Regional development and foreign direct investment in transition countries: a case-study for regions in Ukraine

ABSTRACT

This paper evaluates the effects of foreign direct investment (FDI) on the economic development of Ukrainian regions based on selected indicators (gross regional product, change of gross regional product, share of the industrial (manufacturing) sector, and employment and unemployment rates) in the 2003–2016 period. Employing an exploratory panel time-series approach, the results suggest that there is only a limited impact of FDI on the economic performance of the regions. The small influence of foreign direct investments is, among others, a consequence of political instability, weak governance, the military conflict in the East of the country, and incompleted reforms of the Ukrainian economy. We suggest that it is necessary to focus on targeted programs at both the regional and state levels in order to enhance the existing state of the economy. It is also important to ensure stability and transparency in the legislative processes, tax reforms, and other policy fields to facilitate the attraction of foreign direct investment, the creation of new jobs, and the increase of the income levels of the Ukrainian population.

KEYWORDS:

• Foreign direct investment
• Ukraine
• regional development
• vector-error-correction

1. Introduction and background

For transition countries, foreign direct investments (FDI) may be of significant importance for the structural change of the economy, for improving international competitiveness, and for economic and social development in general. However, the empirical evidence on the importance of FDI in transition countries is mixed (see the literature review in section 2). Ascertaining the effects of FDI in transition countries may be difficult both on methodological and economic grounds (e.g., Welfe, 2013; Weyerstrass, 2008; Weyerstrass et al., 2001). In general, economic, political and social development might be weakened by the lack of stable governance institutions, disrupted by military or ethnic conflicts, or enhanced by democratic movements and political stability (e.g., Aisen and Veiga, 2013; Clemens, 2010; Jong-A-Pin, 2009). Besides the improvement of these frameworks and institutions, predicting economic and social indicators or testing for the influence of potentially significant contributors to development can be challenging tasks in countries in transition from post-communist to democratic social-market economies. For Ukraine, military-ethnic conflicts are of specific importance and have been posing great risks to the territorial integrity as well as to the economic and social development of the whole country (e.g., Lutz, 2017).

For many transition countries, the design of economic models can be especially complicated from a methodological point of view. First, time series may be limited, and statistics and data may be unreliable owing to the transformation and transition of official statistical systems, as well as to a rapid change of economic structures alike. Second, high inflation rates, largely varying exchange rates, political turbulences, and eventually military or ethnic conflicts may lead to variations in data, which cannot be accounted for in economic models with sufficient precision. Third, some economic and social data may be distorted by statistical artefacts such as altered regional or economic and social statistical classifications.

For Ukraine, empirical studies on the effects of foreign direct investment (FDI) on economic or social development are rare. None of the existing studies deals with a regional perspective that would assess the economic effects of FDI over time and across regions. This paper adds to the literature on the effects of FDI by studying the panel characteristics of economic development, and FDI, in the Ukrainian regions. Methodologically, vector-error-correction (VEC) estimations are used to identify potential relationships between regional GDP and FDI. Besides regional GDP, this paper also tests whether significant correlations can be found between FDI, the unemployment rate, the ratio of employment, and as well industrial development measured by the GDP share of industrial production. Thus, the goal of the paper is to evaluate the effect of foreign direct investment on economic development of Ukrainian regions based on the selected indicators in the 2003–2016 period. Our underlying hypotheses to be tested empirically – based on these considerations as well as on the empirical literature review presented in section 2 below – are twofold. First, we hypothesise that FDI promotes economic and social development in terms of GDP growth, employment, and industrial production. Second, FDI itself is influenced by the absorption capacity of regions. Therefore, the regional effects of FDI are not as clear-cut as it may be expected. Furthermore, we argue that the economic and political problems, especially in the Eastern part of Ukraine, aggravate the various obstacles for a smooth economic and social development in all Ukrainian regions.

The results of this paper suggest that there is no clear causal and unidirectional connection between FDI and important development indicators. Rather, the unstable political conditions and the military conflicts in Eastern Ukraine seem to diminish the potentially large supporting role of foreign direct investment.

The structure of the paper is as follows: A brief literature review on the empirical evidence of the effects of FDI on GDP especially in transition countries is presented in Section 2. Descriptive and econometric results are presented and discussed in Section 3, while in Section 4, the results are summarised and conclusions are drawn.

2. Regional effects of foreign direct investment in transition countries

2.1 Effects of FDI on the whole economy

It should be noted that there are many theoretical and empirical studies dealing with the various effects of foreign direct investment (FDI) on economic development of countries and regions. Nevertheless, there is no consensus among researchers about the role of FDI in economies in transition. We divide selected publications relevant for our study into two main groups. The first group of papers reviewed is related to the study of the influence of FDI on the development of countries, economic branches, and businesses from a macro-economic perspective considering the whole economy. The second strand of papers selected deals with the effects of FDI on the regional level.

