Trade, Foreign Direct Investment and Economic Growth in Albania.

Evidence from Time Series

ISMET VOKA

Department of Management, Faculty of Economics,

University College of Business,

Gjergj Legisi, nr. 23, Tirana,

ALBANIA

ARDI BEZO

Department of Finance, Faculty of Economics,

University College of Business,

Gjergj Legisi, nr. 23, Tirana,

ALBANIA

BARDHYL DAUTI*

Department of Economics, Faculty of Economics,

University of Tetovo,

Ilindenska road, nn, 1200, Tetovo,

REPUBLIC OF NORTH MACEDONIA

*Corresponding Author

Abstract: - Trade and Foreign Direct Investment has been treated as crucial factors underlying the relative

growth rates experienced by the Albanian Economy, especially during the late years, thus, boosting economic

growth in the country and improving the degree of integration of the Albanian economy into the World

markets. The paper aims to provide an empirical assessment of the relationship between Trade, Foreign Direct

Investment, and Economic Growth in Albania, by examining Trade and FDI nexus growth interactions using

yearly time series data for a time span of 1993-2018. For this purpose, we employed cointegration analysis and

Granger causality analysis. The co-integration tests, based on Vector Error Correction Mechanism (VECM),

confirm the presence of a long-run relationship between the variables. VECM results support a negative

relationship between trade and GDP in the long-run and a positive relationship between trade and FDI. Granger

Causality tests support the causality evidence of one-directional reinforcement of GDP on trade in Albania and

changes in GDP and trade are causing changes in FDI. The VAR analysis confirms that changes in GDP and

FDI are encouraging changes in trade. The paper outlines policy implications with respect to promoting

relevant institutional policies for the enhancement of trade and FDI activities in the country, which potentially

could enhance economic growth in the country.

Key-Words: - Albania, FDI, Trade, Growth, Cointegration analysis, time series

Received: January 8, 2023. Revised: May 27, 2023. Accepted: June 5, 2023. Published: June 15, 2023.

1 Introduction

The role of trade and foreign direct investment in

economic growth in transition countries is

considered a crucial ingredient of the globalization

process, involving the principal channel through

which the liberalization process can affect the

output level and therefore the growth prospects of

the economy. The expansion of trade increases

productivity by offering greater economies of scale,

and greater access of the national economy to

international markets, [17], [13], [18], [10]. Foreign

Direct Investment, on the other hand, is perceived as

an important catalyst of economic growth in

transition countries, by producing positive spillover

effects on domestic firms, [4], [11]. The empirical

evidence, which examines a causal relationship

between trade, FDI, and economic growth is

ambiguous, supporting the positive and reverse

association. Trade and FDI are simulative to growth,

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1336

Volume 20, 2023

and on the other hand, there are cases where growth

induces trade and FDI. However, causality analysis

often imposes misrepresentations in the final

inference process, [10]. Therefore, the paper, relying

on yearly time series data for a time span of 1993-

2018, in addition to examining the nature of the

causal relationship between trade, FDI, and

economic growth, employs suitable techniques

regardless of the integration (co-integration) of the

data in the multivariate model. The main motivation

of the study is to empirically investigate the

relationship between trade, GDP, and FDI in

Albania. The research question addressed in the

study is: What is the nature of the causal relations

between trade, GDP, and FDI in Albania in the long

and short term. The Vector Error Correction

Mechanism is used to capture the long-run

relationship between trade, FDI, and GDP. The

result of the paper finds that trade is positively

associated with FDI potentials, supporting the

vertical nature of FDI and the complementary

relationship between trade and FDI, at both short-

run and long-run models, implying that Albania’s

trade potential is highly impacted by FDIs which are

based on a geographically fragmented production

process by stages, [8]. The VECM results are

supporting a negative relationship between trade and

GDP, which is evident due to trade deficits that

occurred in the country in the long term and the high

dependency ratio of Albania’s economy from

imports. The Granger–Causality analysis implies

that trade performance in Albania is associated with

changes in past values of trade and GDP and

changes in trade and GDP are reinforcing changes in

FDI. In addition, the results from VAR analysis

confirm that Albania’s trade flow is reinforced by

the changes in GDP, FDI levels, and the lagged

value of trade. Furthermore, the VAR analysis

confirms that the variation in Albania’s GDP level is

motivated by the agglomeration factor of GDP in a

one-time lag and variations of trade at a three-time

lag. In addition, the same results find that changes in

FDI level are triggered by changes in trade

performance and GDP. The coming section of the

paper focuses on the literature review. Section three

outlines the descriptive nature of the data employed

in the empirical part of the study. Section four

describes the methodology used, econometric

techniques, and estimation results. Section five

discusses the results and section six concludes the

study and gives policy recommendations.

