Product innovation, process innovation and export propensity: persistence, complementarities and feedback effects in Spanish firms

ABSTRACT

Drawing on a longitudinal data of Spanish manufacturing firms, this study explores the persistence of technological innovation and exports, their potential complementary relations and feedback effects. Empirical results suggest the presence of both true and spurious state dependence in all three activities. True state dependence in technical innovation and exports implies intertemporal spillovers relevant to the evaluation of innovation and export policy measures. However, given that results also suggest spurious state dependence, firm-specific characteristics should be taken into account in promoting technological innovations and exports. In addition, we find a strong complementarity between product and process innovation both through a contemporaneous effect and via unobserved firm characteristics. However, concerning complementarity between innovation and exports, results suggests complementarity only through contemporaneous effects. Finally, we find no support for the causal link from past product and process innovations to current export activities.

KEYWORDS:

• Product innovation
• process innovation
• exports
• persistence
• complementarities

JEL CLASSIFICATION:

• O31
• C23
• F10
• F14

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ As noted by Esteve-Pérez and Rodríguez (2013, 221): ‘A main challenge in the empirical analysis is to disentangle these two sources of this persistent behaviour.’

² In our empirical strategy, we take into account both arguments: (i) the simultaneous engagement in innovation and exports, and (ii) the endogeneity of innovation with respect to exports.

³ However, our results do not depend on this particular decision as our findings sustain when the first equation has process innovation or exports as the dependent variable. All results are available from the authors upon request.

⁴ Separate regressions overestimate state dependence and the feedback effects. Thus, our preferred model yields more conservative results and does not overestimate relationships that may not necessarily exist. These results are available from the authors upon request.

Additional information

Funding

Sara Ayllón acknowledges support from the projects ECO2016-76506-C4-4-R and 2017-SGR-1571.