Recent and comprehensive studies include the one by Whalley and Xin (2010) who analyse the effect of inward FDI on economic growth in China. Their results confirm that foreign direct investment significantly contributes to the country’s economic development. China’s economic growth and exports would be much lower without FDI flows. Chan et al. (2014) use a systematic approach aimed at the consideration of short- and long-run flows of causality of foreign direct investment and its possible determinants. In both cases, growth in FDI has a substantial direct impact on GDP as well. However, there is also a reverse positive effect of GDP on foreign direct investment (higher GDP growth attracting more FDI). Zhang (2014) evaluates this role of foreign direct investment on China’s industrial competitiveness, and suggests that FDI have significant favorable effects on Chinese industrial development, especially in low-tech manufacturing industries compared to medium- and high-tech industrial sectors. In our paper, we therefore test, based on data availability, the effects of FDI on GDP and industrial development.¹

Investigating the paths of influence of FDI on the Croatian economy, Sohinger and Harrison (2004) note the favourable role of FDI in supporting macroeconomic stability of the country, improving the business and institutional environments, increasing total factor productivity as well as a more efficient allocation of resources. Based on a firm-level database for eight Central and East European countries, Falk (2015) explores the relationship between FDI and innovation activities of domestically owned companies. The results confirm that there is a positive association between local firms and the presence of foreign-owned companies in customer industries if the productivity gap among these two types of firms is not substantial. The researcher also finds a direct positive link between FDI and technological progress and innovations of domestic firms.

In the same line of arguments, Kim and Li (2014) examine the impact of FDI flows on entrepreneurial activity of 104 countries with special emphasis on emerging and transition economies. They find that foreign direct investment has a positive effect on the creation of new businesses. Moreover and interestingly, FDI may have the highest impact in countries with the lowest level of institutional development, political stability, and human capital. Shepotylo (2012) explores spatial factors of FDI location in 25 transition countries during 1993–2010. There are positive spatial FDI spillovers from neighbouring countries, which contribute to FDI stock and inflows of the country. Though these spillover effects differ by sector, from strong and positive in the non-tradables and services sectors to weak and negative in the manufacturing sectors. It therefore seems that the empirical evidence on the stabilising or promoting functions of FDI is mixed.

Škuflić and Botrić (2006) study the determinants of foreign direct investment in Southeastern European countries. The researchers identify both GDP growth and labour cost to be essential for attracting FDI. Their findings show that the main FDI destination is the service sector, which is connected to the specific features of the privatisation processes. It is revealed that the share of the service sector, and the foreign exchange markets have strong effects on FDI decisions in this sector. Similarly, exploring the influence of FDI on the Croatian economy, Jakšić et al. (2018) suggest that this investment has an unfavourable impact on the export of goods and is mostly directed to services, and not to the industrial sector. As stated by Estrin and Uvalic (2016), FDI does not have a significant impact on manufacturing sectors (namely, their value added, employment, and exports) of the Western Balkans economies. These results are very interesting for our study since we test the correlations between FDI and industrial production (as complement to the service sector).

Jimborean and Kelber (2017) suggest that, in general, FDI had a favourable impact on economies of the Central Asian and Eastern European countries in 1993–2014. This positive influence increased during the 2007 crisis, while it became non-significant for the 2011 crisis period. It therefore seems to be relevant as well to consider economic and political shocks when exploring the role of FDI in economic development.

Research results opposite to the literature reviewed so far, are obtained by Chen et al. (2011). They argue that foreign direct investment negatively influences the wage level of Chinese firms, which leads to an increase of inter-enterprise wage inequality. McLaren and Yoo (2017) pay attention to the fact that inward FDI flows may result in a small decrease of the standard of living of Vietnamese households in which a family member does not work in a foreign company. At the same time, the investment leads to a modest income growth of households that have a family member employed in a foreign firm. Furthermore, investments in infrastructure development do not promote economic growth if the quality of infrastructure is already high (e.g., Getzner, 2012).

Eren and Zhuang (2015) study the effect of various types of foreign direct investment on economic development of 12 new EU member states during 1999–2010. It is determined that mergers and acquisitions, and greenfield investment do not play an important role on growth of the countries’ economies. Mergers and acquisitions need to be complemented by an advanced financial system in order to have a positive effect on national economies, while greenfield investment ought to be accompanied by human capital development at the appropriate minimum rate.

Zulfiu-Alili (2014) estimates how inward FDI affects wage inequality within Macedonian enterprises. The analysis proves that foreign-owned companies pay higher salaries for workers than domestically owned firms. Another fact revealed is that larger wage inequality is observed in foreign-owned companies, compared with their domestic counterparts.

Nath (2009) investigates the role of FDI flows on growth of per-capita real GDP of 13 transition countries located in Central and Eastern Europe and the Baltic region in 1991–2005. It is revealed that, unlike trade and domestic investment, FDI flows do not have any substantial impact on growth of these economies.

For our study, the literature review so far has indicated that FDI may affect economic development in various ways. Not only GDP or industrial production are important; relevant factors are also wage inequality and external trade relations. However, for some of these potentially interesting effects, data are not available on the regional level over time. We therefore now turn to the effects of FDI in a regional perspective.

2.2 Effects of FDI on regional economies

The other group of papers is devoted to the influence of FDI on growth and inequalities at the regional level. Analysing the spatial distribution of FDI in China, Zhao et al. (2012) study the concentration of this kind of FDI in a few Chinese provinces. The polarisation and concentration of FDI in a limited number of regions largely lead to social inequality and further expansion of labour-intensive manufacturing industries. As will be seen in this paper in the empirical sections below, FDI in Ukrainian regions are very unevenly distributed as well, leading to large spatial disparities of regions both regarding FDI and economic and social development.