2 Literature Review

Trade for developing countries, such as the case of

Albania as a small and open economy, may induce

the progression of skills through imports of

advanced technology and expertise, through

international markets, hence, reinforcing capital-

intensive production facilities, [14]. Trade openness

typically utilizes encouraging economic growth due

to the enhanced accumulation of physical capital,

sustained technological transfer, and improvement

of the country's macroeconomic conditions, thus

creating a suitable economic environment for

boosting FDI performance, [6], [7]. Inward FDI

enhances the positive spillover effect by promoting

sustained domestic productivity, thus endorsing

capital formation in the host country, [4]. Inward

FDI can stimulate domestic investment through

links in the supply chain where foreign firms

operate internationally by buying locally made

inputs and selling intermediate inputs to local firms,

[8]. Therefore, Trade and FDI generally have been

widely accepted as an important catalyst of the

economic growth process, in the literature on FDI

and Trade nexus growth interactions, [8]. Both, FDI

– growth nexus and Trade – growth nexus literature

have concluded FDI and trade enhance economic

growth, [20]. FDI has a significant impact on the

growth prospects of transition countries by

improving the host country's economic conditions

with respect to the employment situation, incomes,

exports, and economic welfare, [10]. [10], by

employing co-integration analysis, investigating the

relationship between trade, FDI, and economic

growth in Greece, over a yearly time span from

1960-2002, found a long-run relationship between

the three factors. Some other late empirical studies

on the relationship between trade, FDI, and

economic growth in transition countries are

presented in the tables below. Having regard to the

summarized empirical studies in Table 1, it is

evident the outlined impact of Trade and FDI on a

country's economic growth. However, due to the

heterogeneous nature of different countries in terms

of macroeconomic performance, it is suggestive that

such analysis, which involves the relationship

between different economic variables, should be

considered at the country level. The main objective

of this paper is to evaluate the relationship between

FDI, trade, and economic growth in Albania. The

study will add value to the knowledge of the

existing body of literature. We use annual data over

the yearly period: 1993-2018 and employ co-

integration technique to estimate the long-run

relationship between the variables as well as

Granger causality analysis and VAR analysis to

check for the direction of the causal impact between

Trade, FDI, and economic growth in Albania.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1337

Volume 20, 2023

Table 1. Presented empirical studies on the relationship between Trade, FDI, and economic growth

Author Investigation

Sample and period

Methodology

Findings

[1]

Sample: Bulgaria, the Czech

Republic, Estonia, Hungary,

Latvia, Lithuania, Poland,

Romania, Slovakia, and

Slovenia

Period: Quarterly data from

1994 to 2008.

ADRL and Granger

causality test

The three variables have long-term co-

integration in four countries (the Czech

Republic, Slovakia, Poland, and Latvia).

Foreign Direct Investment seemed to be a

more important factor in driving economic

growth than export in these countries.

[2]

Period: 1980–2012

Sample: Turkey.

Time series

techniques, Granger

Causality.

There is no long-term relationship between

foreign direct investment and economic

growth in Turkey.

[3]

Sample countries: BRIC

countries, period: 1989–2012.

Co-integration

analysis

FDI and economic growth are cointegrated at

the panel level and the long-run relationship

between FDI and economic growth is present.

[44]

Has the Foreign

Direct Investment

Boosted Economic

Growth in the

European Union

Countries.

Sample: EU countries, period:

1987–2012.

Feasible GLS (FGLS),

and General Method

of Moments (GMM).

The higher foreign direct investment (FDI)

and portfolio investment (FPI) triggered by

the European Monetary Union (EMU) have

not contributed to growth.

Notes: Summary papers with empirical studies.

3 Data and Stylized Facts

Table 2. Relationship between Trade, FDI, and GDP in Albania

Year

Trade

Absolute

change

Percentage

change

FDI

Absolute

change

Percentage

change

GDP

Absolute

change

Percentage

change

2000

63.45

6.95

2001

66.49

3.04

4.79

1.18

28.64

8.29

1.34

19.28

2002

68.53

2.03

3.06

-2.18

-41.26

4.54

-3.75

-45.24

2003

67.02

-1.52

-2.23

0.07

2.19

5.53

0.99

21.81

2004

67.05

0.03

0.04

1.58

49.72

5.51

-0.02

-0.36

2005

64.27

7.22

10.77

-1.1

-23.06

5.9

0.39

7.08

2006

83.21

8.93

12.03

2.45

67.15

5.98

0.08

1.35

2007

77.45

-5.75

-6.91

3.57

58.49

7.52

1.52

25.42

2008

75.09

-2.36

-3.04

1.49

15.37

3.35

-4.15

-55.33

2009

76.54

1.45

1.93

9.14

-2.03

-18.2

3.71

0.36

10.65

2010

81.22

4.68

6.11

8.14

-1.07

-10.97

2.55

-1.16

-31.34

2011

76.51

-4.71

-5.81

7.45

-0.68

-8.41

1.42

-1.13

-44.31

2012

75.87

-0.64

-0.83

9.82

2.36

31.74

1.08

-0.42

-29.31

2013

75.41

-0.47

-0.61

8.69

-1.12

-11.44

1.77

0.77

77.14

2014

71.83

-3.61

-4.78

8.69

0,05

-0.03

2.22

0.44

25.04

2015

74.81

3.01

4.19

8.81

0.11

1.32

3.31

1.15

49.46

2016

78.19

3.38

4.52

7.86

-0.95

-10.78

3.84

0.49

14.77

2017

76.86

-1.34

-1.71

7.95

0.12

1.22

4.07

0.27

7.08

2018

76.89

0.04

0.05

7.86

-0.09

-1.14

2.24

-1.83

-44.98

Average

2000-10

72.76

n.a

6.21

n.a

5.44

n.a

Average

2011-18

75.79

3.04

4.17

8.39

2.18

35.11

2.48

-2.96

-54.38

2017

76.86

-1.34

-1.71

7.95

0.1

1.22

4.07

0.27

7.08

Notes: Trade is the sum of exports and imports of goods and services measured as a share of gross domestic product.

