(De)industrialisation and lessons for industrial policy in Central and Eastern Europe

Abstract

Over the past two and a half decades, the economic landscape of Central and Eastern European countries (CEECs) has considerably changed. The demise of traditional industries and the rise of the service sector during the 1990s inclined economic structure towards deindustrialisation. In years that followed, new industries emerged in many of these countries and brought them back on the route of reindustrialisation. Across countries, this process developed at an uneven pace. The recent rise of awareness about the importance of industrial development for the well-being of nations makes it relevant to investigate the sources behind changes in the economic structure of CEECs. Our findings reveal that reindustrialisation takes place at an uneven pace. No support was found for horizontal economy-wide industrial policies advocated within new classical economics. Strong impulse to reindustrialisation comes through improvements in export sophistication. Such findings are in line with those on the position of CEECs in global value chains.

Keywords:

• CEEC
• deindustrialisation
• reindustrialisation
• industrial policy

Notes

1. For universities, this type of engagement is named ‘third mission’ (in addition to education and research), which generally relates to the social mission of engaging in external partnerships related to community needs and supporting economic development, coupled with new challenges determined by the emergence of the ‘learning economy’.

2. Research and innovation strategies for Smart Specialisation.

3. For a detailed explanation, see the Appendix.

4. For detailed results, see Table A1 in the Appendix.

5. Eurostat classification of high-tech manufacturing industries at two-digit NACE rev2 level applied.

6. Several robustness tests were applied. The fixed effects panel estimation with endogenous covariates can be undertaken with means of instrumental variables and generalised method of moments (GMM) estimation. As a robustness check both techniques were applied. Results remain unchanged. The model was also estimated in first differences form. Model diagnostics continue to support our specification and all variables retain their signs. The only variable to lose its significance is the measure of patent applications (epo). Printouts of all estimations are available upon request.