However, some other authors obtain different results with respect to China. For instance, Mah (2013) states that, unlike international trade, there is mixed evidence of the influence of FDI on income inequality in China. Yu et al. (2011) also note that FDI has an insignificant effect on China’s regional income inequality, in contrast to provincial per-capita physical assets. Moreover, this impact has been constantly decreasing since 2002.

Lal (2017) investigates relationships among FDI, trade openness, and GDP in China, India, and Mexico in the short and long run. The findings show that while substantial differences on the indicators between the chosen countries are identified in short-run causal relationships, the existence of long-run causal relationships among FDI, trade openness, and GDP are found for China and Mexico. Methodologically, the study by Lal (2017) suggests that an empirical model should consider both short- and long-term relations between variables. The method employed in our paper takes up this suggestion by using a vector error-correction model that distinguishes between different temporal adaptions to foreign direct investment.

Anwar and Nguyen (2014) explore the effect of FDI and FDI spillovers on total factor productivity (TPF) of manufacturing firms located in eight regions of Vietnam. This analysis shows that their influence is different among the investigated regions. It means that the presence of foreign companies has a positive impact on regional technological development, but the rate of this advancement differs substantially across regions.

Bresslein et al. (2019) investigate the effects of agglomeration externalities on the allocation of FDI in Polish regions. The research results confirm that while agglomeration raises the probability of FDI attraction at the regional level, the unemployment rates and low wage levels have an unfavourable impact on its location probability. Some of these variables are also used in our approaches such as regional GDP (gross regional product), attraction of regions for FDI, and unemployment rates.

Lengyel and Leydesdorff (2015) determine that FDI and foreign companies have a mixed impact on synergies of innovation systems in Hungarian regions depending on their position. They have a substantial positive effect on the synergy of regions which are located between the Hungarian capital and the Austrian border. At the same time, FDI has an unfavourable impact on synergy in the country’s Eastern and Southern regions, which suggests that the geographical location of regions influences both the attraction and the effects of FDI.

According to Mallick and Zdražil (2018), FDI, together with domestic physical investment and human capital, plays a key role on regional income disparity in the Czech Republic. The researchers argue that the proper allocation of FDI, internal investment, and human capital might substantially reduce per-capita income variation among the country’s regions, contrary to some studies of which few are cited above.

Similarly, Völlmecke et al. (2016) point out that FDI on its own is not capable of sustaining income convergence within the European Union or Central and East European countries. From their point of view, to achieve such convergence, FDI flows have to be accompanied by the development of human capital and technological advancements.

It is worth noting that a limited number of studies are available about the influence of foreign direct investment on Ukraine’s economy. We would like to draw the reader’s attention to several publications on this topic. According to Saha et al. (2018), FDI flows play a significant positive role in the development of industries of the country. Various aspects and peculiarities of FDI flows from Poland to Ukraine are considered by Cywiński and Harasym (2014) and Nowak et al. (2013). This research suggests that there are many factors that hamper the attraction of FDI into the Ukrainian economy. There is also a need to improve the investment climate in the country, and enhance political stability, governance frameworks, and end the military conflicts. Crane and Larrabee (2007) point out that it is important to elaborate a special strategy on FDI in Ukraine, setting clear priorities and sequencing.

Summing up, the literature on the effects of FDI is mixed both for whole economies (economic development, growth) and regional economies. While some papers find a significant effect of FDI on economic development, others stress the importance of complementary economic policies (human capital formation) to support the effects of FDI. Some papers even argue that regions developing more strongly than others are those attracting higher FDI; the latter may point to a reverse causal relationship between FDI and economic development. In addition, institutional and political stability influence both the attraction and the effects of FDI in a whole country as well on the regional level.

3. Descriptive and econometric results

3.1 Descriptive results

As briefly outlined in the introduction, the main goal of this paper is to ascertain the effects of FDI on regional GDP (Gross Regional Product, GRP), employment and unemployment, and the GDP share of industrial production, in a panel setting using both the time-series characteristics of the variables, and the cross-section variations. The approach is therefore exploratory and employs vector error-correction estimations (for more details, see Section 3.2.). Table 1 contains an overview and short description of the variables used in the estimations.

Table 1. Variables considered in the analysis.

Variable	Description
FDIit	Foreign direct investment (FDI), per capita, in the ith Ukrainian region; constant 2010 prices (UAH)
GRPit	Gross regional product (GRP), per capita, in the ith Ukrainian region; constant 2010 prices (UAH)
GRPCit	Change of gross regional product (GRP), per capita, in the ith Ukrainian region; constant 2010 prices (UAH), from t-1 to t (%)
INDit	Share of the industrial (manufacturing) sector (% of total GRPit)
EMPLit	Employment (ratio of employment to total population aged 15–70 years, %)
UNEMPLit	Unemployment rate (%)

t = period from 2003 to 2016; i = 27 regions in Ukraine.

Own calculations, 2019.

