Access the full text.
Sign up today, get DeepDyve free for 14 days.
This paper investigates how typical macroeconomic indicators affect the economic growth of Western Balkans countries. A static panel empirical investigation for the period 2010 to 2019 has been conducted us- ing GDP growth rate as the dependent variable, while independent variables in focus include foreign direct investments, remittances, unemployment rate, population growth rate, and control of corruption. The most interesting finding is that a rising share of remittances positively affects economic growth. This might indi- cate that even when remittances are used for non-investment purposes, they might increase domestic pro- duction of consumption and intermediate goods. There is also evidence of a non-linear relationship between FDI and economic growth, which may be a consequence of undeveloped capacities to use the positive side of FDI. To foster economic growth policy-makers should focus on reforms that target sectors that show sharp declines in FDI and remittances inflows, including also a need for better control of corruption in the region. Keywords: Economic growth, panel data, FDI, remittances, unemployment, corruption JEL Classification: E60, O10 1. Introduction The Western Balkans countries (Northern Besime Ziberi, PhD Macedonia, Kosovo, Albania, Bosnia and Herzegovina, Assistant Professor Montenegro, and Serbia) have transitional economic Faculty of Economics, AAB College, Kosovo systems with many problems and challenges. During E-mail: firstname.lastname@example.org the 1990s, the Western Balkan region suffered from ORCID number: 0000-0003-2891-3832 severe conflicts, which ended after intervention by the United Nations and NATO forces. In the early and Merita Zulfiu Alili, PhD (corresponding author) mid-2000s, the prospect of EU accession and the Associate Professor global boom facilitated rapid economic recovery in Faculty of Contemporary Social Sciences, the Western Balkans and boosted economic and insti- South East European University tutional reforms (Dabrowski and Myachenkova 2018). E-mail: email@example.com The region is known for its instability and recent histo- Address: Ilindenska no. 335, 1200 Tetovo, ry of wars and civil conflicts. During the transition pe - North Macedonia riod, the WBC embarked on a process of far-reaching ORCID number: 0000-0002-6367-6193 reforms aimed at transforming the economic structure Copyright © 2021 by the School of Economics and Business Sarajevo 68 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS from a socialist to a market-based one (Shimbov, Western European levels, including in terms of wages Alguacil, and Suárez 2019). (Amoh, Awuah-Werekoh, and Ofori-Boateng 2020). Six Western Balkans countries are known as coun- DIG-Labour models feature segmented labour mar- tries with young populations (but aging once due the kets, efficiency wages and open unemployment, and high emigration rate of youth in the region (Efendic an informal non-agricultural sector. Thus investment 2016; Michael and Isilda 2020)) and rising numbers of in education affects labour productivity with a long university graduates, on the other hand, insufficient lag, it takes 15+ years before net national income, the demands for profiles from the labour market and thus private capital stock, and real wages for the poor, and high unemployment rates of persons with a university formal sector employment to surpass their counter- degree. According to the World Bank Group (2017), parts in a program that invests mainly in infrastructure over 67 percent of respondents in the Western Balkans (Buffie et al. 2020). Thus the 2025 EU Accession target cited unemployment as their top concern - double the for the Western Balkans represents a highly ambitious share in the European Union, where it appears that a best-case scenario, which could serve as a power- stronger labour-market response and more gener- ful incentive for countries in the region to speed up ous unemployment benefits have counteracted the their reform agendas (Grieveson, Grübler, and Holzner effects of slowing growth. Remittances and Foreign 2018). Direct Investment are considered as driving forces for This study examines the factors that impact the sustainable economic growth and development in the economic growth of Western Balkans countries. WB region. Population growth, remittances, FDI and control of The resulting inflows of capital and technology corruption are factors expected to boost economic affected the demand for labour and accordingly em- growth. In the viewpoint of the determinants that ployment, productivity and wage levels (Zulfiu Alili negatively impact the economic growth in WB, we and Adnett 2018). Regarding the attraction of foreign take into account unemployment as a main chal- direct investments, a series of legal and economic re- lenge that remains unsolved and a crucial factor for forms were applied to improve the business climate. development. However, despite the efforts of governments to im- The rest of the paper is organized as follows: In prove the conditions for doing business and attract Section 2, a review of relevant previous empirical re- FDI, decreasing the unemployment rate thus increas- search is presented. This discussion informs the speci- ing the level of employment, corruption and unem- fication of the model of determinants of economic ployment remain significant factors that hamper growth presented in Section 3. Section 3 also explains the economic stability of these countries. The main the data used and presents the descriptive statistics of concern of any macroeconomic policy is to lower the the variables of interest. The empirical approach taken unemployment rate and to have economic growth, so to estimate the determinants of the economic growth these two indicators play a vital role in the economic model is explained in Section 4. The following section performance of any country (Ziberi and Avdiu 2019). presents and interprets the results, with both Fixed The Western Balkan countries are steadily stated and Random effects models providing a robustness in the viewpoint of economic growth with a long check. Section 6 is devoted to concluding remarks and process of transition and concretely from the year consideration of the policy implications of the main that the countries gained the status of candidate for findings. the European Union are still far away from negotia- tions to join the European Union. The transition from stagnation to growth is one way to think about the 2. Literature Review historical pattern: what can trigger such shift and what is the underlying engine of growth that could Two main factors of great importance for eco- explain not just the shift from a low level of develop- nomic growth in WB countries are Foreign Direct ment into persistent growth in per-capita income, but Investment and Remittances (Loxha 2019). For exam- also the demographic transition that has accompa- ple, the effects of foreign direct investment inflows nied such a growth pattern in many other parts of the in the industrial, construction, and services sectors world in more recent decades? (Isaac and Pei 2020). on economic growth are analysed from Miteski and The increasing trend of youth unemployment in the Stefanova (2017) in a panel of sixteen Central, Eastern, Western Balkans will hamper economic growth. In the and South-eastern European CESEE countries using long run, this will result in the loss of an important data for the period 1998-2013. Their findings indicate share of the human capital of these countries, which that total FDI contributes positively to the growth in might affect their prospects for convergence with the countries into the analysis. