Fraunhofer-Institut für System- und Innovationsforschung ISI

Filters

2
2
2
Reset filters

Settings
Now showing 1 - 2 of 2
  • Publication
    Final report on the CDM model including updated literature, estimations from the econometric model, and a discussion of the policy implications
    ( 2022)
    Schubert, Torben orcid-logo
    ;
    Türkeli, Serdar
    ;
    Ashouri, Sajad
    ;
    Bäck, Asta
    ;
    Deschryvere, Matthias
    ;
    Jäger, Angela
    ;
    Visentin, Fabiana
    ;
    Hajikhani, Arash
    ;
    Suominen, Arho
    This final report for Work Package 2 (Deliverable D4) implements the econometric estimations concerning the relationship between the specifics of the innovation process and company productivity as outlined in the interim report (D3). This final report consists of two main parts. In the first, we present the overall project work related to the development of a multistage R&D-innovation-productivity (CDM) model inspired by Crépon et al. (1998). We start with an updated review of the literature relating to the secular stagnation hypothesis, that is, the claim that the depletion of technological opportunities has heralded a phase of low or no productivity growth. Where more recent works have become available, this literature review has been updated. In order to align the literature review with the empirical evidence, we also focus more on the role of digitalization, where web-scraping based data collection has provided very useful indicators in the area of intangibles. Beyond digitalization trends, cooperation, and open innovation, knowledge spillovers and servitization are increasingly important drivers of productivity. We then describe and implement the augmented Crépon-Duguet-Mairesse (CDM)-type model, supplementing the original specification with the influence factors of digitalization, cooperation, spillovers, and servitization. The CDM model is further amended from the original outline in D3 in order to achieve alignment with the existing data. This part also contains a description of the processes underlying data generation and management, including a discussion of data quality issues and missing data, and discusses the potential and limitations of scraping data from company websites for empirical innovation economics. Concluding the first part of this report, we discuss the policy implications resulting from our analysis. In the second part of this report, we present a more detailed analysis of the role of AI use on productivity using a panel on Finnish firms for the period 2013-2019. This exercise was not originally planned as part of the work package. However, the availability of a specialized data source enabling us to infer company use of AI based on data on job offerings provided the opportunity for complementary analysis. Because this data became available as panel data, it also allowed us to control for a number of econometric issues that the web-scraped data from part 1 was unable to resolve.
  • Publication
    Productivity effects of process vs. product digitalization
    ( 2022)
    Schubert, Torben orcid-logo
    ;
    Ashouri, Sajad
    ;
    Deschryvere, Matthias
    ;
    Jäger, Angela
    ;
    Visentin, Fabiana
    ;
    Pukelis, Lukas
    ;
    Hajikhani, Arash
    ;
    Suominen, Arho
    Digitalization is considered an important driver of upcoming societal and economic transformations. However, holding both promises and challenges, its effects on the performance of individual firms are still underexplored. In this paper, we disentangle the phenomenon into two distinct factors: the digitalization of processes and the digitalization of product offering. We analyse the effects of the two digitalization factors on firm-level productivity. This analysis is based on a large European-wide unique dataset combining structured information from ORBIS and PATSTAT with web-scraped information on the firms involved in high-tech manufacturing. Building on a triangular structural equation model -- including a patenting equation and a productivity equation -- we find that digitalization boosts productivity both directly and indirectly. The direct effects occur through immediate effects on productivity, while the indirect effects occur through increased patenting. However, the positive effects occur largely for product digitalization, while process digitalization on average does not significantly contribute to productivity. Interestingly quantile regression estimates show that the effects of product and process digitalization show significantly contrasting patterns across the productivity distribution. While the effects of product digitalization are largest for highly productive firms, there are mildly positive effects of product digitalization for lowproductivity firms