Foreign direct investment, net inflows (percentage of GDP). The value of foreign direct investment refers to direct

investment equity flows in an economy. GDP is calculated without making deductions for the depreciation of fabricated

assets or for the depletion and degradation of natural resources.

Source: World Bank: World Development Indicators and author’s calculations.

The data set contains information for Albania, based

on aggregate level data (GDP growth, Trade, and FDI). The following table outlines the descriptive

relationship between the data. Here, we illustrate the

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1338

Volume 20, 2023

dynamics of GDP growth in relation to the

dynamics of FDI inflow and Trade flow, during the

period 2000-2018. The data presented in Table 2

outlines a positive correlation between trade and

foreign direct investment. The increase of trade as a

share of GDP between two periods (2000-2010) and

(2011-2018) by 3.04 percentage points, was

followed by the increase of FDI by 2.18 percentage

points. However, following the increase of both FDI

and Trade, the output level decreased marginally by

2.96 percentage points, a result, which can be

attributed to other specific factors affecting the

determining factor of GDP growth, which is beyond

the scope of this research.

4 Methodology and Econometric

Framework

4.1 Unit Root Test

This study tests the relationship between Trade,

FDI, and GDP in Albania. The used methodology in

this paper is based on the Vector Error Correction

Model (VECM) analysis and Granger Causality

analysis. The so-called “co-integration analysis”,

which has provided further support for the vector

error correction model (VECM thereafter), and has

greatly enhanced the approach to non-stationary

time series is employed additionally to capture the

long-run relationship between variables. The

Granger Causality analysis is applied to the study to

capture the nature of the causal relationship between

the variables. The potential causality patterns can be

represented by bi-variate VARs for Albania as

follows:

 

 

   (1)

Where TRjt denotes trade level estimated as the sum

of exports and imports of goods and services, in

millions of US dollars, FDIjt is the Foreign Direct

Investment, net inflows in millions of US dollars,

and GDPjt, is the real GDP estimated with the price

level of 2010, ut is the standard error. All values are

in the logarithm. Testing for unit root is the first step

in macroeconomic time series and essential to

confirm the process by which data could have been

generated is a stochastic one, [19]. For this purpose,

we apply the Augmented Dickey-Fuller test (ADF)

to determine whether the various time series are

integrated at the order of zero I (0), [9]. Co-

integration refers to the fact that two or more series

share a stochastic trend, [21]. [12], suggested a two-

step process to test for cointegration (an OLS

regression and a unit root test), the EG-ADF test. If

the residuals of the OLS regression will be

stationary, the co-integrating regression is

considered as a long-run relationship and we

proceed to the second step, where an Error

Correction Model (ECM), including those lagged

residuals as an error-correction term, is postulated in

order to consider the long-run dynamics. The

starting point in the unit root test is:



  

        (2)

The null hypothesis in the Augmented Dickey-

Fuller test is that the underlying process that

generated the time series is non-stationary. This will

be tested against the alternative hypothesis that the

time-series information of interest is stationary. If

the null hypothesis is rejected, it means that the

series is stationary i.e., it is integrated to order zero.

If, on the other hand, the series is non-stationary, it

is integrated to a higher order and must be

differenced until it becomes stationary, [5]. When

testing for unit root we want to find out whether

equation (2) is equal to one. If

is smaller than

one, the series is stationary. If, on the other hand,

is greater than one, than it would be an explosive

series. Subtracting 

 from both sides in equation

(2), we get equation (3), which is estimated by the

Dickey–Fuller and Augmented Dickey–Fuller test.



  

   (3)

In addition, constant – testing for a random walk

with drift, and time trend – testing for a

deterministic feature, are incorporated. Since the

null hypothesis in equation (2) is that  is equal to

one, in equation (3) it must be that  is equal to

zero. Hence, when  is zero, there is a unit root, and

we have insufficient evidence to reject the null

hypothesis of non -stationary. The Augmented DF

Test is performed on each variable separately, on

the following regression.



     



 

   (4)

The variable 

 equation (4) expresses the first

differences with k lags and the final

is the

variable that adjusts the errors of autocorrelation.

The coefficients    and  are estimated. In

order to test for the stationary of time series, we

have to lag the variables.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1339

Volume 20, 2023

4.2 Augmented Dickey-Fuller Test

When comparing the t statistics with their critical

values as shown in Table 3, we notice that the

variables of Trade, FDI, and GDP are becoming

stationary on their first difference. This means that

the null hypothesis that a given series (Trade, FDI,

or GDP data), contain a unit root and is non-

stationary, was rejected at the first difference of the

respective variables of Trade, FDI, and GDP. The

MacKinnon approximate p-value for Z(t), of 0, as

shown in Table 3, suggests that the null hypothesis

of a unit root and is non-stationary is rejected when

the ADF test is applied to the first difference for all

variables at 1, 5 and 10 percent level of significance.