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 69 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Biørn and Han (2017) use GMM estimation of au- spillover benefits to purely domestic enterprises – toregressive equations in error-ridden variables with which represent the broader advantages of FDI for error memory considered in exploring the impact of economic development – are found to be few and foreign direct investment (FDI) on GDP using country far between, and indeed often appear to have been panel data. Their study comes to the conclusion that negative rather than positive. In the absence of posi- results from Monte Carlo simulations suggest that ag- tive spillovers – and a fortiori in the presence of nega- gregate FDI has a positive, but insignificant effect on tive ones – the restructuring and development of the aggregate GDP based on the full country panel; (2) domestic enterprise sector may be inhibited, thereby for the developing Asian countries, FDI significantly reinforcing fears that an “enclave” economy might be improves GDP growth; and (3) manufacturing FDI emerging where a technologically advanced FDI sec- impacts both manufacturing and service GDP growth tor pulls ahead but has little if any positive impact on positively. the rest of the economy. Of a great contribution is the study from Dritsakis The effect of Foreign Direct Investment (FDI) and Stamatiou (2018) which examines the causal- on economic growth is analysed by Keisuke and ity relationships between foreign direct investments Sovannroeun (2010) using the data of 132 countries (FDI), exports, unemployment, and economic growth for the period from 1995 to 2008, considering the in the fifteen old EU members using panel data for the role of corruption in each country as an absorptive period 1970 to 2015. The results indicate that there factor. The estimation results indicate that, although are three bidirectional causalities among economic FDI alone does not promote economic growth, it has growth and exports, exports and FDI, and exports a significant effect on economic growth if the interac - and unemployment, and three unidirectional cau- tion term between FDI and corruption is examined. salities running from FDI to economic growth, FDI Tamar and Luca (2020) results suggest FDI benefits do to unemployment, and from economic growth to not accrue mechanically and evenly across countries. unemployment. They find an inverted-U-shaped relationship between Brenner (2014) studies the effects of foreign direct countries’ income levels and the size of FDI impact on investment (FDI) on national economic growth taking growth. The determinants and the policies that affect into consideration the effects on productivity growth, economic growth in the Western Balkan are also stud- capital and labour inputs, and innovation. This study ied by Fetai, Mustafi, and Fetai (2017) over the period analyses separately less and more developed coun- 1994 to 2015. The study used OLS, fixed and random tries for the periods 1974-1991 and 1992-2009. Results effects model, and instrumental variables (IV). The suggest that the effects of FDI on production ca- results indicate that foreign direct investments, gross pacities are found to be positive for more developed savings, and domestic credit to the private sector have countries in the first time period, while effects on a positive effect on per capita growth in the case of innovation activities are found to be positive for the Western Balkan Countries. medium developed countries in the latter time period. Sabir, Rafique, and Abbas (2019) analyze the Effects in less developed countries are rather negative. impact of institutional quality on Foreign Direct The empirical analysis of Mehic, Silajdzic, and Babic- Investment (FDI) inflows using panel data for low, Hodovic (2014) on the impact of FDI on economic lower-middle, upper-middle, and high-income coun- growth in the transition countries of southeast Europe tries for the sample period of 1996-2016 using the indicates a positive and statistically significant effect Generalized Method of Moments (GMM). They find of FDI on economic growth. This effect is robust when that the magnitude of the coefficients of control of including data on domestic investments. corruption, government effectiveness, political stabil- A study from Djankov and Hoekman (2000) con- ity, regulatory quality, rule of law, and voice and ac- cludes that the wider benefits of FDI are contingent countability for FDI inflows are greater in developed on the domestic economic and institutional environ- countries compared to emerging countries. The re- ment – there is nothing automatic about them. Also, lationship between institutional improvement and the study concludes that FDI in the transition econo- economic performance analysed by Efendic and Pugh mies since 1990 has largely flowed to just a few central (2015) find that GDP per capita in transition econo - European countries, which are also the leading can- mies is determined by the history of the institutional didates for EU membership. These have indeed ben- reform and conditional on this history, GDP per capita efited from significant FDI financing of the balance of adjusts to institutional changes. payments, and enterprises with foreign investment, The threshold level of corruption separating the not surprisingly, have had high rates of growth of out- negative and positive effects of FDI on economic put, productivity, and exports. However, the expected growth is approximately in the 10th percentile from 70 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS the least corrupt countries. The existence of a corrup- Remittances and their impact on economic tion threshold implies a counter-intuitive proposi- growth is analysed by Meyer and Shera (2017) using tion: that FDI inhibits economic growth in countries panel data of six high remittance-receiving countries, where corruption is below a corruption threshold Albania, Bulgaria, Macedonia, Moldova, Romania, and and promotes economic growth