Evalution of FDI in CE, SEE and Kosovo in Relation to Growth Rates and Other Indicators

Abstract

Foreign Direct Investment, and related to it, economic growth are among the main indicators in all countries that are looking for their growth. The challenge is for developing and transition countries such as SEE and Kosovo. The high level of FDI globally (2007) has not yet been repeated. In economic literature, it is considered as a “need” for the analysis of the factors that influenced the chronic level of FDI over a decade. Facing the theoretical criteria, the model Borensztein, et al. (1998), FE and RE technique through macro Panel Stata, judged in its entirety, it is concluded that the improvement of FDI requires money also to improve the specific factors in each country, as we conclude that the FDI determinants in CE, SEE and Kosovo are economic growth, domestic (local) investments, and government spending. Despite the importance of foreign capital flows, the paper emphasizes the importance of cooperation between domestic and foreign firms and the idea that the expectations of foreign investors and the benefits of the local economy are realized on the basis of improving their long-term profits.

Appendix

See Tables 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10.

Table 10. Regression variables

The sample of the countries included in the study

See Table 11.

Table 11. The sample of the countries included in the study (developed countries, developing countries, transition countries)

Definition of data

Empirical literature provides different opinions on the influence of independent variables (growth determinants) on economic growth. In our study, the determinants of growth are simultaneously determinants of FDI.

• 1. Economic growth—represented by real annual GDP growth per capita (GDPgpc).

• 2. Gross Domestic Product (GDP_it) at the beginning of 2003 (Y₀): Log_InitialGDP (Y₀) or the natural logarithm of real GDP (logY₀) at the beginning of each period, so that it is equal to the initial year of one and two year intervals. For the Initial_GDP per capita, the scale of 2003 is chosen, i.e. the year before the start of the sample. GDP is taken at purchase prices, or the amount of the added value of all resident producers in an economy, plus product taxes and minus subsidies. So, the initial GDP variable (Y₀) in the study is expected to reach the catching-up effect (N/N*).¹⁸

• 3. Foreign Direct Investment (FDI as % of GDP) are used as a standard in literature (Asiedu 2002; Choi 2009), in the model is conceptually analogous to parts of goods produced by foreign firms, (n * /N).¹⁹ The FDI net inflow was used for measurement of FDI. It includes the amount of capital, reinvestment of profits, long-term capital and short-term capital as shown in the balance of payments.

• 4. FDI in interaction with human capital (FDI/GDP_it * HC/Population_it): FDI in interaction with the variable of absorption capacity (HC) of the host country, i.e. gross registration in the secondary education as a percentage of the country’s population, expresses the term of interaction or the indirect impact of FDI on economic growth. This will also determine the level of education threshold.

• 5. Domestic Investment (DOM.INV./GDP_it): Is represented by formation of the fixed gross capital minus FDI. GCFCs are expenditures or additions to fixed assets of the economy minus net changes in the inventory level.

• 6. Variables in interaction with each other: [(FDI/GDP_it) × (DOM.INV./GDP_it)], in our case, is an interaction variable between FDI and domestic investment.

• 7. Secondary level of education (lag_HC/pop_it): The gross registration rate is the total census report, regardless of age, of the age group that officially corresponds to the level of education (secondary education—the high secondary school level) in relation to the population of the country.

• 8. Remittances (REM/GDP_it): are remittances to their homes by emigrants working abroad (through formal channels).

• 9. Total expenditure of government final consumption (Exp.Gov.Fin.Conc../GDP_it): The government expenditure ratio to GDP is taken by WDI as a measure of government involvement in the economy. The general government consumption includes all current government expenditure on purchase of goods and services (including employee compensation and government transfers). It also includes most of the expenses for national defense and security, but excludes government military expenses that is part of government capital formation.

The description of command (xtreg ²⁰ ), in Stata vs. 12, ²¹ to the use of methods (FE/RE).