The data for this paper are taken from the State Statistics Service of Ukraine (www.ukrstat.gov.ua). The raw data (e.g., GRP for the 27 regions) are normalised in order to reflect the size of the regions, and are adjusted for inflation (GDP deflator) which varied largely in the observation period (2003 to 2016), resulting in several time series (e.g., GRP_it per capita in UAH, constant 2010 prices). Furthermore, FDI was available in USD and therefore had to be transformed based on the respective average exchange rate for the single years of the observation period. Regarding availability and reliability of data, the database does not include the occupied territory of the Autonomous Republic of Crimea, the city of Sevastopol, as well as the zone of the military operations in the Eastern part of Ukraine (the database therefore includes 27 regions [cross-sections]).

Figure 1 presents an overview of the economic development of Ukraine from 2003 to 2016. While GDP substantially increased from 2003 until the beginning of the world-wide economic and financial crisis in 2008 (from about UAH 800 billion to close to UAH 1,400 billion), the loss of total income can clearly be detected after 2008. Since then, the Ukrainian economy even faced more difficult times, aggravated by political turmoil, regional military conflicts, and the annexation of Crimea (see, e.g., Ivanov et al., 2016). This also led to high inflation rates (up to 38% in 2015), and to increased exchange rates of the Ukrainian hryvnia (UAH) compared to the EUR or USD. As foreign direct investments are usually denominated in USD, foreign direct investment (FDI) denominated in UAH at current prices are substantial since inflation and the varying and increasing exchange rates have to be taken into account. As can be seen from Figure 1, FDI may be of crucial importance for the Ukrainian economy given the size of inflows from international investors.

Figure 1. GDP and FDI in Ukraine (billion UAH, constant 2010 prices).

Regarding the regional distribution of foreign direct investment, the concentration of economic activities as well as foreign direct investment is presented in Figure 2. Substantial regional disparities with respect to GRP and FDI can be detected, with the capital city of Kyiv standing out both in terms of GRP as well as FDI. In addition to the country’s capital, the highest rates of foreign direct investment are observed in Dnipropetrovsk and Kyiv regions, while the strongest economic development is seen in Poltava and Dnipropetrovsk regions. At the same time, Ternopil, Kirovohrad, and Chernivtsi oblasts have the lowest FDI flows, and Luhansk, Chernivtsi, and Zakarpattya oblasts are the weakest regions in terms of GRP.

Figure 2. Mapping foreign direct investment and GRP of Ukrainian regions (2016, FDI/GRP per capita at constant 2010 prices).

3.2 Econometric results

In order to study the effects of FDI on regions in Ukraine, and given the mixed results of the various papers discussed in sections 2.1 and 2.2, we employ an exploratory panel time-series approach by presenting estimations of the time-series characteristics of the variables, as well as the results of panel VEC (vector error correction) estimations in order to discuss the potential connections between FDI and economic development in Ukrainian regions. Error-correction models basically distinguish between short-term fluctuations and long-term tendencies towards the mean values of variables.

In general, an error-correction model consists of two parts, one that describes short-term influences and fluctuations, and another that mirrors the long-term development towards the mean value of the variables. Under certain conditions such as similar degrees of integration, and cointegration between variables, VEC models can be estimated by means of ordinary least squares.

For instance, if we assume that the dependent variable is GRP (gross regional product), and FDI (foreign direct investment) is the explanatory variable, we can write our empirical model as a panel error-correction equation as follows (Equation 1(1) $\left(GR{P}_{it}-GR{P}_{it-1}\right)=\alpha +{\beta }_{1}GR{P}_{it-1}+{\beta }_{2}FD{I}_{it-1}+{\sum }_{j=1}^L{\beta }_{3j}\left(GR{P}_{it-j}-GR{P}_{it-j-1}\right)+{\sum }_{j=1}^L{\beta }_{4j}\left(FD{I}_{it-j}-FD{I}_{it-j-1}\right)$(1) ):

(1) $\left(GR{P}_{it}-GR{P}_{it-1}\right)=\alpha +{\beta }_{1}GR{P}_{it-1}+{\beta }_{2}FD{I}_{it-1}+{\sum }_{j=1}^L{\beta }_{3j}\left(GR{P}_{it-j}-GR{P}_{it-j-1}\right)+{\sum }_{j=1}^L{\beta }_{4j}\left(FD{I}_{it-j}-FD{I}_{it-j-1}\right)$(1)

GRP and FDI denote gross regional products and foreign direct investment, respectively. The index i denotes the region, t is the year considered. L denotes the lag length (from j = 1 to L, in years), and the α and β parameters are to be estimated. In order to estimate equation [1(1) $\left(GR{P}_{it}-GR{P}_{it-1}\right)=\alpha +{\beta }_{1}GR{P}_{it-1}+{\beta }_{2}FD{I}_{it-1}+{\sum }_{j=1}^L{\beta }_{3j}\left(GR{P}_{it-j}-GR{P}_{it-j-1}\right)+{\sum }_{j=1}^L{\beta }_{4j}\left(FD{I}_{it-j}-FD{I}_{it-j-1}\right)$(1) ], one has to test whether the time series are of the same order of integration, and whether they are cointegrated.