in countries where Bosnia Herzegovina during the period 1999 to 2013. corruption is above the threshold. Amoh, Awuah- The study comes with conclusions that remittances Werekoh, and Ofori-Boateng (2020) examine the ef- have a positive impact on growth and that this impact fect of corruption on economic growth in Ghana. To increases at higher levels of remittances relative to establish the strength of relationships among corrupt- GDP. ing activities they used structural equation modeling Astrov et al. (2020) have used the econometric on selected data from the World Economic Forum vector autoregression (VAR) estimation, which takes executive opinion survey on corrupting activities and account of endogeneity and dynamic effects. Results data on economic growth measures from the World indicate that the Phillips curve mechanism, which Development Indicators data from 2008 to 2017. transmits declining unemployment to wage dynam- The study concludes that corrupting activities, inde- ics, is not yet mature in less-developed countries, such pendently and mutually impede Ghana’s economic as the Western Balkans, Moldova and Ukraine. growth. Using a set of determinants of economic growth Sbaouelgi (2019) examines the impact of corrup- suggested by literature in a group of transition coun- tion on investments and economic growth in coun- tries from 1997 to 2017, Szymańska (2021) panel data tries from the MENA region over the period 1990 to results indicate that the effect of demographic factors 2017 using the dynamic panel data model. Results are not significant, however, the role of investment suggest that political institutions have a positive ef- has been emphasized. fect on investment and growth. Moreover, corruption Our literature review reveals the number of deter- harms economic growth through its impact on invest- minants to be important for the economic growth of ment. The way how corruption impacts the economic WB countries, but this applies to standard growth fac- growth of Western Balkans is also analysed through tors, in particular to FDI, corruption, and remittances, indirect channels taking into consideration the im- which have specific importance in the region. Thus, we pact of corruption in Small and Medium Enterprises s focus our empirical investigation on these influences. (SMEs) as they are considered a key factor of economic growth. SMEs impact the economic growth through job creation, innovation, and creativity, production, 3. Model Specification, data and knowhow, etc. descriptive statistics The study of Rehman et al. (2019) analyses the barriers that hinder labour productivity (LP) of SMEs Hundreds of previous empirical studies have at- in WBCs. The study uses cross-sectional data and con- tempted to identify the determinants of economic ducting a factor analysis and ordinary least square growth. However, as emphasized in these studies, (OLS) regression. The study comes with directions that growth theories are open-ended. Hence the diversity the variability for different countries’ access to finance, of theoretical views makes it difficult the identifica- tax rates, tax administration, corruption, inadequately tion of the most effective growth-promoting policies educated labor force, competition in the informal (Moral-Benito 2009). Based on the previous literature sector, and political instability appears to be some of and data availability for the countries included in the the main obstacles that are negatively affecting LP of sample the following -growth model is specified: SMEs in WBC. Another study (Durguti et al. 2020) used panel 𝛾𝛾 �� �� 𝑋𝑋 �� (1) �� � � �� �� data for the Western Balkan countries for the period 2001 to 2017 and pooled OLS methods, fixed and ran- Where: dom effects model, to test economic growth rate as i indexes countries; t the time period; γ repre- it the dependent variable, and explanatory variables sents the dependent variable GDP growth (annual %); such as working remittances to GDP, exports to GDP, x represent a set of growth determinants, including it imports to GDP, foreign direct investment to GDP and those suggested by previous empirical studies, hence inflation rate as independent variables. The study con- population growth rate (POP), remittances (REM), for- cludes that foreign direct investments and inflation do eign direct investments (FDI), unemployment (UNR) not impact the economic growth in Western Balkans and control of corruption (CORR), and e is the error it countries. term. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 71 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS To capture the effect of demographic changes affecting negatively the economic growth. Whereas population growth (POP) is included in the model. control of corruption (CORR) and government effec - Population growth rate, if considered as independent tiveness are expected to positively impact economic from GDP growth, is expected to lead to higher eco- growth. All data are obtained from the World Bank nomic growth rates. In this case population growth Indicators (The World Bank 2020). Analysing the rank is stimulating technological innovations that would of control of corruption for the period 2000 to 2019 allow food production to keep up with the growing for the countries under investigation the data show population (Wesley and Peterson, 2017). However, that Albania went down by 9.3 in rank for control of Malthus (1993) argues that population growth harms corruption, Montenegro went down by 1.9 in rank and economic growth since the population tends to grow Serbia went down by 29.4 in rank. Whereas Bosnia and faster than food supplies and other goods needed for Herzegovina went up by 3.2 in rank and Kosovo went human life. High population growth in low-income up by 37.3 in rank for control of corruption. countries may slow the economic growth (Wesley and Among the selected countries, Albania has the Peterson 2017). highest year-on-year average growth rate at 3.96%, Population growth affects the age structure and whereas North Macedonia has the lowest year-on- the workforce of a country’s population, international year average growth rate at -53.99%. Appendix 1 migration, economic inequality, etc. As shown in Table presents more details on variable definitions and data 1, the mean of population growth (annual %) is nega- sources. tive in Albania, Serbia, and Bosnia and Herzegovina, hence the population size is getting smaller for the period 2000 to 2019, mainly as a result of high emigra- 4. Empirical methodology tion rates in these countries. Data reveal low popula- tion growth of 0.1% in other SEEC included in the The empirical methodology used to select the sample. The size of the population in this region is correct estimation technique is discussed in this sec- suspicious due to high emigration rates. tion. We compared the Ordinary Least Squares (OLS), Remittances (REM) are expected to contribute to Fixed Effects (FE), and Random Effects (RE) models for economic growth through their positive impact on panel data using the standard approach. Both the FE consumption, savings, and investments. These coun- and the RE models appear to be appropriate estima- tries have experienced a significant increase in remit - tors, as discussed below. The preferred model is FE, tances and in most of the countries they represent which is more appropriate for small samples and can more than 10 percent of GDP on average exceeding be estimated for unbalanced panels, as is the case the flows of FDI. Kosovo as one of the countries with here, whereas RE is estimated as a robustness check. the weakest economic development in the region is Diagnostic tests were performed on the chosen model more dependent on remittances (mean of 16.6 % for to check if it is misspecified. The slope homogeneity, the period 2000 to 2019) (Table 1). cross-sectional dependence in the error, groupwise The role of FDI is generally considered to impact heteroscedasticity, serial correlation in the errors, and positively economic growth. Nevertheless, as dis- normality of the errors have all been evaluated as cussed above, the empirical evidence is mixed. The common concerns with panel estimations. These tests data in Table 1 show that FDI inflows during the past are explained below and summarised in Appendix 3. decade in Western Balkan countries have been stag- To account for the heteroscedasticity problem, robust nant on average. Among the countries included in standard errors are used. the sample, Montenegro has been more successful in To explore the determinants of economic growth, attracting FDI for the period under investigation with we start with the classic OLS estimator testing wheth- a mean of 15.5% of GDP. FDI squared is added to test er fixed effects model is more appropriate than OLS. for any non-linearity in the relationship with the GDP To choose between these two models we test whether growth. the deviations (FEs) are jointly significant. Results pre - Unemployment (UNR) is a macroeconomic indica- sented in Appendix 2 indicate that a one-way FEM is tor having high values (the mean value in the sample supported. The use of the OLS model would produce is 25%, Table 1) presents the inability of these econo- unreliable estimates since heterogeneity across coun- mies to make full use of labour resources. Kosovo ex- tries exists. Choosing between one-way RE versus OLS, perienced the highest unemployment rate (mean of Breusch and Pagan’s LM statistic is performed. In this 40.1 %) among the other Western Balkans countries. case H0: s2u=0 with test restriction that ui=α (whether Corruption leads to high investments costs and random effects are equal to OLS intercept). The result low profit, hence discouraging investments and of the test indicates the estimation of RE compared 72 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Table1. Summary statistics of the variables of interest by country and total, 2000-2019 N mean sd min max Albania GDP growth 20 4.3 2.1 1.0 8.3 Population growth (POP) 20 -0.4 0.2 -0.9 -0.1 Remittances (REM) 20 13.4 2.8 9.6 17.8 Foreign Direct Investment (FDI) 20 6.9 2.5 3.1 11.2 Unemployment rate (UNR) 20 15.2 1.7 12.3 17.5 Corruption rank (CORR) 19 30.8 6.3 22.9 42.3 Bosnia and Herzegovina GDP growth 20 3.4 2.7 -3.0 8.8 Population growth (POP) 20 -0.6 0.7 -1.7 0.2 Remittances (REM) 20 15.3 5.7 10.5 28.7 Foreign Direct Investment (FDI) 20 3.8 2.6 0.8 11.7 Unemployment rate (UNR) 20 26.2 3.6 18.4 31.1 Corruption rank (CORR) 19 42.4 6.0 30.3 50.7 Kosovo GDP growth 19 5.2 5.6 -0.7 27.0 Population growth (POP) 20 0.1 1.0 -3.6 0.9 Remittances (REM) 16 16.6 2.0 14.6 20.1 Foreign Direct Investment (FDI) 16 5.9 3.1 1.5 13.0 Unemployment rate (UNR) 19 40.1 9.6 25.6 57.0 Corruption rank (CORR) 19 38.5 11.8 26.7 69.7 North Macedonia GDP growth 20 2.9 2.3 -3.1 6.5 Population growth (POP) 20 0.1 0.1 0.0 0.5 Remittances (REM) 20 3.3 0.7 2.0 4.1 Foreign Direct Investment (FDI) 20 4.6 2.6 0.5 12.7 Unemployment rate (UNR) 20 30.2 5.6 17.8 37.2 Corruption rank (CORR) 19 45.6 9.7 26.8 57.3 Montenegro GDP growth 20 3.1 3.2 -5.8 8.6 Population growth (POP) 20 0.1 0.1 -0.0 0.4 Remittances (REM) 13 10.1 2.3 5.3 12.4 Foreign Direct Investment (FDI) 13 15.5 8.7 5.2 37.3 Unemployment rate (UNR) 20 22.1 6.0 14.9 30.6 Corruption rank (CORR) 19 50.9 5.7 38.0 58.2 Serbia GDP growth 20 3.6 3.0 -2.7 9.0 Population growth (POP) 20 -0.4 0.2 -0.8 -0.1 Remittances (REM) 13 8.4 0.9 6.8 10.3 Foreign Direct Investment (FDI) 13 6.5 2.3 2.9 10.2 Unemployment rate (UNR) 20 17.1 3.8 12.6 24.0 Corruption rank (CORR) 19 41.6 10.3 7.6 51.4 Total GDP growth 119 3.7 3.3 -5.8 27.0 Population growth (POP) 120 -0.2 0.6 -3.6 -0.9 Remittances (REM) 102 11.2 5.6 2.0 28.7 Foreign Direct Investment (FDI) 102 6.7 5.2 0.5 37.3 Unemployment rate (UNR) 119 25.0 10.0 12.3 57.0 Corruption rank (CORR) 114 41.6 10.5 7.6 69.7 Source: (The World Bank, 2020), Author’s calculations SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 73 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS to OLS. Besides, the OLS does not account for the of dealing with the endogeneity problem (Baltagi group-specific effects. The last models to be tested are 2005, p. 113). However, given the limitations of the a one-way FEM vs. one-way REM. To decide between available data finding a good instrumental variable these two models Hausman’s chi-square statistic can for an endogenous explanatory variable in this case be used. A large value from this test will support FE is very difficult. Assuming that the correlation of FDI over RE. Here, we test the null hypothesis that the with the error term is fixed over time, the problem of unobserved individual-specific random effects and re - potential endogeneity is alleviated by controlling for gressors are uncorrelated. The test suggests that ran- country fixed effects (Figini and Görg 2011; Zulfiu Alili dom effects can be estimated. According to this test, 2021). For this reason, a Fixed Effects model has been the difference between the estimators is small to its employed. variance, i.e. there is no systematic difference between Diagnostic testing should be used to provide the the estimators, which suggests that the RE estimator best possible statistical specification. Diagnostics for is not biased and it is consistent. Therefore, in addition panel analysis, particularly for unbalanced and small to the FE model, we estimate RE model. panels like the one we have here are not well devel- However, we cannot totally exclude the possibility oped. Diagnostic testing, on the other hand, can still that the chosen determinants of the economic growth provide some valuable indicators; hence diagnostics model might be considered to be endogenous or that have been used on the chosen FE model (Appendix 3). we omit some important variables (e.g. institutions). To test for slope homogeneity a random coeffi- FDI can lead in capital accumulation and increase em- cients model by GLS is estimated. This model provides ployment opportunities but also would contribute to the test of parameter constancy: H : β = β. where the o i increasing the economic growth of the host economy. minimum requirement is T>K (where K is the number However, there is no clarity on the direction of cau- of parameters to be estimated). As expected the test sality between FDI and economic growth in the lit- indicates no slope homogeneity, however as sug- erature. There is no consensus on whether economic gested by Pesaran, Shin, and Smith (1996) this is not growth causes FDI or FDI causes economic growth a problem since we are interested in investigating the (Sabharwal 2019). Previous studies find institutions to determinants of growth in the countries in the sam- be an important factor for economic growth and de- ple and we are not interested in making inferences velopment, they should be included in econometric to a wider population or on the individual country models to account for transaction costs (Efendic and coefficients. Panel data models are likely to exhibit Pugh 2007; Efendic and Pugh 2015). cross-sectional dependence in the errors due to spa- However, institution measures (mainly used in the tial dependence, omitted unobserved common com- literature EBRD indices) for countries in the sample ponents, or idiosyncratic pair-wise dependence of the are missing for some years and especially for Kosovo error term (Pesaran 2004). There are two alternative which reduced significantly the number of observa- approaches commonly used to test for cross-sectional tions making it not reliable to include in the model. dependence in panels, namely the Lagrange multi- According to Wooldridge (2002), endogeneity occurs plier (LM) test developed by Breusch and Pagan valid when variables are omitted because data is unavaila- when T>N and Pesaran test valid when T<N. However, ble; measurement error occurs when a variable’s (par- both approaches are problematic with unbalanced tial) effect is measured with only an imperfect meas- panel data. We have used the user-written Stata pro- ure of it; and simultaneity occurs when at least one of gram xtcsd which allows two options in our case: the explanatory variables is determined simultane- the Pesaran and Friedman version of cross-sectional ously with the dependent variable. The GMM estima- independence in the errors. The tests indicate to re- tor presented by Arellano and Bond (1991), Arellano ject the Ho of zero cross-sectional dependence in the and Bover (1995), Blundell and Bond (1998), Roodman errors. Testing for homoscedasticity, i.e. constant vari- (2009), is one technique to cope with the endogeneity ance over cross-section units, the modified Wald test problem. This model is applicable for situations with for GroupWise heteroskedasticity was used. The null “small T, large N panels, meaning few periods and hypothesis of homoscedasticity was rejected suggest- many individuals; independent variables that are not ing the estimation of the model with robust standard strictly exogenous, meaning correlated with past and errors. To obtain consistent estimates in addition to possibly current realizations of the error; fixed effects; robust standard errors the Driscoll-Kraay standard er- and heteroskedasticity and autocorrelation within rors are used, which are robust to heteroscedasticity individuals,” as Roodman argues (p.86), which is not Hoechle 2007). Even though autocorrelation in the er- the case in the current study since N is only six groups. rors is common in panel data analysis, the Wooldridge Instrumental variable (IV) methods are another way test for serial correlation (xtserial) suggests that 74 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS there is no first-order autocorrelation in the errors. Overall, the results are robust and do not differ signifi- Normality of the errors is tested using the Stata user- cantly in terms of the magnitude and significance of written command pantest2 and by plotting the resid- the coefficients. Considering the above discussion of uals (Appendix 3). Both the test and the plot indicate the heteroscedasticity problem the preferred model is that residuals are normally distributed. The empirical FE Driscoll-Kraay standard errors, column 2. According results are discussed in the next section. to the findings, remittances have a favourable impact on economic growth, which is consistent with previ- ous research. A 1 percent rise in remittances (percent of GDP) will result in a 0.3 percent boost in GDP growth 5. Estimation Results for the WB nations in the sample. FDI coefficient is The estimation results of the FE and RE models are positive and significant suggesting that if the share of presented in this section. The results of FE regressions FDI to GDP increases by 1 percent economic growth utilizing robust standard errors and Driscoll-Kraay will increase by 0.6 percent. However, the square FDI standard errors are shown in columns (1) and (2). The is negative and significant, implying that FDI and eco - results of the RE estimation are shown in column (3) nomic growth have a concave connection. As a result, The total number of observations in the model has FDI inflows boost economic growth (but at a slower been decreased to 99 due to missing observations pace). These findings could indicate that the transition for various variables in specific years and countries. countries in the sample are unable to catch up with Table 2. Estimation results of FE and RE models. FE Robust standard FE Drisc/Kraay errors Std.Err. RE (1) (2) (3) VARIABLES GDP growth GDP growth GDP growth Population growth (POP) -0.783 -0.783 0.125 (0.623) (0.418) (0.501) Remittances (REM) 0.300*** 0.300*** 0.072 (0.089) (0.067) (0.046) Foreign Direct Investment (FDI) 0.571*** 0.571** 0.421*** (0.131) (0.204) (0.116) SQFDI -0.017*** -0.017** -0.014*** (0.004) (0.005) (0.004) Unemployment rate (UNR) -0.027 -0.027 0.028 (0.050) (0.033) (0.030) Corruption rank (CORR) -0.080** -0.080** -0.058** (0.034) (0.024) (0.025) Constant 1.263 1.263 2.564 (2.057) (1.233) (1.599) Observations 99 99 99 R-squared 0.341 Number of id 6 6 Number of groups 6 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: Author’s calculations SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 75 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS the trend of technological inventiveness and exper- not possible in our case. A bigger sample would be an tise, causing domestic enterprises to be constrained advantage in the future, taking into account also the and unable to achieve market share competitiveness. effect of the Pandemic. Melnyk, Kubatko, and Pysarenko (2014) find that addi- tional FDI inflows in firms may push out of the market other