Foreign acquisition and post-privatization exit of state-owned firms

Abstract

This paper analyzes the impact of foreign and domestic ownership on the exit rates of privatized state-owned enterprises (SOEs) in transitional countries. The exit of privatized SOEs can have a profound impact on employment and on the development of local economies of transitional countries. An oligopoly model that incorporates country-level trade costs and individual SOE's productivity is developed to assess the exit of SOEs under either foreign or domestic ownership. The model shows that market competition between firms can lead to liquidation of the SOE by a domestic firm when trade costs increase. When the productivity of SOE is high, neither foreign nor domestic firm will liquidate. The predictions of the model are tested using firm-level privatization data from Central and Eastern Europe. By controlling for productivity, trade costs, and other attributes of SOEs after privatization, it is found that foreign ownership significantly reduces the probability of SOE's exit as compared to domestic ownership. Furthermore, there is evidence that as trade costs increase, the exit probability of domestically owned SOEs increases and the exit probability of foreign-owned SOEs declines.

JEL Classifications:

• F23
• P31

Acknowledgements

I gratefully appreciate the helpful comments from James Markusen, Scott Savage, Anna Rubinchik, Jennifer Lamping, and three anonymous referees.

Notes

 1. According to Privatization Barometer Report (2009), world privatization was valued at €65.84 billion in 2009. Countries studied in this paper, for example, had value of €1,626.26 million in Poland, and €18.25 million in Czech Republic. Poland planned to privatize another 700 firms between 2008 and 2011.

 2. For example, World Investment Report (2005) reports that in South-East Europe large privatization has contributed to FDI inflows by $11 billion in 2004.

 3. For a review of the theory on MNEs and FDI see Markusen (1995).

 4. As shown by Melitz (2003), and Helpman, Melitz, and Yeaple (2004) multinational firms are more productive than domestic firms which leads them to engage in international trade and FDI.

 5. In our model we omit the possibility of a foreign firm entry through Greenfield investment as this form of entry is not present in our data which includes only direct acquisition of SOEs by domestic and foreign firms.

 6. We assume that parameters α and β are such that consumer does not attain a satiation point. Furthermore, the quadratic utility function assumes that the marginal utility of income is fixed. By adding the numeraire good z, the utility function becomes quasi-linear and the marginal utility of income is then unity.

 7. This assumption is made to simplify the exposition as we do not address the impact of acquirers' productivity on SOEs exit in the paper. We also make this assumption as we cannot measure the acquirers' productivity using our data due to lack of information.

 8. Cournot–Nash framework presented in the model fits well with privatization; for example, the majority of our privatization data contains manufacturing industries producing final goods for sale.

 9. In the paper, we are mainly concerned with horizontal FDI although we acknowledge that there are situations where FDI is of vertical nature where MNEs acquire SOEs for production of intermediate goods. Data used in the paper do not provide a clear indication of whether a given SOE produces final or intermediate goods. However, examining individual firms majority appear to be final goods producers. There are also firms as steel and cement manufacturing plants that produce both final and intermediate goods.

10. Auction framework is used as the government does not know the valuation that each bidder has for asset k; the bidders have full information regarding asset k. This is consistent with privatization examples where bidders conduct full due diligence to learn about the target. Furthermore, the second-price seal-bid auction has well-known properties which allow us to focus on firm interaction after privatization and not on the actual privatization mechanism. Also, assuming that firms were sold via an auction fits well with the privatization data used in the paper where we have SOEs that were only directly sold by the government.

11. BvDEP (n.d.) provides firm-level data used in papers such Helpman, Melitz, and Yeaple (2004), and Egger and Loretz (2010).

12. There are other forms of privatization that governments used to dispose SOEs including sale to managers, employees, and voucher privatization.

13. The main databases used in the search were: Bloomberg Business week ‘ http://investing.businessweek.com/research/company/overview/overview.asp’, ISI Emerging Markets ‘ http://www.securities.com/’, Alacra Store ‘ http://www.alacrastore.com/’. A more complete list for each country is available upon request.

14. As a robustness check, all analysis is also conducted by excluding Bulgaria due to the high number of firms that exited in Bulgaria. The main results do not change and are available upon request.

15. Related papers that use this method of calculating productivity include Alvarez and Görg (2009), and Frazer (2005).

16. Because productivity is an important variable in our paper, we also obtain a third measure of productivity using index numbers (INs). We calculate IN productivity using Caves, Christensen and Diewert (1982b) index where productivity level of firm i at time t is

where Q is output, L is labor, K is capital, and is the fraction of wage bill in output. The mean productivity using INs in our data is −0.631 with a standard deviation of 1.621, and the median is −0.504. Given the restrictive assumption of IN productivity described in Caves, Christensen and Diewert (1982b) and that for our data majority of firms have negative productivity using this method, we do not report our results with IN productivity in the main body of the paper but only in the Appendix in Table A2. Our main results are unchanged.

17. These definitions for capital intensity and size are used by Bernard and Jensen (2007); they also show that controlling for multi-plant firms is important.

18. For an exact step-by-step with table examples explanation of how to obtain transport margin of an industry please refer to Eurostat Manual of Supply, Use and Input–Output Tables (2008) on page 77.

19. For a full description of the creation of quality of political institutions for each country see Kuncic (2012). The data also include quality of economic institutions and quality of legal institutions which are highly correlated with the quality of political institutions and were used as robustness check; results did not change.

20. For all our productivity measures in Table 5, we remove several extreme observations which were more than five standard deviations above the mean. As robustness, all our estimations were also redone with this and there was no change in our results.

21. We follow the literature on exit in selecting probit estimation as studies such as Bernard and Jensen (2007) and Alvarez and Görg (2009) only use probit analysis as other panel estimation is not appropriate because exit of a given firm is observed only once across all years.

22. As described by Wooldridge (2002, 621–33), and used for example by Sabirianova, Svejnar, and Terrell (2005), we estimate 2SLS where the first stage in the following probit model

where D_i is the deal value of a firm during privatization and R_t−1 is lagged governments non-tax revenue.

23. For a complete description of the SRE, see Dong and Lewbel (2011).

24. Results in Table 6 are robust to using other available firm control variables such as revenue, and profit. Also, results are robust to logit specification where error term has a logistic distribution. We estimate results in Table 6 dropping Bulgaria; the main results do not change. Furthermore, we estimate the results in Table 6 for manufacturing firms only and find that results are consistent.

25. We also run all the IV estimations with both Government non-tax revenue and deal value, the results do not change.

26. In appendix Table A4, we also present the IV estimations where we use IN methodology to calculate productivity.