Furthermore, the approach of using a vector error-correction model allows to test for influences in both directions. That means that equation [1](1) $\left(GR{P}_{it}-GR{P}_{it-1}\right)=\alpha +{\beta }_{1}GR{P}_{it-1}+{\beta }_{2}FD{I}_{it-1}+{\sum }_{j=1}^L{\beta }_{3j}\left(GR{P}_{it-j}-GR{P}_{it-j-1}\right)+{\sum }_{j=1}^L{\beta }_{4j}\left(FD{I}_{it-j}-FD{I}_{it-j-1}\right)$(1) can easily be reversed in order to insert FDI as the dependent and GRP as the explanatory variables. By means of this empirically driven estimation procedure, we are able to test for manifold different possible connections between variables without a priori assumptions. We therefore do not presuppose a certain theoretical framework (e.g., some type of regional development theory explaining the determinants of regional economic development) in our paper.

First of all, all time series are tested with respect to their characteristics under a diverse set of assumptions. With respect to limited space, we only present the results of unit-root tests (stationarity of variables) for the two main variables considered in the estimations, GRP_it and FDI_it.² Table 2 presents the results of panel unit root tests for GRP_it by taking several assumptions and specifications into account. The unit root tests with GRP levels indicate that the time series is stationary; this statistical result correlates with the economic development of Ukraine discussed above since income did not grow significantly over the observation period.

Table 2. Panel unit-root tests (variable: GRP_it).

	Levels				1^st difference
	Individual effects		Individual effects, individual linear trends		Individual effects
	Statistic	Prob.	Statistic	Prob.	Statistic	Prob.
Common unit root processes assumed
Levin, Lin & Chu t*	−5.896	0.000	−8.870	0.000	−13.543	0.000
Breitung t-stat			−1.593	0.056
Individual unit root processes assumed
Im, Pesaran & Shin W-stat	−4.793	0.000	−1.952	0.026	−7.959	0.000
ADF – Fisher Chi-square	111.279	0.000	72.025	0.051	164.664	0.000
PP – Fisher Chi-square	297.549	0.000	259.271	0.000	319.448	0.000
Period	2003-2016
n (observations)	291
N (cross sections)	27

H₀: unit-root.

Own calculations, 2019.

A similar result is achieved for the variable FDI_it; again, the tests along a range of different assumptions indicate a stationary time series in levels (see Table 3).

Table 3. Panel unit-root tests (variable: FDI_it).

	Levels				1^st difference
	Individual effects		Individual effects, individual linear trends		Individual effects
	Statistic	Prob.	Statistic	Prob.	Statistic	Prob.
Common unit root processes assumed
Levin, Lin & Chu t*	−3.350	0.000	−5.400	0.000	−8.871	0.000
Breitung t-stat			2.156	0.985
Individual unit root processes assumed
Im, Pesaran & Shin W-stat	−0.163	0.435	−0.749	0.227	−7.542	0.000
ADF – Fisher Chi-square	60.752	0.246	62.964	0.189	159.141	0.000
PP – Fisher Chi-square	68.118	0.094	98.764	0.000	319.936	0.000
Period	2003-2016
n (observations)	291
N (cross sections)	27

Own calculations, 2019.

However, for further econometric estimations, the time series are also tested regarding their potential cointegration. As Table 4 shows, the tests are ambiguous; depending on the chosen test statistic and the specification of the underlying cointegrating equation, the cointegration of the two variables (GRP_it, FDI_it) is confirmed in same cases. With respect to the influence of foreign direct investment on regional GRP, this result which is in line with the literature discussed in section 2 suggests that there are no clear-cut unidirectional (causal) effects of FDI_it on GRP_it.

Table 4. Panel cointegration test (variables: FDI_it, GRP_it).

	Individual intercept				Intercept and trend
	Statistic	Prob.	Statistic (weighted)	Prob. (weighted)	Statistic	Prob.	Statistic (weighted)	Prob. (weighted)
Alternative hypothesis: common AR coefs. (within-dimension)
v-Statistic	2.542	0.006	2.134	0.016	−1.368	0.914	−1.610	0.946
rho-Statistic	−2.403	0.008	−2.900	0.002	0.465	0.679	0.125	0.550
PP-Statistic	−5.916	0.000	−8.057	0.000	−6.141	0.000	−7.110	0.000
ADF-Statistic	−5.763	0.000	−8.432	0.000	−6.050	0.000	−7.120	0.000
Alternative hypothesis: individual AR coefs. (between-dimension)
rho-Statistic	−0.148	0.441			2.399	0.992
PP-Statistic	−8.936	0.000			−6.661	0.000
ADF-Statistic	−9.350	0.000			−6.569	0.000
Period	2003-2016
n (observations)	405
N (cross sections)	27

H₀: No cointegration

Own calculations, 2019.

Tables 5-9 show the results of the vector-error correction estimations for the relations between foreign direct investment (FDI_it) and regional (economic) development. In these estimations, the level of GRP_it, economic growth (GRPC_it), the share of industrial production (IND_it), the rates of employment (EMPL_it) and unemployment (UNEMPL_it) are considered. A crucial assumption has to be made regarding the lag length of the models. Several tests suggested that a lag length of 4 years would be adequate. There is, however, a certain trade-off regarding the lag length since the observation period is in total 14 years; increasing the lag length further would compromise the time-series character of the estimation and would approximate the estimation to a cross-section study.

Table 5. Vector error-correction estimation (FDI_it, GRP_it).