firms without FDI, this is referred to as a “market 6. Conclusions and Recommendations stealing” effect when domestic firms are not so pro - ductive compared to the foreign ones. Western Balkan This study analyses the impact of macroeconomic countries in recent years have decreasing trends of FDI indicators on economic growth in Western Balkans in general and also in regard to technology and know- countries. The factors influencing economic growth how which are considered as main channels through are divided into two groups: factors expected to posi- which FDI is expected to positively impact growth. tively impact economic growth such as population In this regard, results might suggest that FDI invest- growth rate, remittances, Foreign Direct Investments, ments should be oriented to productive sectors thus control of corruption and one of the main factors that increasing the rate of employment and productivity hamper the economic growth such as unemployment. in the labor market. In addition, the negative sign of Using data from World Bank Indicators for the period square FDI might be a consequence of the relatively 2000 to2019 we estimated FE and RE models. Results high level of corruption in many transition countries . suggest that remittances and FDI positively impact In addition, the unemployment rate is used as an inde- the economic growth of the Western Balkan countries pendent variable, which is also an outcome variable. whereas unemployment and control of corruption Including the unemployment rate raises the ques- have a negative impact on the economic growth. This tion: does reducing the unemployment rate increases study suggests that remittances are an important fac- a real GDP growth or does real GDP growth causes tor for economic growth and they should be more a reduction in unemployment or do both affect one oriented to investment instead of personal consump- another. However, the unemployment variable is not tion. The study also suggests for policymakers to build significant and excluding it from the model does not business zones and also to improve the rule of law in impact the results. A low level of control of corruption order to attract the FDI. Economic growth increases may humper the positive effect that FDI may have on with FDI inflow but this effect diminishes with further human development and economic growth. Evidence increases in FDI. A study from (Jushi et al. 2021) con- indicates that corruption is likely to adversely affect cludes that in the case of Western Balkans countries, long-term economic growth through its impact on remittances; population growth, and labour force investment, taxation, public expenditures, and human participation were not key determinants of growth. development. Corruption is also likely to undermine The impact of FDI and trade was negative yet highly the regulatory environment and the efficiency of state significant. As far as GDP growth, it exhibited strong institutions as rent-seeking distorts incentives and persistence from one year to the next. Trying to keep decision-making processes (Chêne 2014). The vari- stable remittances, FDI, and trade over time can make able of control of corruption is negative and signifi- these factors have a positive significant contribution cant indicating that a one percent increase of control to growth. In addition, these countries are steady- of corruption will negatively impact the economic state in the road of the process of automation and growth by 0.08 percent. Ahmad, Ullah and Arfeen digitalization whereas domestic firms usually remain (2012) show that a decrease in corruption raises the out of the competitive market in front of foreign firms. economic growth rate in an inverted U-shaped way. Western Balkans countries are late with the har- They find corruption to be growth-enhancing at low monization of legal acts and even more in delay levels of incidence and growth-reducing at high levels with the implementation of innovations. Having in of incidence. mind the nature of the process of digitalization, it It should be emphasized that the relatively small should go in line with the market requests otherwise size of the sample in this study affects the number of it threatens to remain at all times outdated and too possible variables to be included in the growth analy- expensive. Furthermore, the digitalization of services sis, due to the data availability. Therefore, the role of in the Western Balkans region is mostly driven by the institutions in explaining economic growth could financial aspect and most of the electronic services (e- not be included. As indicated above a dynamic panel services) provided are intended for the business sec- model could be an advantage to account for endoge- tor (Jashari 2020). neity which arises due to omitted variables that not be In sum, the previous empirical evidence supports included due to data unavailability, however, this was the hypothesis that an increased FDI inward stock as 76 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS References a share of GDP increases economic growth; however, this relationship may be a complex one in transition economies. This study highlights for policymakers to Ahmad, E., Ullah, M. A. and Arfeen, M. I. 2012. Does corrup- make efforts on reduction of corruption as mainly the tion affect economic growth?. Latin American Journal of corruption affects also the trend of FDI and in general Economics 49(2): 277-305. hampers the economic growth of WB countries. Amoh, J. K., Awuah-Werekoh, K. and Ofori-Boateng, K. 2020. As a result of the pandemic crisis, all countries in Do corrupting activities hamper economic growth?. the region have moved to support enterprises to re- Journal of Financial Crime. https://doi.org/10.1108/ tain jobs and protect the unemployed and as the re- JFC-11-2019-0150. covery phase starts, authorities there should not lose Arellano, M. and Bond, S. 1991. Some tests of specification sight of addressing longer-term challenges (World for panel data: Monte Carlo evidence and an application Bank Group 2020). The COVID-19 pandemic and the to employment equations. Review of Economic Studies ensuing economic crisis have exposed how informal 58: 277–297. firms and workers are more vulnerable to shocks, Arellano, M. and Bover , O. 1995. Another look at the instru- and the difficulty of providing support to businesses mental variable estimation of error-components mod- operating in the shadow (informal) economy. All six els. Journal of Econometrics 68: 29–51. Western Balkan economies have large informal sec- Astrov, V., Leitner S., Mara, I., Podkaminer, L., and Weinberger, tors, which account for an estimated 25–35 percent V. H. 2020. Wage developments in the Western Balkans, of both the GDP and the number of people employed Moldova and Ukraine, Vienna: The Vienna Institute for — much higher than in the EU member states. For the