Performance Effects of External Search Strategies in European Small and Medium‐Sized Enterprises¹

1 External search, search strategies, external knowledge flows, and external knowledge sources are used interchangeably in the open innovation literature.

Abstract

There is little evidence regarding the performance impact of open innovation on small and medium‐sized enterprises (SMEs), especially across different firm‐size categories and sectors. Using new survey data from 28 European countries, we specify ordered logit and generalized proportional odds models to explore how seven individual external search strategies (knowledge sources) affect SME innovation performance across different size categories and sectors. While we find some consistently positive effects, in particular from using customers as an external knowledge source, we also find that some search strategies may not be beneficial. These findings suggest managerial and policy implications.

1 External search, search strategies, external knowledge flows, and external knowledge sources are used interchangeably in the open innovation literature.

1 External search, search strategies, external knowledge flows, and external knowledge sources are used interchangeably in the open innovation literature.

Notes

1 External search, search strategies, external knowledge flows, and external knowledge sources are used interchangeably in the open innovation literature.

1 External search, search strategies, external knowledge flows, and external knowledge sources are used interchangeably in the open innovation literature.

2 We thank an anonymous referee for this point.

3 The dataset used in the analysis contains information about the share of innovation expenditure in total expenditure, but innovation expenditure encompasses both internal and external (extramural) R&D expenditures. Given that outsourcing R&D is considered as a type of open innovation, the share of innovation expenditure is omitted from the model. Furthermore, this variable is highly correlated with the variable measuring the share of R&D personnel (the correlation coefficient is 0.79), suggesting a potential problem with multicollinearity if both variables were to enter the model (Greene 2005).

4 If the proportional odds assumption is not rejected then either ordered logit or ordered probit are appropriate. However, these yield qualitatively similar results. Ordered logit is chosen to be consistent with the partial proportional odds model, our preferred estimator when the proportional odds assumption is rejected.

5 The “standard interpretation of the ordered logit coefficient” (UCLA, 2015b) is that “for a one unit increase in the predictor, the response variable level is expected to change by its respective regression coefficient in the ordered log‐odds scale while the other variables in the model are held constant.”

6 Using the whole sample, we also tested whether participation in innovation networks moderates the impact of other OI practices on innovation performance, as recently suggested by Mazzola, Perrone, and Kamuriwo (2015). We would like to thank an anonymous referee for this suggestion. This hypothesis is tested by augmenting our model with six interaction variables calculated by multiplying each of our six variables representing other sources of external knowledge with the variable representing participation in innovation networks (Source_networks) as a knowledge source. Due to multicollinearity, each of these interactions proved to be statistically insignificant at any conventional level. Accordingly, we entered each interaction term individually. The coefficients on two of these interaction terms (informal network with research organization—Source_research—moderated by participation in innovation networks; and strategic alliances—Source_strategic—moderated by participation in innovation networks) proved to be negative and statistically significant at the 5 percent level in the highest category of innovative sales. These two significant interactions suggest that Source_networks exerts an indirectly negative effect on SME innovation when combined with either Source_research or Source_strategic. Of course, this suggestion is tentative, because it is based on only two results from six (i.e., two interaction effects on each of three categories of innovative sales). To this extent, this result conflicts with the positive interactions reported by Mazzola, Perrone, and Kamuriwo (2015), although their findings refer to aggregate measures of open innovation whereas ours refer to individual OI practices.

7 In footnote 6, we tentatively advance evidence suggesting that Source_networks also exerts an indirectly negative effect when it accompanies other external search strategies, by diminishing their effects.

Additional information

Notes on contributors

Dragana Radicic

Dragana Radicic is Senior Lecturer in Economics at the University of Lincoln.

Geoffrey Pugh

Geoffrey Pugh is Professor of Applied Economics at Staffordshire University.