Hence, in the first difference, all the variables are

becoming stationary and we have sufficient

evidence to reject Ho of unit root presence in our

data. This means that all the variables are integrated

to order I (1).

Table 3. Augmented Dickey-Fuller test of the selected variables in levels

Dickey-Fuller, Log of Trade (Levels)

Lag limit

Absence/Presence of trend

Test statistic

1% critical value

5% critical value

10% critical value

MacKinnon

approximate p-

value for Z(t)

Without trend

-1.207

-3.730

-2.992

-2.626

0.6704

With trend

-3.165

-4.352

-3.588

-3.233

0.0916

Without trend

-1.750

-3.736

-2.994

-2.628

0.4056

With trend

-2.408

-4.362

-3.592

-3.235

0.3755

Without trend

0.039

-3.743

-2.997

-2.629

0.9617

With trend

-2.256

-4.371

-3.596

-3.238

0.4583

Dickey-Fuller, Log of FDI (Level)

Without trend

-1.571

-3.730

-2.992

-2.626

0.4979

With trend

-1.858

-4.352

-3.588

-3.233

0.6760

Without trend

-1.547

-3.736

-2.994

-2.628

0.5103

With trend

-1.864

-4.362

-3.592

-3.235

0.6729

Dickey-Fuller, Log of GDP real (Level)

Without trend

0.006

-3.730

-2.992

-2.626

0.9590

With trend

-4.309

-4.352

-3.588

-3.233

0.0030

Without trend

-1.673

-3.736

-2.994

-2.628

0.4450

With trend

-1.977

-4.362

-3.592

-3.235

0.6141

Dickey-Fuller, Log of differenced TRADE (difference)

Without trend

-7.771

-3.736

-2.994

-2.628

0.0000

With trend

-7.804

-4.362

-3.592

-3.235

0.0000

Dickey-Fuller, Log of differenced FDI (difference)

Without trend

-5.169

-3.736

-2.994

-2.628

0.0000

With trend

-5.261

-4.362

-3.592

-3.235

0.0000

Dickey-Fuller, Log of differenced GDP real (difference)

Without trend

-7.166

-3.736

-2.994

-2.628

0.0000

With trend

-6.904

-4.362

-3.592

-3.235

0.0000

4.3 Cointegration Analysis

Co-integration refers to the fact that two or more

series share a stochastic trend, [21]. [12], suggested

a two-step process to test for cointegration (an OLS

regression and a unit root test), the EG-ADF test.

First, we run the OLS regression on differenced

variables suggested by the stationary test and check

for the presence of unit roots on the residuals

obtained after running the OLS regression. The

stationary test suggests that the regression model

should be estimated in different terms for one-time

lag (Equation:5). Hence, here we can only look at a

short-run relationship among these variables, [5].

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1340

Volume 20, 2023

The final short-run models estimated have the

following form:

   

  

   (5)

Following, [12], we estimate by OLS regression

equation (1), and the obtained residuals from this

regression we test for the presence of unit root

(Table 4).

Table 4. Augmented Dickey-Fuller test of the obtained residuals after estimating OLS regression

Dependent variable LNTRADE: Explanatory variables: LNGDP and LNFDI

Variable

Lag

*limit

Test statistic

1% critical

value

5% critical

value

10% critical

value

MacKinnon approximate p-

value for Z(t)

Residual (e)

-5.068

-3.736

-2.994

-2.628

0.0000

Notes: *Lag limit of one is suggested by the HQIC test

If the residuals were found to be stationary, the co-

integrating regression might be taken as a long-run

relationship and we could then proceed to the

second step, where an Error Correction Model

(ECM), including those lagged residuals as an error-

correction term, would be postulated in order to

consider the long-run dynamics. When we test for

the presence of unit root on the residuals obtained,

after the OLS estimation of equation (1), as shown

in Table 4, we find that the residuals are stationary.

We conclude that the test statistics exceed the

critical values, suggesting no unit root presence on

the obtained residuals (Table 4). The residuals are

stationary, thus confirming the presence of a long-

run relationship between variables. The series are

co-integrated and we continue with the second step,

by analysing the Error Correction Mechanism

(ECM) model. To consider the formal analysis we

regard the postulation of the lagged residuals as an

error correction term, obtained from the OLS

estimation of equation (1). The final ECM model,

capturing the long-run relationship among variables,

has the following form.

   

  

  

(6)

The results from the ECM regression output

(equation 6: column 3), are suggesting that the error

correction mechanism which implies a long-run

equilibrium relationship is statistically significant.