	Dependent variables
	Est. 1		Est. 2
	GRPit-GRPit-1		FDIit-FDIit-1
Explanatory variables	Coefficient	t-stat.	Coefficient	t-stat.
Constant	910.7534	−1.2716	−2,232.1266	1.8473
GRPit-1	−0.0635	−1.9967	0.1365	6.0837
FDIit-1	0.0592	13.7062	−0.1272	−41.7604
GRPit-1-GRPit-2	0.0336	0.5932	−0.3446	−8.6315
GRPit-2-GRPit-3	−0.4911	−7.5156	−0.0867	−1.8806
GRPit-3-GRPit-4	0.0298	0.4865	−0.0535	−1.2394
GRPit-4-GRPit-5	−0.1724	−2.8920	0.0381	0.9066
FDIit-1-FDIit-2	−0.3714	−3.3480	0.1014	1.2966
FDIit-2-FDIit-3	0.3559	3.5220	−0.1666	−2.3383
FDIit-3-FDIit-4	−0.1892	−1.6387	0.1949	2.3936
FDIit-4-FDIit-5	0.3457	3.2903	−0.1573	−2.1233
Adj. R2	0.2891		0.3182
S.E. equation	2,594.8320		1,829.9210
F-statistic	11.6616		13.2359
Log likelihood	−2,194.3030		−2,111.5310

Period: 2003–2016; N (cross-sections): 27; n (observations): 237.

Own calculations, 2019.

Table 6. Vector error-correction estimation (FDI_it, GRPC_it).

	Dependent variables
	Est. 3		Est. 4
	GRPCit-GRPCit-1		FDIit-FDIit-1
Explanatory variables	Coefficient	t-stat.	Coefficient	t-stat.
Constant	−1.1093	3.4143	19.0269	0.2978
GRPCit-1	−2.1704	−12.1605	−19.1084	−0.5657
FDIit-1	0.0000	−0.9997	−0.0001	−0.0465
GRPCit-1-GRPCit-2	1.0884	6.9946	−18.7809	−0.6378
GRPCit-2-GRPCit-3	0.7057	5.1675	−16.5838	−0.6417
GRPCit-3-GRPCit-4	0.5104	5.6430	−17.9343	−1.0479
GRPCit-4-GRPCit-5	0.1639	2.6410	−17.8552	−1.5205
FDIit-1-FDIit-2	−0.0003	−0.5878	0.1045	1.1315
FDIit-2-FDIit-3	0.0003	0.5595	−0.2184	−2.1570
FDIit-3-FDIit-4	−0.0005	−0.9726	0.1894	2.0956
FDIit-4-FDIit-5	0.0005	1.1215	0.0132	0.1445
Adj. R2	0.6476		0.0572
S.E. equation	11.7347		2,220.7020
F-statistic	43.6746		2.4092
Log likelihood	−809.9896		−1,911.0260

Period: 2003–2016; N (cross-sections): 27; n (observations): 210.

Own calculations, 2019.

Table 7. Vector error-correction estimation (FDI_it, IND_it).

	Dependent variables
	Est. 5		Est. 6
	INDit-INDit-1		FDIit-FDIit-1
Explanatory variables	Coefficient	t-stat.	Coefficient	t-stat.
Constant	0.1054	−7.5091	1,295.1103	−1.5370
INDit-1	−0.1430	−4.8639	−5.8856	−3.0559
FDIit-1	0.0006	7.9008	0.0234	4.9638
INDit-1-INDit-2	−0.0982	−1.8469	6.4306	1.8464
INDit-2-INDit-3	−0.3151	−7.0468	−5.1948	−1.7729
INDit-3-INDit-4	0.1119	2.3751	−3.4261	−1.1098
INDit-4-INDit-5	0.1705	5.4939	−3.0738	−1.5119
FDIit-1-FDIit-2	0.0023	1.7996	0.1006	1.1943
FDIit-2-FDIit-3	−0.0026	−2.1459	−0.1862	−2.3454
FDIit-3-FDIit-4	−0.0004	−0.2698	0.0911	0.9976
FDIit-4-FDIit-5	−0.0020	−1.4510	−0.1436	−1.5765
Adj. R2	0.6157		0.0577
S.E. equation	32.8301		2,151.2280
F-statistic	43.0045		2.6056
Log likelihood	−1,158.6290		−2,149.8690

Period: 2003–2016; N (cross-sections): 27; n (observations): 237.

Own calculations, 2019.

Table 8. Vector error-correction estimation (FDI_it, EMPL_it).

	Dependent variables
	Est. 7		Est. 8
	EMPLit-EMPLit-1		FDIit-FDIit-1
Explanatory variables	Coefficient	t-stat.	Coefficient	t-stat.
Constant	0.1589	−2.4638	−6,096.7960	1.7360
EMPLit-1	−0.0457	−0.8973	1,427.7030	3.1812
FDIit-1	−0.0002	−1.4254	0.0408	5.0537
EMPLit-1-EMPLit-2	0.2307	2.7500	−1,354.3650	−1.8340
EMPLit-2-EMPLit-3	−0.2989	−2.0084	−1,952.1240	−1.4899
EMPLit-3-EMPLit-4	−0.0205	−0.1038	−285.5582	−0.1643
EMPLit-4-EMPLit-5	−0.0753	−0.5727	−361.9655	−0.3126
FDIit-1-FDIit-2	0.0001	−0.4046	0.0766	0.9055
FDIit-2-FDIit-3	−0.0002	0.6766	−0.2057	−2.4761
FDIit-3-FDIit-4	0.0001	0.8813	0.0395	0.4090
FDIit-4-FDIit-5	0.0001	−0.6401	−0.1672	−1.7340
Adj. R2	0.0114		0.0510
S.E. equation	0.2452		2,158.8360
F-statistic	1.3032		2.4098
Log likelihood	1.9321		−2,150.7060

Period: 2003–2016; N (cross-sections): 27; n (observations): 237.

