Skip to main content

Advertisement

SpringerLink
Log in

R&D subsidies and productivity in SMEs

  • Published:
Small Business Economics Aims and scope Submit manuscript

Abstract

This paper examines the effect of R&D subsidies on labour productivity. We use firm-level data on Finnish SMEs from 2000 to 2012 and apply a combined matching and difference-in-differences method to control for selection bias. We find no significant positive effect on labour productivity over the five-year period after a subsidy is granted. However, the results vary over time and indicate a 2–4 % negative effect on SMEs’ annual productivity growth one to 2 years after the subsidy year. Nevertheless, subsidies generate a positive employment effect and enhance firm survival. Additional scrutiny reveals that subsidies positively affect the human capital level of low-skill firms.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. Empirical literature indicates that returns on R&D differ by firm characteristics (e.g. Hall et al. 2008; Hall et al. 2009; Ortega-Argilés et al. 2010; Ortega-Argilés et al. 2011).

  2. Many authors have reported that R&D subsidies stimulate private investments (e.g. Almus and Czarnitzki 2003; Hyytinen and Toivanen 2005; Czarnitzki 2006; Görg and Strobl 2007; Özçelik and Taymaz 2008; Aerts and Schmidt 2008; Hussinger 2008; Meuleman and De Maeseneire 2012; Czarnitzki and Lopes-Bento 2013), but studies have also found that these funds partly or fully crowd out some private investments (e.g. Wallsten 2000; Lach 2002; Busom 2000; González and Pazó 2008; Gelabert et al. 2009; Bronzini and Iachini 2014).

  3. Subsidies have been found to enhance employment growth (e.g. Girma et al. 2008; Koski and Pajarinen 2013, Link and Scott 2013; Moretti and Wilson 2014), patent development/innovations (e.g. Czarnitzki et al. 2007; Bérubé and Mohnen 2009) and sales/investments (e.g. Criscuolo et al. 2012; Einiö 2014; Cerqua and Pellegrini 2014).

  4. Recent empirical literature indicates that absorptive capacity (or “learning capacity”) at a regional level is essential for subsidy effectiveness (e.g. Griffith et al. 2003, 2004; Becker et al. 2013).

  5. This information is based on official statistics by Statistics Finland. E-publication (in English) is available at: http://tilastokeskus.fi/til/tkke/2011/tkke_2011_2012-10-31_tie_001_en.html.

  6. For more information on Tekes programs, see: http://www.tekes.fi/en.

  7. Alternatively, productivity can be measured by calculating total factor productivity via a production function rather than calculating labour productivity from the raw data. We use labour productivity to avoid a priori assumptions, which are needed to estimate total factor productivity.

  8. The causal interpretation of the results may also be sensitive to log linearisation (see Fisher and Ciani 2014).

  9. In our sample, 899 firms received only a direct subsidy and 64 firms received only a loan-based subsidy (for more details on R&D subsidies, see Table W1 in the online appendix). The results remained qualitatively similar when we focused on firms that received only a direct subsidy (see online appendix Table W9).

  10. The DID method is based on differencing outcomes before and after a treatment. Thus, if a firm can anticipate that it will receive a subsidy (and changes its behaviour, e.g. hires new employees), then DID estimates can be biased (for details, see Heckman and Smith 1999). As robustness checks, we also repeated the estimations by matching firms three and four years before a subsidy was granted. Again, our results remained qualitatively unchanged.

  11. Following Aerts and Schmidt (2008), employment and turnover are in logarithms to avoid potential biases caused by skewness of the data.

  12. We plot averages of our outcome variables over the three-year period before the subsidy year (and after) to show descriptively that the outcomes in the subsidised firm group developed in parallel to those in the unsubsidised firm group. Figures are available in the online appendix.

  13. This might indicate that funding authorities are attempting to follow the so-called picking-the-winner strategy (i.e. subsidies might be granted to relatively good firms rather than to marginal projects or firms that suffer market malfunctions, e.g. Cantner and Kösters 2012).

  14. Table W2 in the online appendix further shows how the matching method succeeds in removing significant differences for all used covariates between subsidised and unsubsidised firms.

  15. We also rerun our estimations by excluding the covariates that measure employees’ education and earlier subsidies from the matching model. In this case, productivity growth was more rapid, and the effect was significant and positive five years after the treatment. This indicates that without controlling for firm-specific education variables and covariates regarding earlier subsidies in the estimations, the estimation results would be biased upwards [as noted earlier by González and Pazó (2008)]. This finding highlights the need to consider the selection problem in different evaluations of subsidies.

