Modeling of FDI in business services: Additional effects in case of Ukraine's European integration

ABSTRACT

To analyze Ukraine's deep and comprehensive integration with the EU, we develop a multi-regional general-equilibrium simulation model incorporating heterogeneous firms and Foreign Direct Investment (FDI) in business services. This allows for consideration of (a) trade growth in new varieties; (b) aggregate productivity changes attributed to reallocation of resources across and within an industry; and (c) productivity growth in manufacturing due to increased access to business services. The results indicate relatively small gains for the EU, whereas Ukraine benefits with a welfare increase of over 8%. The deindustrialization impact, previously found by Olekseyuk and Balistreri (2014) in a comparison of different modeling structures, is supported by our findings. Ukraine's welfare gains are higher under an Armington structure compared to monopolistic competition. This is due to a movement of resources into Ukraine's traditional export sectors producing under constant returns. Implementation of the FDI modeling approach and liberalization of barriers to FDI, however, mitigates the deindustrialization impact as multinational firms enter the Ukrainian market. This increases the number of available varieties and, consequently, induces productivity growth of manufacturing sectors due to improved access to business services as critical inputs.

KEYWORDS:

• Ukraine
• EU
• DCFTA
• CGE
• Armington
• Melitz
• FDI

JEL classification:

• F12
• F14
• C68
• F21

Acknowledgements

We thank Volker Clausen, Edward J. Balistreri, Hannah Schürenberg-Frosch, Miriam Frey and participants at 18th Annual Conference on Global Economic Analysis (Melbourne), 7th Joint IOS/APB/EACES Summer Academy on Central and Eastern Europe, 17th Annual Conference of the European Trade Study Group (ETSG), Research Seminar at the University of Duisburg-Essen for valuable comments and helpful suggestions. All errors are our own.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The signature process of the AA/DCFTA was completed in June 2014; the Ukrainian Verkhovna Rada and the European Parliament ratified the agreement in September 2014. See, e.g. European Council (2014c), European Commission (2014b) for details (available at http://europa.eu/index_en.htm).

2. The EU has temporarily removed customs duties on Ukrainian exports as an autonomous trade measure since April 2014. This unilateral transitional trade measure allows Ukraine to benefit from the advantages offered by the DCFTA even before its implementation in January 2016. See European Council (2014d), European Council (2014a), European Council (2014b), European Council (2014c) and European Council (2014e) available at http://eeas.europa.eu/ukraine/news/.

3. The implementation of the DCFTA was delayed under Russia's pressure until December 2015 while the autonomous trade measures of the EU were extended for this period. See, e.g. European Commission (2014a), European Parliament and the Council (2014) for details.

4. The number is based on data for the first two quarters of 2014 provided by the State Statistics Service of Ukraine, available at http://www.ukrstat.gov.ua/.

5. See Knuth, Giucci, and Chukhai (2010, 9). This is also supported by Kirchner, Kravchuk, and Ries (2015, 6) and Adarov et al. (2015, 56–57) who additionally state that in contrast to business services the share of manufacturing is rather low (25.3% in 2013) illustrating a lack of export-oriented, efficiency-seeking FDI.

6. A negative long-term welfare effect of −0.06% is found for Ukraine by Emerson et al. (2006).

7. Francois and Manchin (2009) present detailed results for seven CIS counties: Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Russia and Ukraine.

8. See Jensen et al. (2005), Maliszewska, Orlova, and Taran (2009), Ecorys and CASE-Ukraine (2007), Francois and Manchin (2009), Movchan and Giucci (2011).

9. Balistreri, Tarr, and Yonezawa (2014) apply a multi-regional simulation model for East Africa and show that the gains from liberalization of FDI in services are dominant in the overall impact (compared to the effects from reduction of NTBs and barriers to efficient trade facilitation) only if countries have relatively high barriers against foreign service providers. This is found for Rwanda, Kenya and Uganda.

10. See, e. g. Cohen and Levinthal (1989) or Mebratie and van Bergeijk (2013, 64) especially for developing and transitional countries.

11. See, e. g. Javorcik (2004), Blalock and Gertler (2008), Kolasa (2008), Havranek and Irsova (2011).

12. The value of 1.25 is adopted from Jensen and Tarr (2012), Jensen, Rutherford, and Tarr (2010), Jensen, Rutherford, and Tarr (2007), Jensen and Tarr (2010), Jensen and Tarr (2008), Balistreri, Rutherford, and Tarr (2009) and Balistreri, Tarr, and Yonezawa (2014).

13. This supply elasticity is used in the partial equilibrium models for Krugman and Melitz formulation.

14. The inter-variety elasticity of substitution sig_i is equal to 3.8 which is consistent with the plant-level empirical analysis of Bernard et al. (2003).