Own calculations, 2019.

Table 9. Vector error-correction estimation (FDI_it, UNEMPL_it).

	Dependent variables
	Est. 9		Est. 10
	UNEMPLit-UNEMPLit-1		FDIit-FDIit-1
Explanatory variables	Coefficient	t-stat.	Coefficient	t-stat.
Constant	0.3490	3.5784	1,453.5204	1.7076
UNEMPLit-1	−0.0093	−0.3278	−172.6033	−3.0924
FDIit-1	0.0001	0.6877	0.0558	6.4878
UNEMPLit-1-UNEMPLit-2	−0.1779	−2.8787	363.2751	2.9788
UNEMPLit-2-UNEMPLit-3	−0.2334	−3.7162	175.7293	1.4181
UNEMPLit-3-UNEMPLit-4	−0.1122	−1.7151	91.5082	0.7087
UNEMPLit-4-UNEMPLit-5	−0.4117	−7.1168	74.5684	0.6533
FDIit-1-FDIit-2	0.0001	0.4847	0.1032	1.2065
FDIit-2-FDIit-3	−0.0002	−0.9481	−0.2288	−2.6949
FDIit-3-FDIit-4	0.0001	0.7553	0.0217	0.2131
FDIit-4-FDIit-5	0.0001	0.7898	−0.1767	−1.7778
Adj. R2	0.2170		0.0619
S.E. equation	1.0878		2,146.3810
F-statistic	8.2678		2.7315
Log likelihood	−351.1224		−2,149.3350

Period: 2003–2016; N (cross-sections): 27; n (observations): 237.

Own calculations, 2019.

The results of Table 5 suggest that there are both long-term relations between FDI and GRP; the coefficient for FDI seems to have a higher explanatory power for GRP (Est. 1) than vice versa (Est. 2). However, the short-term fluctuations of FDI and GRP, respectively, depend on the year of observation. There is no consistent picture with regard to a relation between FDI and GRP as the signs of the coefficient change with the lagging of the variables. Concerning the statistical quality of the estimations, Est. 2 has a slightly better explanatory power than Est. 1. As a first important result of this paper, the estimations provide further evidence that there is no clear-cut and uniformly positive causal relationship effect of foreign direct investment on the gross regional product. It seems that foreign direct investment both enhances GRP; however, higher GRP also attracts higher FDI, thus supporting the hypothesis that the absorptive capacity for funding respective projects attracts FDI as well. A graphical representation of an impulse-response function is depicted in Figure 3. Based on the estimation results, there is again no clear-cut picture; e.g. a higher FDI leads to a drop in GRP in the next two periods while it leads to a substantial long-term increase of GRP.

Figure 3. Impulse-response function based on vector error-correction estimations (FDI_it ←→ GRP_it).

Table 6 presents the estimations with regard to the connections between FDI and regional economic growth (variable GRPC_it). The long-term fluctuations in both estimations (Est. 3 and 4) suggest that foreign direct investment does not influence growth differentials; regional economic growth is significantly correlated to past growth both in the long-term and in the short-term. In turn, economic growth does not attract FDI significantly. While Est. 3 exhibits a satisfactory explanatory power, Est. 4 rejects the notion that higher economic growth positively influences foreign direct investment. Comparing the results of Tables 5 and 6, the levels of FDI and GRP are positively (but not uniformly) correlated. Economic growth, however, is not influenced by FDI, which is also shown in Figure 4 with the substantial drop of GRP growth as a response to higher FDI, and a somewhat erratic response of FDI to higher growth rates.

Figure 4. Impulse-response function based on vector error-correction estimations (FDI_it ←→ GRPC_it).

Figure 5. Impulse-response function based on vector error-correction estimations (FDI_it ←→ IND_it).

Besides the potential effects on economic growth and development, it is also of interest to explore the question whether FDI affects industrial development (variable IND_it). Table 7 again indicates that – similar to the results displayed in Est. 3 (Table 6) – industrial development very much depends on past development (Est. 5). However, the influence of FDI seems to be significant in the long-term term but again erratic in the short-run. Est. 6 suggests that industrial development is not consistently influencing FDI – the explanatory power of this estimation is not satisfying. In combination with the impulse-response function of Figure 4, foreign direct investment enhances industrial development in the short-run (see also Figure 5).

With respect to the level of employment (ratio of employment to total population), the estimations of Table 8 (Est. 7 and Est. 8) and the graphical presentation in Figure 6 suggest that there is basically no consistent correlation between foreign direct investment and the level of employment.

Figure 6. Impulse-response function based on vector error-correction estimations (FDI_it ←→ EMPL_it).

