Does innovation activity affect trade openness? An ARDL bounds testing approach for 10 European countries

Abstract

International differences in technological activity are considered a fundamental factor in explaining differences in both levels and trends between countries’ exports, imports, and income, especially among industrialized nations. This study examines the effects of innovation activity (measured by R&D expenditure and patent applications) and economic growth (GDP per capita) on trade openness (the ratio of imports plus exports to GDP) in 10 European countries from 1983 to 2018 using time series analysis. We also investigate the impact of the Lisbon and Europe 2020 strategies on trade openness. We use autoregressive distributed lag methodology and Granger causality tests based on the vector error correction model to conduct the analysis. The empirical findings indicate a strong long-run relationship among the variables in all the countries examined. In most countries, the long-run coefficients of R&D expenditure, patent applications, and GDP per capita are positive and statistically significant. In short-run dynamics, error correction models are well-defined and provide interesting results for each country. The effects of the Lisbon Strategy and Europe 2020 differ across the countries examined. Finally, the Granger causal relationships among the variables vary across countries.

KEYWORDS:

• Trade openness
• innovation
• Europe
• ARDL
• causality

JEL Classifications:

• C22
• F14
• O32

Acknowledgments

The authors would like to thank the editor David Giles and an anonymous reviewer for their very helpful critical feedback in an earlier version of the paper. The authors are also grateful to Slavo Radosevic for his insightful remarks which tackle issues in this paper. The authors would also like to thank their colleague Vasiliki Georgatzi for helpful comments and suggestions. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Their sample consists of Germany, Denmark, Spain, Finland, France, Italy, Norway, the Czech Republic, the United Kingdom, Sweden, and Switzerland.

2 In each country, the optimal lag lengths of variables were determined based on the Schwarz Bayesian Criterion (SBC). ARDL bounds testing results for cointegration are available upon request.

3 The Breusch–Godfrey test indicates high autocorrelation, at the 1% significance level of the ARDL model for Belgium.

4 Further, the CUSUM and CUSUMSQ tests confirm that all models are stable. The correlation between the two innovation proxies has been found weak or moderate across countries. All results are available upon request.

5 This terminology (‘Bismarckian’, ‘Scandinavian’, ‘Southern’, and ‘Anglo-Saxon’) and the groups of welfare state typologies have been used by Ferrera (1996).

6 For each country, the lag length selection of k was based on the Akaike information criterion (AIC), final prediction error (FPE), and Hannan–Quinn (HQ) information criterion.

7 The outcomes are not shown for the sake of brevity but are available upon request.

Additional information

Funding

This work was partially supported by the Hellenic Foundation for Research and Innovation (HFRI) under the HFRI PhD Fellowship grant (Fellowship Number: 455).