Recombinant innovation and the boundaries of the firm

Sokbae Lee, Rachel Griffith, Bas Straathof2014 · DOI

TL;DR

This paper quantifies the friction in knowledge diffusion between firms, finding that it's about 30% more costly to build on an idea from another company than from within your own. This "firm boundary effect" is significantly larger than barriers from national borders or technological distance, highlighting substantial imperfections in the market for technology.

What Problem Does This Paper Address?

The innovation landscape increasingly relies on combining existing ideas (recombinant innovation). While economic theory suggests that market frictions like incomplete contracts and transaction costs hinder the flow of ideas between firms, empirical evidence on the magnitude of these frictions has been limited. This paper aims to quantify these frictions, specifically focusing on how the boundaries of a firm affect the speed and cost of accessing and utilizing external knowledge compared to internal knowledge. Understanding this is crucial for designing effective innovation policies and understanding firm strategy.

Methods

The authors employ a multiple-spell duration model, a statistical technique used to analyze the time it takes for an event to occur, especially when multiple such events can happen for the same entity. In this context, the "event" is a patent citation, and the "entity" is a cited patent. They model the hazard rate – the instantaneous probability of a patent being cited – as a function of various factors, including whether the citation is within the same firm or between different firms.

A key methodological innovation is the use of a conditional likelihood approach inspired by Griffith et al. (2011). This allows them to estimate the model's coefficients without needing to explicitly model the baseline hazard function or deal with unobserved patent quality (fixed effects). This is achieved by comparing adjacent citations for the same patent, effectively "first-differencing" away unobserved characteristics. They also address censoring (patents not yet cited) by weighting observations.

Datasets

The study utilizes a comprehensive dataset of European patent applications from the European Patent Office's (EPO's) Worldwide Statistical Patent Database (PATSTAT), covering the period from 1985 to 2004. To identify firm ownership structures, they integrate this with commercial data from Bureau Van Dijk (BVD), specifically the Amadeus database for European firms and Icarus for US firms with European subsidiaries. This integration allows them to distinguish between within-firm and cross-firm citations.

Key Insights and Ideas

  • Firm boundaries are significant barriers: The most striking finding is that the cost of accessing and utilizing an idea from another firm is substantially higher than doing so internally. This suggests that firms are not just passive recipients of knowledge but actively manage and control its flow.
  • Recombination is costly: The paper empirically confirms the theoretical notion that recombining existing knowledge into new innovations is not frictionless. The "market for technology" is far from perfect.
  • Firm boundaries vs. other barriers: The magnitude of the firm boundary effect is significantly larger than barriers imposed by national borders or technological distance, suggesting that internal organizational structures and knowledge management play a more critical role in innovation diffusion than often assumed.
  • Quality heterogeneity matters: The authors highlight that differences in patent quality can bias estimates of citation delays. Their fixed-effects approach helps to control for this, revealing that the cross-firm effect is even stronger when unobserved quality is accounted for.

Key Results

  • 30% higher cost for cross-firm innovation: On average, it is approximately 30% more costly to successfully discover and utilize new ideas created in another firm compared to ideas developed within one's own firm.
  • Smaller barriers for national and technological distance: The increased cost of accessing ideas across national borders is around 5%, and across different (4-digit) technology classes is around 20%.
  • Significant impact on citation speed: Cross-firm citations take, on average, 28% longer for the first citation and 47% longer for the second citation compared to within-firm citations.
  • Industry variations: While the effect is present across industries, it is slowest in chemical engineering (37% delay) and fastest in pharmaceuticals (27% delay).

Limitations and Future Work

  • Data limitations: The study focuses on European patents. While they use data on US firms with European subsidiaries, the primary dataset is European. Future work could explore similar analyses using US patent data.
  • Censoring correction: While they implement a weighting scheme to address censoring (patents not yet cited), the assumption of independence between censoring and covariates is crucial.
  • Unobserved factors: Although they control for unobserved patent quality, other unobserved factors might still influence citation patterns.
  • Future directions: The authors suggest that their findings point towards substantial imperfections in the market for technology, which could inform policy and firm strategy. Further research could explore the specific mechanisms behind these frictions and how they evolve over time.

Summary updated

Abstract

There is considerable interest in understanding how important market frictions are in stifling the transmission of ideas from one firm to another. Although the theoretical literature emphasizes the importance of these frictions, direct empirical evidence on them is limited. We use comprehensive patent data from the European Patent Office and a multiple spells duration model to provide estimates that suggest that they are substantial. It is around 30% more costly to successfully discover and utilize new ideas created in another firm than in your own. This compares to the increased costs of accessing new ideas across national borders of around 5%, and across technologies of around 20%. These result point towards substantial imperfections in the market for technology.
BibTeX
@misc{lee2014recombinant,
  author = {Sokbae Lee and Rachel Griffith and Bas Straathof},
  title = {{Recombinant innovation and the boundaries of the firm}},
  year = {2014},
  doi = {10.1920/wp.cem.2014.4014},
}
Figure