  16. By higher tertiary education, we mean employees who have master’s, licentiate or PhD degrees. The median share of employees with higher tertiary education (in our sample of subsidised firms) is 4.6 %. The correlation between firm size and the share of higher tertiary education is relatively small (−0.12); see online appendix Figure W2 for illustration. As a robustness test, we also study the subsidy effect on productivity by firm size, but the results are mainly insignificant (online appendix Table W5).

  17. It should be stressed that the goal of the R&D subsidies is also to enhance firms’ innovation capacity, the benefits of which (such as enhanced productivity) are observed in the long term.

  18. We focus on the three-year period after the treatment year because employee education data were not available for later years.

  19. In online appendix Table W6, we show that our results by firm skill groups are robust even if we use different cut-off points for low- and high-skill firms.

  20. We focus on subsidy years 2004–2007 to study possible placebo effects due to data constraints. Placebo effect is calculated by moving one year “window” (a difference between reference year and comparison year) to pre-treatment period.

  21. In online appendix (Panel D of Table W3), we match firms exactly by sector, because propensity score matching might compare firms from different sectors. Results show that this has no impact on our results. Table W10 of online appendix shows that initial decline in productivity is, on average, more significant (also statistically) in industrial sector than in service sector.

  22. In our study period, 32 % of subsidised firms exited our sample during the five-year period.

  23. From our data, we cannot distinguish between mergers, acquisitions and firms that went out of business. Still, most business exits indicate that the business was unsuccessful (Coad 2013). Firm survival is studied similarly in earlier literature (e.g. Hyytinen et al. 2014).

  24. We studied descriptively how starting/initial (year) productivity is related to firm survival. Simple cross-tabulations show that subsidised firms whose initial productivity is below median productivity are more likely to survive than subsidised firms whose initial productivity is above the median (see online appendix Table W11). Descriptive examination suggests that our results are not driven by the possible survivorship bias. This might indicate that subsidies are more important for low-productivity firms.

  25. It should be stressed that we cannot observe whether R&D subsidies also affect unsubsidised firms. Our empirical approach does not allow spillovers (the so-called stable unit treatment value assumption; see Imbens and Wooldridge 2009). If subsidies also affect unsubsidised firms, then the subsidy effect on productivity might be under- or overestimated. When more accurate databases become available for research, it would be interesting to evaluate whether different research results that focus on subsidy effects are affected by this possibility.

References

Download references

Acknowledgments

We are grateful to Mika Haapanen, Mika Maliranta, Ari Hyytinen, Jaakko Pehkonen, Hannu Tervo, Petri Böckerman and Terhi Maczulskij for their helpful comments. We thank the editor and two anonymous reviewers for their constructive suggestions, which helped us to improve the manuscript. We would also like to thank Tekes and Statistics Finland for their help with the data, which were made available through remote access. Karhunen gratefully acknowledges the financial support from the Alfred Kordelin Foundation, the Yrjö Jahnsson Foundation (Grant number 6029) and the University of Jyväskylä. The usual disclaimer applies.

Author information

Authors and Affiliations

  1. Jyväskylä University School of Business and Economics, University of Jyväskylä, P.O. Box 35, 40014, Jyväskylä, Finland

    Hannu Karhunen

  2. Pellervo Economic Research PTT, Eerikinkatu 28 A, 00180, Helsinki, Finland

    Janne Huovari

Authors
  1. Hannu Karhunen
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Janne Huovari
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Hannu Karhunen.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (DOCX 570 kb)

Appendix

Appendix

See Tables 6, 7, 8 and 9.

Table 6 Variable definitions
Full size table
Table 7 R&D subsidy effect (ATT) on employment and value added by employee skill level (CDID)
Full size table
Table 8 R&D subsidy effect on labour productivity before and after the subsidy (CDID)
Full size table
Table 9 R&D subsidy and survival propensity over the five-year period after the subsidy year
Full size table

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Karhunen, H., Huovari, J. R&D subsidies and productivity in SMEs. Small Bus Econ 45, 805–823 (2015). https://doi.org/10.1007/s11187-015-9658-9

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11187-015-9658-9

Keywords

JEL Classifications

Search

Navigation