This coefficient of ut-1 shows us how fast the trade

level in Albania changes to disequilibrium changes

in the explanatory variables. Hence, a 1 percent

increase in the speediness of disequilibrium changes

in the GDP and FDI is associated, with average

faster changes in trade level in Albania, at about 0.2

percent. The results are also suggesting that GDP

and FDI are statistically significant in short-run and

long-run models. Focusing on the long-run results

(equation 6: model 3), a negative relationship

between trade and GDP is found, whereas, trade is

positively associated with FDI. Hence, a 1 percent

increase in GDP will affect the average decrease of

trade flow by 2.5 percent in the long-run, whereas, a

1 percent increase in FDI will increase trade flow by

0.6 percent, on average, ceteris paribus. Focusing on

the results which capture short-run relationships

among variables (equation 5: column 2), we outline

that GDP and FDI are positively associated with

trade level. Hence, a 1 percent increase in GDP and

FDI, in the short-run, is associated with the average

increase of Albania’s trade level by 0.5 percent and

0.2 percent, respectively, ceteris paribus. The results

of the macroeconomic factors affecting trade in

Albania are presented in Table 5.

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1341

Volume 20, 2023

Table 5. Results of macroeconomic factors affecting trade in Albania

(1)

(2)

(3)

Dependent variable is log of TRADE

OLS in levels

Equation (2)

OLS in differenced

Equation (5)

Error Correction

Mechanism

(Equation 6)

Log of GDP

.5960***

.5784**

-2.569***

[2.07]

[1.92]

[5.52]

Log of FDI

.2117***

.2149 ***

.5978**

[-2.70]

[2.66]

[1.79]

Lagged residuals (ut-1)

.2017***

[2.24]

Constant

4.722426

5.06131

-52.792

[0.91]

[0.93]

[-11.72]

Observations

R-squared

0.89

0.94

Notes: Dependent variable is log of the TRADE level. t-statistics in brackets, ***, **, and * indicate the significance of coefficients at 1,

5, and 10 percent, respectively. Model 1 shows the results of the OLS equation in levels (specified in equation 1), Model 2 shows the

results of the OLS equation in differenced terms (specified as in equation 5), Model 3 shows the results of the Error Correction

Mechanism (ECM) as specified in equation 6.

4.4 Formal Analysis of Granger Causality

Test

According to [15], Y is said to “Granger-cause” X if

and only if X is better predicted by using the past

values of Y than by not doing so with the past

values of X being used in either case. Essentially,

Granger’s definition of causality is motivated in

terms of predictability. With the regression analysis,

we want to estimate whether trade promotes GDP

and FDI in Albania and whether the GDP and FDI

can encourage the level of Trade. Namely, we want

to find out if the changes in the level of Trade will

respond to changes in the level of FDI and GDP and

vice versa. The Granger causality test applied for

the relationship between trade, FDI, and GDP is as

follows:

    



  



 

 







 

   (7)



    



 

 



 

 





    (8)



    



 

 



 

 





    (9)

Where  

 and 

 are stationary time

series sequences,,  and  are the respective

intercepts, , and  are white noise error terms,

and k is the maximum lag length used in each time

series. The optimum lag length is based on

Granger’s definition of causality and Akaike’s

minimum final prediction error criterion, [15], [16].

If in equation (7), 



 and 





 are

significantly different from zero, then we may

conclude that FDI and GDP granger cause trade. If

in equation (8), 



 and 



 are significantly

different from zero, then we may conclude that FDI

and trade granger cause GDP. Similarly, if in

equation (9), 



 and 



 are significantly

different from zero, then we conclude that GDP and

Trade Granger cause FDI.

4.4.1 Results from Granger Causality Test

For example, the low p-value of 0.000 in the first

row is evidence that the coefficients on the lags of

Gross Domestic Product (LNGDP) are jointly

different from zero in the equation for Trade. This

result indicates that there is sufficient evidence to

reject the null hypothesis of Granger Causality, that

Gross Domestic Product (LNGDP) does not

Granger cause Trade (LNTRADE). On the other

hand, the relatively large p-value of 0.075 in the

second row, favours the conclusion that the

coefficients on the lags of Foreign Direct

Investment (LNFDI) are jointly zero in the equation

for Trade. This result indicates that there is

insufficient evidence to reject the null hypothesis of

Granger Causality, that Foreign Direct Investment

(LNFDI) does not Granger cause Trade. In other

words, the tests show that changes in Albania’s past

values of GDP are causing changes in the trade

performance of Albania, whereas, Albania’s past

values of FDI level are not causing changes in

Trade. Following the same logic, focusing on rows 7

and 8, changes in trade and GDP are causing

changes in FDI. To define the influence of

explanatory variables on the dependent variable, we

employed a Granger causality analysis, which

should point out which occurrence precedes the

other, i.e., whether the trade follows the changes of

the explanatory variables or vice versa, the

explanatory variables follow up the changes in

trade. A Wald test is commonly used to test Granger

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1342

Volume 20, 2023

Causality. The Wald table, reports a Wald test that

the coefficients on the lags of the variable in the

“excluded” column are zero for the variable in the

“equation” column. The results from the causality

analysis using the Granger methodology are

presented in Table 6.