Finally, this paper also includes estimations regarding the linkages between FDI and the unemployment rate (Table 9, Figure 7). While the unemployment rate does not influence FDI (Est. 10) in the short-term, higher unemployment rate might lead to a decrease of FDI in the long-run. However, Est. 9 also suggests that the unemployment rate is rather determined by past realisations than by being influenced by FDI.

Figure 7. Impulse-response function based on vector error-correction estimations (FDI_it ←→ UNEMPL_it).

Thus, the results obtained in the estimations indicate positive effects of foreign direct investment on industrial development in the long-run. The empirical analysis also suggests that the unemployment rate is correlated to a decline of FDI flows in the long-term period. In all other cases, the influence of foreign investments is only short-term or absent. Therefore, we can only see limited effects of foreign direct investment on the chosen indicators of regional development. The results of our research are discussed in more detail in the next section.

4. Discussion, summary and conclusions

The results of the empirical model estimations confirm that there are long-term relationships between FDI and GRP (gross regional product, in UAH per capita). However, these relationships are rather unstable and have changeable directions during the period analysed. We do not find a clear positive influence of foreign direct investment on this indicator of regional economic development in the sense of a stable causal relationship. In our opinion, the instability and absence of a pronounced favourable effect of FDI on GRP might be the result of inconsistent ‘rules of the game’ on the Ukrainian business market, frequently changing legislation in the tax and other policy fields, as well as the impacts of non-market factors. Among others, the military conflict in Eastern Ukraine may be one of the disruptive factors influencing the attraction and the effects of FDI on regional development. That is why it is quite problematic for a significant number of companies with foreign investment to work effectively in the long-term, and, as a consequence, their effect on the economic development of regions is variable and limited, even in regions where FDI flows are quite substantial. In addition, flows FDI are unevenly distributed between regions, which is a result that is also supported by other studies on FDI such as the one by Iwasaki and Suganuma (2007).

Large industrial companies with foreign investment, for example, may also experience certain difficulties in their activities in the short-run due to the influence of the above-mentioned unfavourable factors. Though these companies are able to adapt to such economic environments and operate stably in the long-term perspective. We suggest that this is an explanation of the substantial impact of FDI on industrial development, revealed in this study.

The research findings also indicate that there is no interdependence between FDI and the level of employment. From our point of view, that might be a consequence of the existing imbalance in regional labour markets. This result can be also confirmed by the fact that higher unemployment rates may lead to a reduction of FDI in the long-run. Currently, there is a shortage of workers with the required professional skills in the labour markets of Ukrainian regions (cf. Moroz, 2015). Due to this shortage, enterprises with foreign investments are not capable of operating to the full extent, which negatively affects the attraction of foreign direct investment in the long-term perspective.

Results of some recent publications differ from our findings. For example, Iamsiraroj (2016) investigates the association between foreign direct investment and economic growth in 124 selected countries. The findings confirm that FDI flows promote the growth of countries’ economies and vice versa.

Gomes Neto and Veiga (2013) estimate the influence of FDI on economic growth through the development of technology and innovation using data for 139 countries. The obtained results also indicate that this investment has both direct and indirect effects on the productivity growth and GDP increase. The researchers state that these research results are also applicable to developing countries.

Kottaridi and Stengos (2010) study the nexus among FDI and economic growth in 25 OECD and 20 non-OECD countries. The researchers reveal that a positive influence of flows on the economies’ growth takes place only for countries with a minimum threshold of absorptive capacity (higher income levels). They also find that this investment has a non-linear impact on economies’ growth.

Alvarado et al. (2017) examine the impact of foreign direct investment on economic development of 19 Latin American countries. According to their research, FDI has a favourable impact on countries’ economic growth only in case of high-income states. In contrast, the negative effect of this investment is observed in lower-middle-income economies.

Exploring the role of FDI in economic growth of developed and developing countries, Makiela and Ouattara (2018) note that foreign direct investment positively affects the input growth component of economic growth. However, the empirical studies also show that FDI flows do not exert a favourable influence on total factor productivity growth.

Taking into account the above-mentioned explanations and the empirical studies cited above, we argue that the limited influence of FDI identified on the chosen regional indicators is a consequence of political instability, weak governance, and the military conflict in the east of the country and uncompleted reforms of the Ukrainian economy. That is why there is a need to focus on targeted programs at both the regional and state levels, which could really help to enhance the existing economic situation. It is also necessary to ensure stability and transparency in the legislative, tax, and other policy fields to facilitate the attraction of foreign direct investment, creation of new jobs, and increase of the income level of the population in Ukraine.

Acknowledgments

The authors thank two anonymous reviewers and the editor who provided many helpful suggestions for substantial improvements of the paper. Furthermore, the authors would like to thank the participants of the 2020 Annual Conference of the Austrian Economic Association (NOeG), February 24-25, Vienna, Austria, for their helpful comments. All errors are, of course, the responsibility of the authors.

Disclosure Statement

The authors declare that there is no conflict of interest.

Notes

1. Another group of papers considers the external trade effects of FDI. Anwar and Nguyen (2011) consider the influence of FDI flows on exports, imports, and net export of Vietnam. They discover the existence of a complementary relationship between FDI, and exports and imports. Besides, FDI have a substantial positive impact on net-exports during the post-Asian crisis period.

2. All detailed results for the other variables can also be sent by the authors upon request.