15. Due to data availability, we implement this FDI structure only for Ukraine as the country of interest.

16. A detailed specification of the Krugman model equations is provided in the Appendix A.1.

17. Similar structures are implemented by Markusen et al. (2005), Jensen, Rutherford, and Tarr (2007), Jensen, Rutherford, and Tarr (2010), Jensen and Tarr (2012), Balistreri, Rutherford, and Tarr (2009) and Balistreri, Tarr, and Yonezawa (2014).

18. A detailed description of the model equations can be found in Olekseyuk and Balistreri (2014) and Balistreri and Rutherford (2013). See also Balistreri, Hillberry, and Rutherford (2011) and Bernard et al. (2003) for the parametrization of the model and Balistreri, Hillberry, Rutherford (2010), Balistreri and Rutherford (2012), Corcos et al. (2011), Costinot and Rodríguez-Clare (2014) for discussion of divergence in results across different model structures (e.g., Armington and Melitz) and their empirical relevance.

19. See Tables A2 and A3 for detailed mapping of GTAP regions and sectors to our model.

20. For NTBs, we use the AVEs estimated by Kee, Nicita, Olarreaga (2009); for barriers to trade facilitation see Minor (2013), Hummels (2007), Hummels et al. (2007), Hummels and Schaur (2013); detailed description of aggregation procedure is provided by Olekseyuk and Balistreri (2014).

21. They provide tariff equivalents for the following sectors: accounting and legal services; air, rail, road and maritime transport; banking and insurance; fixed and mobile line; retail. All values except of retail sector are used in our calculations.

22. A bilateral liberalization is not simulated due to the lack of data concerning the FDI firms in the EU (shares of output captured by multinational firms and shares of specialized imported inputs).

23. Melitz structure is incorporated for the following sectors: CNM, MEQ, OMF, TEX, TRD, WPP.

24. Including CMN, FNI, OBS, TRS.

25. Relatively weak trade links of the EU constitute the main reason for the difference in magnitude. Following Olekseyuk and Balistreri (2014), the import shares of the EU from Ukraine are very low with the maximum of 10.6%. The situation is opposite in Ukraine with import shares from the EU over 40% for the IRTS goods.

26. Similar specialization of Ukraine is found by Frey and Olekseyuk (2014) and Olekseyuk and Balistreri (2014).

27. Only manufacture of machinery and equipment (MEQ), textiles (TEX) and wood and paper industry (WPP) demonstrate an increase of imported varieties in Ukraine However, it is not enough to compensate for the losses of Ukrainian domestic varieties.

28. Comparing equilibria t versus t − 1, Feenstra (2010) shows that the variety gains can be measured by deviations in the following ratio from unity: where λ^z_hr is region-r's share of expenditures at equilibrium z on good-h varieties available in both equilibria to the total expenditures on good-h varieties at z.

29. Only in S1 there are some losses in Ukrainian trade sector (TRD) form tariff reduction induced changes in the number of varieties.

30. Only for Ukraine in scenario S1.M we observe a qualitative switch from negative to positive results with the lower value for esubm and sig as well as upper value for esupply, which happens for different reasons. For instance, lower substitutability between foreign goods from differen regions (esubm) leads to the lower increases in trade flows and number of available varieties decreases less, which allows for a small welfare increase. A rise of esupply causes an increase of the top nest elasticity of substitution between sector-specific capital and the rest of the inputs which leads to the slight welfare increase after tariff elimination.

31. According to data for the first two quarters of 2014 provided by the State Statistics Service of Ukraine (available at http://www.ukrstat.gov.ua/), 93.5% of Ukrainian FDI outflows are directed to the EU member countries. A reduction of European barriers to FDI will, therefore, increase the Ukrainian outflows.

32. This is expected as the DCFTA agreement provides for the right of establishment in services and integration of Ukraine into the internal market of the EU in telecommunications, financial services, maritime services and postal and courier services once Ukraine effectively implements the EU acquis in these sectors (see, e.g. Hoekman, Jensen, and Tarr (2014)).

33. The export spillovers from FDI have been found by a number of previous studies, e. g. Greenaway Sousa, and Wakelin (2004) and Kneller and Pisu (2007) for the UK, Sun (2009, 2010) for China.

34. Index ss is also used to indicate a destination region, but other than s which is already used in the equation.

35. The price elasticity of demand is assumed to be equal 0.75.

36. Fixed cost is measured in composite input units as well as the iceberg trade cost τ_krs.

37. This supply elasticity is taken into account by calibrating the top nest elasticity .

38. This technique is also used by Balistreri, Hillberry, and Rutherford (2011), Balistreri, Tarr, and Yonezawa (2014) and Olekseyuk and Balistreri (2014).