Table 6. Results from the causality analysis using Granger methodology

VAR Granger Causality result

Sample

Yearly time span: 1993-2018

Row

Equation

Excluded

chi2

Prob > chi2

LNTRADE

LNGDP

34.517

0.000

LTRADE

LNFDI

6.916

0.075

LTRADE

ALL

38.116

0.000

LNGDP

LNTRr

7.7415

0.052

LNGDP

LNFDI

6.6118

0.085

LNGDP

ALL

8.7145

0.190

LNFDI

LNTRADE

35.633

0.000

LNFDI

LNGDP

17.469

0.001

LNFDI

ALL

56.611

0.000

4.4.2 Results from Vector and Auto Regression

Model (VAR) Model Table 7. Estimation results from VAR analysis

Notes: t-statistics in brackets, ***, **, and * indicate the significance of coefficients at 1, 5, and 10 percent, respectively (*** p<0.01, **

p<0.05, * p<0.1), standard errors in the brackets (Standard errors in parentheses).

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1343

Volume 20, 2023

Independent variables

Sample:

1993 - 2018

Nr of obs

Log likelihood

102.0095

AIC

-5.539195

HQIC

-5.121172

SBIC

-4.087545

Equation

R-sq.

chi2

P>chi2

LNTRADE

0.9813

1367.846

0.0000

LNGDP

0.9909

2827.241

0.0000

LNFDI

0.9717

892.5728

0.0000

Dependent variables

(1)

(2)

(3)

VARIABLES

LNTRADE

LNGDP

LNFDI

L.LNTRADE

0.719***

-0.0188

2.338***

(0.155)

(0.0761)

(0.460)

L2.LNTRADE

-0.149

0.0752

-0.399

(0.103)

(0.0505)

(0.305)

L3.LNTRADE

0.0791

0.101**

0.769***

(0.0926)

(0.0454)

(0.274)

L.LNGDP

0.977***

0.933***

0.473

(0.364)

(0.179)

(1.079)

L2.LNGDP

0.276

0.0455

1.835

(0.479)

(0.235)

(1.420)

L3.LNGDP

-0.133

0.0655

-3.213***

(0.318)

(0.156)

(0.941)

L.LNFDI

0.103*

-0.0279

0.158

(0.0567)

(0.0278)

(0.168)

L2.LNFDI

-0.000733

-0.0475

-0.252

(0.0603)

(0.0296)

(0.179)

L3.LNFDI

-0.0748

-0.0128

0.217

(0.0479)

(0.0235)

(0.142)

Constant

-14.25***

-2.786

-23.07*

(4.280)

(2.098)

(12.68)

Observations

To make a more formal analysis of the influence of

FDI and GDP on Trade and the influence of the

lagged value of Trade on further trade flow, we

apply the methodology of Vector Autoregression

(VAR), as shown in Table 7, [22]. In the

specification of the model, when we consider Trade

as a dependent variable, (equation 7), the results

showed that statistically significant are the changes

in the first-time lag of trade, Gross Domestic

Product, and FDI. The model set in this manner

gives a satisfactory explanation for the relation

between the changes in Trade, Gross Domestic

Product, and the changes in Foreign Direct

Investment at the first lag, which is evident from the

R square from 0.9813. The first-time lag of the

coefficient of GDP (0.977) in the equation of trade

is highly significant at a 1% level of significance,

(indicated by a low p-value of 0.000), with regard to

the changes in trade, (which points to high trade

flow motivated by the increase of Albania’s GDP).

The coefficient of the first-time lag of Trade

(0.719), in the equation of Trade, is also highly

significant at a 1% level of significance. In addition,

the first-time lag of the coefficient of FDI (0.103),

in the equation of trade is significant at 10 percent

of significance. Thus, according to the VAR model,

the increase of GDP and FDI in the current year by

1 percent will act on an average increase of trade in

the forthcoming period by 0.9 and 0.1 percent

respectively. In addition, the current increase of

trade by 1 percent will impulse the forthcoming

increase of trade flow in the next period by 0.7

percent. When applying VAR analysis to equation 8

(considering GDP as a dependent variable), we see

that the influence of the lagged value of GDP on the

current value of GDP is based on only a one-time

lag, with an estimated impact of 0.933 percent. In

addition, in the equation of GDP, the influence of

the lagged value of trade on the current value of

GDP is based on three-time lags, with an estimated

impact of 0.101 percent. The high explanatory

power of the model of 0.99 gives a satisfactory

explanation for the variation of the explanatory

variables (GDPt-1, TRADEt-3), per unit variation of

the dependent variable (GDP). In the model, the

coefficient of FDI is statistically insignificant,

pointing to the low dependency level of economic

development from the foreign sources of capital,

while the coefficient of TRADE at the third lag is

statistically significant at a 5% level of significance.

Applying VAR analysis to equation 9 (Considering

FDI as a dependent variable), we outline the

significant impact of the first- and third-time lag of

trade on FDI, at one percent level of significance,

with an estimated coefficient of 2.338 and 0.769

percent, respectively and the third time lag of GDP

on FDI, with an estimated coefficient of 3.213.