International Economic Studies (wiiw) 444. Western Balkan economies, high informality has held Baltagi, B. 2005. Econometric analysis of panel data. back growth, created unfair competition, and delayed Chichester: John Wiley and Sons. convergence with the rest of Europe (World Bank Bartlett, W. 2008. Europe’s troubled region: economic de- Group 2021). velopment, institutional reform and social welfare in the Government reforms need to be oriented toward Western Balkans. London: Routledge. the reduction of high unemployment rates through Biørn, E. and Han, X. 2017. Revisiting the FDI impact on GDP creating jobs and promote youth employment growth in errors-in-variables models: a panel data GMM through creativity and innovation as key drivers of analysis allowing for error memory. Empirical Economics: economic growth in these countries. The study goes A Journal of the Institute for Advanced Studies. further with recommendations for policy-making in Blundell, R. and Bond, S. 1998. Initial conditions and mo- the Western Balkan countries to strengthen law en- ment restrictions in dynamic panel data models. Journal forcement, reduce corruption, apply policies to raise of Econometrics 87:115–143. public awareness to ask for fiscal coupons when mak - Brenner, T. 2014. The impact of foreign direct investment ing purchases, promote democratic and economic on economic growth: an empirical analysis of different sustainability to attract foreign investors in export- effects in less and more developed countries. Marburg: oriented industries, regulate contract issues in eco- Philipps University Marburg, Department of Geography. nomic zones and facilitate administrative procedures Buffie, E. F., Luis-Felipe, Z., Christopher, A., Lacina, B., Dawit, for foreign investors and to use private remittances for T., and Kangni, K. 2020. Debt, investment, and growth business purposes in this way to generate new jobs. in developing countries with segmented labor markets. Washington, DC: International Monetary Fund / IMF Working Paper. Chêne, M. 2014. The Impact of corruption on growth and in- equality. Transparency International. Dabrowski, M. and Myachenkova, Y. 2018. The Western Balkans on the road to the European Union. Bruegel Endnotes Policy Contribution 4. Djankov, S. and Hoekman, B. 2000. Foreign investment and 1 We included an interaction term between FDI and corruption, however, in this case the model was not productivity growth in Czech enterprises. Economic well specified and the coefficient of interaction term Survey of Europe 14(1): 49-64. was not significant. Dritsakis, N. and Stamatiou, P. 2018. Causal nexus between FDI, exports, unemployment and economic growth for the old European Union members: evidence from panel data. International Journal of Economic Sciences: IJoES. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 77 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Durguti, E. A., Gashi, E., Kunoviku, F. D. and Mehmeti, M. Meyer, D. and Shera, A. 2017. The impact of remittances on 2020. DAF determinants affecting economic growth – economic growth: an econometric model. Economia. evidence for Western Balkans countries. International Revista da ANPEC 18(2): 147-155. Journal of Finance and Banking Studies: JJFBS, 9(1). Michael, L. and Isilda, M. 2020. Massive migration and its ef- Efendic, A. and Pugh, G. 2015. Institutional effects on eco - fect on human capital and growth: the case of Western nomic performance in post-socialist transition: a dynam- Balkan and Central and Eastern European countries. ic panel analysis. Acta Oeconomica 65(4): 503-523. Vienna: The Vienna Institute for International Economic Efendic, A. 2016. Emigration intentions in a post-conflict Studies (wiiw). environment: evidence from Bosnia and Herzegovina. Miteski, M. and Stefanova, D. J. 2017. The impact of secto- Post-Communist Economies 28(3): 335-352. rial FDI on economic growth in Central, Eastern and Efendic, A. and Pugh, G. 2007. Institutions and economic Southeastern Europe. Skopje: National Bank of the performance: an overview of empirical research with Republic of Macedonia. the main focus on transition economies. SEE Journal of Moral-Benito, E. 2009. Determinants of economic growth. A Economics and Business 2(1):. 25-30. Bayesian Panel data approach. Policy Research Working Fetai, B. T., Mustafi, B. F. and Fetai, A. B. 2017. An empirical Paper 4830: The World Bank. analysis of the determinants of economic growth in the Pesaran, M. H. 2004. General diagnostic tests for cross Western Balkans. Scientific Annals of Economics and section dependence in panels. Faculty of Economics, Business 245-254. University of Cambridge, Cambridge Working Papers in Figini, P. and Görg, H. 2011. Does foreign direct investment Economics 0435. affect wage inequality? An empirical investigation. The Pesaran, M. H., Shin, Y. and Smith, R. J. 1996. Testing for World Economy 34(9): 1455-1475. the ‘existence of a long-run relationship’: University of Grieveson, R., Grübler, J., G. and Holzner, M. 2018. Western Cambridge. Balkans EU accession: is the 2025 target date realistic?, Rehman, N. U., Çela, A., Morina, F. and Gura, K. S. 2019. Vienna: The Vienna Institute for International Economic Barriers to growth of SMEs in Western Balkan countries. Studies (wiiw). Journal of Management Development 38(1). Hoechle, D. 2007. Robust standard errors for panel regres- Roodman, D. 2009. How to do xtabond2: An introduction to sions with cross-sectional dependence. The Stata Journal difference and system GMM in Stata. The Stata Journal 7(3): 281–312. 9(1): 86-136. Isaac, E. and Pei, Y. 2020. Human capital as engine of growth: Sabharwal, K. 2019. Causal link between FDI and eco- the role of knowledge transfers in promoting bal- nomic growth? Panel data analysis of MINT econo- anced growth within and across countries, Germany: mies. International Journal of Business Analytics and Discussion paper series/IZA. Intelligence 7(2): 35-41. Jashari, B. 2020. Digital agenda absorvatory. Regional Sabir, S., Rafique, A. and Abbas, K. 2019. Institutions and FDI: Baseline Research. evidence from developed and developing countries. Jushi, E., Hysa, E., Panait, M. and Voica, M. C. 2021. Financing Financial Innovation: FIN, 5(8): 1-20. growth through remittances and foreign direct invest- Sbaouelgi, J. 2019. The impact of corruption on econom- ment: evidences from Balkan countries. Journal of Risk ic growth in MENA region. Romanian Economic and anf Financial Management 14(3). Business Review 40-54. Keisuke, O. and Sovannroeun, S. 2010. How Does corruption Shimbov, B., Alguacil, M. and Suárez, C. 2019. Export struc- influence the effect of foreign direct investment on eco - ture upgrading and economic growth in the Western nomic growth?. s.l.:Munich Personal RePEc Archive. Balkan countries. Emerging Markets Finance and Trade Loxha, A. 2019. Do remittances reduce poverty in Kosovo?