The xtreg command is suitable to the regressions into models with panel data. In particular, xtreg is more suitable to assess the regression of the random effect model (RE), using the estimator (GLS estimator), the results given by the matrix-weighted average of the between and within. The selection of the model (RE) requires that the evaluation is done with the GLS method by specifying the small samples with the Swamy-Arora evaluation and by evaluating the variance component at the individual level as a choice for evaluation.²² The vce type of assessment (Robust) reports the type of standard error that derives from the conventional theory (asymptotic) and is a powerful evaluation when, within the group, there is a correlation (cluster) due to the lack of specificities, therefore the supportive methodology or the Jackknife method has been used.²³ Even FE requires regression to be evaluated with fixed-effects (within) and the GLS method. These two methods (fe, re) are considered more appropriate²⁴ for evaluating the panel data through Stata, in our case STATA vs. 12.

The Wald test: xttest3

Modified Wald test for heteroskedasticity GroupWise in the FE regression model

$$ \begin{aligned} & {\text{H}}0:{\text{sigma}}({\text{i}})^{ \wedge } 2 = {\text{sigma}}^{ \wedge } 2\,{\text{for}}\,{\text{all}}\,{\text{i}} \\ & {\text{chi}}2\left( {24} \right) = \quad 956.05^{{\prime }} ; \quad {\text{Prob}} > {\text{chi}}2 = \quad 0.0000 \\ \end{aligned} $$

Testing for the correlation in the series:

The series correlation tests are applied in macro panels in long time series (over 20–30 years). It is not a problem in micro panels (just a few years). In our case we have only 11 years, so there is no serial correlation. The serial correlation causes the standard error of the coefficients to be smaller, while R² are currently higher.

FE model is used in the data panel for GDPgpc growth rate, dependent:

The Hausman test, Table 18, suggests if: Prob ch2 = 0.0000, smaller than 0.05 (<0.05), is considered important and suggests the Fixed Effects methods to be used. Therefore, in Table 19 (Chapter Four), the Fixed Effects (FE) method is used.

The Random Effects model is used in the data panel for FDI—dependent:

The Hausman test, Table 20, suggests if: Prob 0.156 > 0.05, bigger than 0.05 (>0.05), is considered important and suggests the Random Effects methods to be used. Therefore, in Table 22, our Case Study has used the Random Effects method (RE). Results after the final certification of the tests are presented in chapter IV, Tables 16, descriptive statistics, Table 19 FE method, and Table 22 RE method.

Correlation Matrices

See Tables 12 and 13.

Table 12. cor. Log_GDPpc., FDI, HC, FDI * HC, Dom.Inve., FDI * Dom. Inves., Remittances (obs = 210)

Table 13. corlog_GDPgpc, logGDP_initial, FDI, HC, FDI * HC, Dom.Invest., FDI * Dom.Invest, Remitances (obs = 210)

Chart representation of variables used in a 24 country—panel (graphical-scatter plot, Stata. vs. 12). Source: WDI, 2015, Graphic 1–14, by the author.

See Charts 1, 2, 3, 4, 5 and 6 and Graph 7.

Chart 1.

Growth rate of GDP per capita in 24 European countries in 2004–2014

Chart 2.

LoG_GDPgpc/domestic investments/GDP

Chart 3.

FDI/GDP in US $ (2004–2014)

Chart 4.

Domestic investments/GDP

Chart 5.

Human capital/population

Chart 6.

Remittances/GDP

Graph 7.

General government consumption final expenditures/GDP

Interaktion between variables (Y-X):

See Charts 8 and 9 and Graph 10.

Chart 8.

GDP growth rate per capita and human capital

Chart 9.

FDI/GDP (dependent) and Log_GDPGpc (independent)

Graph 10.

Log_GDP pc and FDI in interaction with HC

Interaction between log, _GDPpc log and initial log_GDP (negative)

See Graph 11.

Graph 11.

Logging GDPgpc per capita and log_GDP_initial)\(lfitlog_GDPgpc logGDP_initial)