Based on these results, the increase of trade in the

current year by 1 percent will act on an average

increase of FDI in the coming first and third years

by 2.3 and 0.7 percent, respectively. In addition, the

increase of GDP by 1 percent in the current year,

will act on an average increase of FDI in the coming

third year by 3.2 percent. The high explanatory

power of the model of 0.97 gives a satisfactory

explanation for the variation of the explanatory

variables (TRADEt-1, TRADEt-3 and GDPt-3 ), per

unit variation of the dependent variable (FDI).

5 Discussion of the Results

The results of the study outline that Albania’s trade

is subject to disequilibrium changes in GDP and

FDI, in the long-run. The VECM results in the long-

run, confirm the deteriorating effect of GDP on

trade, probably due to the fact that trade deficits are

likely to occur in the long-run, owing to the high

dependency ratio of Albania’s economy from

imports. In the short-run, both FDI and GDP

enhance trade. The positive relationship between

FDI and trade, in both the short and long-run,

supports the vertical nature of FDI in Albania, thus,

making the FDIs in Albania to be in a

complementary relationship with trade, [8]. On the

grounds of causality analysis, the results of the

study outline that changes in the trade performance

of the country are subject to changes in past values

of GDP, and not to changes in past values of FDI.

On the other way around, changes in FDI are subject

to changes in trade and GDP. The VECM model

proved that there is a long-run relationship between

GDP, FDI, and Trade. Due to the existence of the

long-run relationship between these variables, we

advocate that it is very important for Albania to

create trade promotion policies and FDI-friendly

policies to boost economic growth. In addition, the

same conclusion is reached from the Granger -

Causality test, which points out that the changes in

trade prospects are triggered by the changes in the

past values of trade and GDP and variations in trade

and GDP are triggering changes in FDI. The results

from the VAR analysis are suggesting that the

forthcoming increase in trade is subject to the

agglomeration factor of trade, as well the increase of

GDP and FDI in the current period. When applying

VAR analysis to GDP, the results outline a

significant impact of the one-time lag of GDP and

three-time lag of trade, to the current level of GDP,

whereas, the VAR analysis to FDI confirms the

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1344

Volume 20, 2023

significant impact of the first and third-time lag of

trade on FDI and third-time lag of GDP on FDI. In

general, the VAR results applied to trade, GDP, and

FDI, confirm reasonable clarification for the high

variation of the explanatory variables on the

dependent variable in the three cases, respectively.

6 Conclusions and Recommendations

In this paper, we have estimated the short-run and

long-run relationship between macroeconomic

variables of Trade, Gross Domestic Product, and

FDI, using yearly data for the period 1993-2018, as

well as causality analysis between the three

macroeconomic indicators. i.e. whether the changes

in trade performance are caused by the other

macroeconomic factors associated with FDI and

GDP, and vice versa, considering a bivariate

analysis, whether the changes in GDP and FDI are

caused by the changes on the right-hand side factors

in the second and third equation, respectively. The

vector error correction mechanism results suggest

that the inward stock of FDI is statistically

significant and positively influences the trade

potentials, hence supporting the vertical nature of

FDI in the country. However, the VECM results

outlined the negative impact of GDP on trade, in the

long-run, whereas in the short-run the impact was

found to be positive. The VAR results confirm that

Albania’s trade performance, in addition to GDP

and past values of trade, is also caused by the

changes in FDI. On the other hand, variations in

Albania’s GDP are reinforced by agglomeration

factors of GDP and trade variations. In addition,

changes in FDI level are driven by changes in trade

and GDP. The results of this paper suggest that

Albania’s GDP level is largely dependent upon

trade potentials and agglomeration factors.

Therefore, FDI and trade promotion policies are

expected to play a significant role in the long-term

economic growth of Albania's economy. One trade

promotion policy for Albania, which could be

applied, is securing tariff-free access to the markets

of developed countries, an analysis that is beyond

the scope of this paper. Albania's government could

do much better in terms of FDI promotion policies

with respect to fiscal preferences that potential

foreign investors could benefit in case they locate

their investment potential to the country’s economic

sectors which contain competitive advantage, in

relation to other surrounding countries, for example

in the tourism sector. Since FDI and trade are

verified as important catalysts for Albania’s

economic growth, especially in the long-run, it is

almost of utmost need for the country to build

relevant policy frameworks that will promote

economic growth, which will mainly be FDI-led

growth policies. Another institutionally related

factor that could lead to growth prospects for

Albania, is political stability, generally promoted

through good governance policies, for instance,

improvements in the rule of law and government

effectiveness, which is encouraged through positive

developments in terms of civil, criminal and

informal justice and private sector developments,

respectively. The institutional-related factors are of

crucial importance for Albania’s EU approximation

path, which on the other hand are referring to the

limitations of this study. However, the objective of

this study was to estimate the long-run relationship

of the factors that contribute to the trade prospects

of Albania, like FDI and GDP, and not to provide an

indication of determinants of trade, therefore, these

issues are not critical, but could serve as a milestone

for future economic research for Albania’s trade

performance.

References:

[1] Acaravci, A., & Ozturk, I., Foreign direct

investment, export and economic growth:

Empirical evidence from new EU

countries. Romanian Journal of Economic

Forecasting, Vol. 2, No.2, 2012, pp. 52-67.