- 55(10): 2185-2210. A counterfactual analysis. The South East European Szymańska, A. 2021. Economic performance in post-soviet Journal of Economics and Business 14(2): 117-132. and post-communist countries–evidence from panel Malthus, T. 1993. An essay on the principle of population. data and multivariate statistical analysis. The South East New York: Oxford Press. European Journal of Economics and Business 16(2): 1-16. Mehic, E., Silajdzic, S. and Babic-Hodovic, V. 2014. The im- Tamar, B. and Luca, G. 2020. Impact of FDI on economic pact of FDI on economic growth: some evidence from growth: the role of country income levels and institu- Southeast Europe. Emerging Markets Finance and Trade tional strength. European Investment Bank (EIB). 49(1): 5-20. The World Bank, 2020. World Bank Indicators. Available at: Melnyk, L., Kubatko, O. and Pysarenko, S. 2014. The impact ttps://data.worldbank.org/indicator (accessed January of foreign direct investment on economic growth: case 26, 2021) of post communism transition economies. Problems and The World Bank, 2020. World Bank Open Data: World Bank Perspectives in Management 12(1). (accessed January 26, 2021). 78 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Wesley, E. and Peterson, F. 2017. The Role of population in Zulfiu Alili, M. 2021. The Impact of foreign ownership economic growth. SAGE Open 7(4): 1-15. on wage inequality and skill intensity. Germany: LAP Wooldridge, J. 2002. Econometric analysis of cross section LAMBERT Academic Publishing . and panel data. Cambridge: The MIT Press. Zulfiu Alili, M. and Adnet, N. 2018. Did FDI increase wage in- World Bank Group, 2017. Job creation picks up: World Bank. equality in transition economies?. International Journal World Bank Group, 2020. Recovery of Western Balkans firms: of Social Economics 54(9): 1283-1304. crowding in private investment and expanding access to advisory support, s.l.: Western Balkans Regular Economic Report 18. World Bank Group, 2021. Subdued recovery, s.l.: The World Bank. Ziberi, B. and Avdiu, M. 2019. Econometric analysis to ex- amine the relationship between unemployment and macroeconomics aggregates: evidence from Kosovo. Academic journal of economic studies 6(2): 33-41. SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 79 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Appendix 1. Variables definition and the source of data Variable Variable label Variable definition Data source name GDP GDP growth (an- Annual percentage growth rate of GDP at market prices based on World Bank nual %) constant local currency. Aggregates are based on constant 2010 Indicators U.S. dollars. GDP is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. POP Population Annual population growth rate for year t is the exponential rate World Bank growth rate of growth of midyear population from year t-1 to t, expressed as Indicators a percentage. Population is based on the de facto definition of population, which counts all residents regardless of legal status or citizenship. REM Remittances Personal remittances comprise personal transfers and compensa- World Bank tion of employees. Personal transfers consist of all current transfers Indicators in cash or in kind made or received by resident households to or from non-resident households. Personal transfers thus include all current transfers between resident and non-resident individu- als. Compensation of employees refers to the income of border, seasonal, and other short-term workers who are employed in an economy where they are not resident and of residents employed by non-resident entities. Data are the sum of two items defined in the sixth edition of the IMF’s Balance of Payments Manual: personal transfers and compensation of employees. FDI Foreign Direct Foreign direct investment are the net inflows of investment to ac - World Bank Investment quire a lasting management interest (10 percent or more of voting Indicators stock) in an enterprise operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earn- ings, other long-term capital, and short-term capital as shown in the balance of payments. This series shows net inflows (new invest - ment inflows less disinvestment) in the reporting economy from foreign investors, and is divided by GDP. UNR Unemployment, Unemployment is defined as the percentage of the workforce that World Bank total (% of is unemployed but looking for job. The terms “labor force” and “un- Indicators total labour employment” are defined differently in different countries. force) (national estimate) CORR Control of Control of Corruption captures perceptions of the extent to which World Bank corruption public power is exercised for private gain, including both petty Indicators index and grand forms of corruption, as well as “capture” of the state by elites and private interests. Percentile rank indicates the country’s rank among all countries covered by the aggregate indicator, with 0 corresponding to lowest rank, and 100 to highest rank. Percentile ranks have been adjusted to correct for changes over time in the composition of the countries covered by the WGI. Higher ‘control of corruption’ percentile ranks denote lower corruption. 80 SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 ECONOMIC GROWTH IN THE WESTERN BALKANS: A PANEL ANALYSIS Appendix 2. Comparison of OLS, FE and RE Tests T-statistic (TS) Prob. <> CV (5%) Decision F test that all u_i=0: F(5, 87)=2.74 0.02 > 2.32 If TS>CV Reject H (Supports Fixed Effects) H : σ =0 chi2(1) = 91.73 0.00 > 3.84 If TS>CV Reject H 0 u 0 (Supports Random Effects) Ho: difference in coefficients chi2(6) =8.68 0.18 > 12.59 If TS<CV There is insufficient not systematic evidence to reject H (RE can be estimated) Source: Author’s calculations Appendix 3. Model diagnostics and plot of the residuals Tests T-statistic (TS) Prob. <> CV (5%) Decision Slope homogeneity chi2(35) = 185.17 0.00 > 49.80 If TS>CV Reject H H :βi =β (No slope homogeneity) Cross-sectional depen- Pesarans’s test= 7.42 0.00 Both tests suggest to reject H . dence in the errors H :Zero cross-sectional Friedman’s test= 42.41 0.00 dependence Groupwise chi2 (6) = 74.34 0.00 > 12.59 If TS>CV Reject H heteroskedasticity (Heteroskedasticity) H :Homoscedasticity Serial correlation F(1, 5) = 2.83 0.06 < 6.61 If TS<CV There is insuffi- st H :No 1 order cient evidence to reject H o 0 autocorrelation Normality of the errors adj chi2(2) =3.21 0.20 < 5.99 If TS<CV There is insuffi- H :Normally distributed cient evidence to reject H Source: Author’s calculations Plot of the residuals Source: Author’s calculations SOUTH EAST EUROPEAN JOURNAL OF ECONOMICS AND BUSINESS, VOLUME 16 (2) 2021 81
South East European Journal of Economics and Business – de Gruyter
Published: Dec 1, 2021
Keywords: Economic growth; panel data; FDI; remittances; unemployment; corruption; E60; O10
Access the full text.
Sign up today, get DeepDyve free for 14 days.