le nombre de brevet par habitant, et les retombées générées par les chaînes de valeur mondiales. Nos résultats montrent que l'implication dans les CVM est positivement corrélée au degré d’innovation d'un pays, en particulier pour les pays en développement importateurs des biens en provenance d'économies avancées, suggérant ainsi que la fragmentation internationale de la production peut être un vecteur permettant le transfert international de technologie des pays développés vers les pays en développement.

Appendix

Further Robustness Checks

Foreign value added in export and offshoring could also come from natural resources industries, which we expect to embody negligible amount of foreign knowledge and therefore to generate lower knowledge spillovers with respect to other industries. Therefore, we have run the regressions of the baseline model (Table  2) splitting our spillover variables, both the OFFIND and the FVSIND, into two variables for each indicator: one with the weights built from natural resource industries only and a second one with the weights built on the rest of the economy. Results show that the coefficient of the spillover variables excluding the natural resource industries is very close to the original ones, while the spillover from natural resources industries is never significant. Since the period we consider in the analysis includes the 2008 crisis, which affected trade volumes and GVC participation (Escaith et al, 2010), we also run the regressions of our preferred specifications (Columns 4 and 8, Table  2) by interacting the GDP per capita and the two spillover variables with a post-crisis dummy. Our results are confirmed. The results for both the checks are available upon request.

Supplementary Tables

See Tables 6, 7, 8, 9, 10 and 11.

Table 6 Models of international knowledge spillovers through GVC participation

Table 7 Models of international knowledge spillovers through GVC participation

Table 8 Gravity models

Table 9 Sample descriptive statistics

Table 10 Cross-correlation table

Table 11 Variables and sources