[2] Aga, A. A. K., The impact of foreign direct

investment on economic growth: A case study

of Turkey 1980–2012. International Journal

of Economics and Finance, Vol.6, No.7, 2014,

pp. 71-84.

[3] Agrawal, G., Foreign direct investment and

economic growth in BRICS economies: A

panel data analysis. Journal of Economics,

Business and Management, Vol.3, No.4,

2015, pp. 421-424

[4] Damijan, J. P., Rojec, M., Majcen, B., &

Knell, M., Impact of firm heterogeneity on

direct and spillover effects of FDI: Micro-

evidence from ten transition

countries. Journal of comparative

economics, Vol.41, No. 3, 2013, pp. 895-922.

[5] Dauti, B., Determinants of Foreign Direct

Investment in Macedonia. Evidence from

times series. Proceedings of FIKUSZ ’09,

Symposium for Young Researchers.

Published by Budapest Tech Keleti Károly

Faculty of Economics, Tavaszmez u. 15-17.H-

1084Budapest, Hungary.

http://kgk.bmf.hu/fikusz, 2009, 41-55.

[6] Dauti, B., Determinants of foreign direct

investment in transition economies, with

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1345

Volume 20, 2023

special reference to Macedonia. Evidence

from gravity model. The South East European

Journal of Economics and Business, Vol.10,

No.2, 2015a, pp.7-28.

[7] Dauti, B., Determinants of foreign direct

investment in South East European countries

and new member states of European Union

countries. Economic and Business

Review, Vol. 17, No.1, 2015b, pp. 93-115.

[8] Dauti, B., Trade and foreign direct

investment: Evidence from South East

European countries and new European Union

member states. Journal of Economics and

Business, Vol. 34, No.1, 2015, pp. 63-89.

[9] Dickey, D. A., & Fuller, W. A., Distribution

of the estimators for autoregressive time series

with a unit root. Journal of the American

statistical association, Vol. 74, No.1, 1979,

pp. 427-431.

[10] Dritsaki, M., Dritsaki, C., & Adamopoulos,

A., A causal relationship between trade,

foreign direct investment and economic

growth for Greece. American Journal of

applied sciences, Vol.3, No.1, 2004, pp. 230-

235.

[11] Edwards, S., Trade orientation, distortions and

growth in developing countries. Journal of

development economics, Vol. 39, No.1, 1992,

pp. 31-57.

[12] Engle, R. F., & Granger, C. W., Co-

integration and error correction:

representation, estimation, and

testing. Econometrica: journal of the

Econometric Society, 1987, 251-276.

[13] Esfahani, H. S., Exports, imports, and

economic growth in semi-industrialized

countries. Journal of development

economics, Vol. 35, No.1, 1991, pp. 93-116.

[14] Frankel, J. A., & Romer, D. H., Does trade

cause growth? American economic review,

Vol. 3, No. 89, 1999, pp. 379-399.

[15] Granger, C. W., Investigating causal relations

by econometric models and cross-spectral

methods. Econometrica: journal of the

Econometric Society, Vol.37, No.3, 1969, pp.

424-438.

[16] Granger, C. W., Some recent development in

a concept of causality. Journal of

econometrics, Vol. 39, No.1-2, 1988, pp. 199-

211.

[17] Helpman, E., & Krugman, P., Market

structure and foreign trade: Increasing

returns, imperfect competition, and the

international economy. 1987, MIT press.

[18] Lucas Jr, R. E., On the mechanics of

economic development. Journal of monetary

economics, Vol. 22, No. 1, 1988, pp. 3-42.

[19] Nelson, C. R., & Plosser, C. R., Trends and

random walks in macroeconmic time series:

some evidence and implications. Journal of

monetary economics, Vol.10, No. 2, pp. 139-

162.

[20] Pahlavani, M., Wilson, E., & Worthington, A.

C., Trade-GDP nexus in Iran: An application

of the autoregressive-distributed lag (ARDL)

model,2005,

https://ro.uow.edu.au/comparers/144.

[21] Tang, D., Has the foreign direct investment

boosted economic growth in the European

Union countries." Journal of International

and Global Economic Studies, Vol. 1, No.8,

2015, pp. 21-50.

[22] Watson, Mark W., Vector autoregressions and

cointegration. Handbook of econometrics.

Vol. 4, 1994, pp. 2843-2915.

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy).

-Bardhyl Dauti carried out the econometric

assessment, methodology development, and design

of the hypothesis and estimations.

-Ardi Bezo carried out the conceptualization of the

study.

-Ismet Voka was responsible for the investigation

process, the execution of the literature review part,

and the conclusion part.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

The authors of the paper funded the study.

Conflict of Interest

The authors have no conflict of interest to declare

that is relevant to the content of this article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US

WSEAS TRANSACTIONS on BUSINESS and ECONOMICS

DOI: 10.37394/23207.2023.20.119

Ismet Voka, Ardi Bezo, Bardhyl Dauti

E-ISSN: 2224-2899

1346

Volume 20, 2023