Copyright © 2016 The Authors. Published by VGTU Press. This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0) license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The material cannot be used for commercial purposes. THE RESEARCH ON INTERFACE BETWEEN LITHUANIAN DIRECT INVESTMENT ABROAD AND FOREIGN TRADE FLOWS Anžela KOZLOVA1, Algita MIEČINSKIENĖ2 Department of Finance Engineering, Faculty of Business Management, Vilnius Gediminas Technical University, Saulėtekio al. 11, LT-10223 Vilnius, Lithuania E-mails: 1kozlovaanzela@gmail.com; 2algita.miecinskiene@vgtu.lt (corresponding author) Received 4 April 2016; accepted 31 May 2016 Abstract. The scientific research results related to foreign trade and direct invest- ment abroad (DIA) are discussed in the article. The relation of the direct invest- ment abroad and foreign trade is still under the discussion as there is no clear answer whether foreign trade is supplemented or replaced by the direct investment abroad. Since 1997 the flows of the direct investment abroad increased greately in Lithuania. Consequently, it is important to define the link between the DIA and foreign trade considering each country separately. Direct investment abroad and trade links in Lithuania in 1997–2014 are analyzed in the article. The research analysis involves Lithuanian direct investments in the developed countries except some countries, such as Belorus, Russia and Ukraine. It is defined that there is a positive bilateral link between Lithuanian direct investment abroad and foreign trade. It is also observed the impact of general development of Lithuanian direct investment abroad (considering certain countries) on the countries economy it- self – imports can exceed exports. Engle-Granger causality test is applied in the research paper for the purpose of defining the impact of the DIA on the import and export range. Keywords: direct investment abroad, export, import, Engle-Granger Causality test, foreign trade. JEL Classification: F14, F21. 1. Introduction Scientists are greately interested in the impact of the direct investment abroad (DIA) on the trade. However, both scientific analysis models and empirical evidence obtained concerning the direct investment abroad impact on foreign trade are not the same. Two attitudes considering direct investment abroad and foreign trade are formed in the scien tific literature: some authors (Mundell 1957; Fonseca et al. 2010) claim that direct investment abroad and foreign trade substitutes each other, while others (Schmitz, Helmberger 1970; Pfaffermayr 1996) argue that direct investment abroad has not only positive links with foreign trade but can also supplement it. B u s i n e s s, Ma n ag e M e n t a n d e d u c at i o n ISSN 2029-7491 / eISSN 2029-6169 2016, 14(1): 136–151 doi:10.3846/bme.2016.321 mailto:2algita.miecinskiene@vgtu.lt http://dx.doi.org/doi:10.3846/bme.2016.321 137 Business, Management and Education, 2016, 14(1): 136–151 Lithuania (the same as other countries) takes active part both, in global trade and stock movement, though recently enough. Following the data of the Statistics Lithuania in 1997 Lithuanian direct investment abroad was 30.08 Eur mill. where 5.88 Eur mill. to EU countries and 23.42 Eur mill. to Commonwealth of Independent States (CIS) countries. In 2015 the flow of Lithuanian direct investment abroad increased by 2052.60 Eur mill. where 1860.68 to EU countries and 170.01 Eur mill. to CIS countries. So, the questions arise: Does Lithuanian direct investment abroad act as a foreign trade supple- ment or replacement? What is the economical impact on the country? The aim of the article is to estimate the impact of Lithuanian direct investment abroad on foreign trade from 1997 to 2014, while analysing the impact of direct invest- ment abroad on foreign trade considering each country separately. Engle-Granger Causality test is applied in the research work for the purpose of defin- ing the impact of the direct investment abroad on the import and export range. 2. The analysis of preceding empirical investigations of direct investment abroad and foreign trade relation Two attitudes are formed in scientific literature: one – direct investment abroad and foreign trade substitutes each other, and the other – direct investment abroad and foreign trade act as a supplement to the trade. While estimating scientific literature it is noticed chronologically by previous research that capital export and foreign trade substitutes each other, however, it is defined by later research that direct investment abroad might supplement foreign trade. Mundell (1957) was one the first to analyse the link between trade and movement of production factors (two countries, two goods, two production factors) referring Heck- scher-Ohlino Model. However, while entering international mobility of production fac- tors between the countries, it was defined that international capital export and foreign trade act as substitutes. It is assumed in Schmitz and Helmberger (1970) model, that production factors are mobile as well as international production functions are not identical. Mundell (1957) model claims production functions to be homogenous and identical in both of the coun- tries. Schmitz, Helmberger (1970) concludes that direct investment abroad is a supple- ment to foreign trade. With the aim to estimate the relation of direct investment abroad with export “verti- cal” (Helpman 1984) and “horizontal” (Markusen 1984) models of integration analy- sis appeared. The case of “vertical” integration involves the intermediate product sale between subsidiaries in rich with stock countries to rich with work places countries, that‘s why foreign trade might be supplemented. “Vertical” integration is considered to be the case when different segments of production process are set out in different coun- tries. “Vertical” model of direct investment seems to be more linked to the investments 138 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... of developing economies (Markusen et al. 1996). “Horizontal” integration is the case when a subsidiary is formed in the country where the same or similar production to the market of the host country is manufactured (Markusen 1984), thus the export to the country might be changed by the investment. Such a conclusion is made by Ekholm et al. (2003). They claim that “horizontal” integration acts as a substitute and “verti- cal” – as a supplement. One more conclusion is made by Markusen et al. (1996) that „horizontal“ model dominates in the countries of the same size and have average or big marketing costs. Empirical research work is made and both, model of gravity and econometric were applied. However, there is no one opinion considering the question, so as it is confirmed by the empirical research that direct investment abroad acts as foreign trade substitutes or might act as a supplement. Austrian direct investment abroad was analysed by Pfaf- fermayr (1996) and positive bilateral relation between direct investment abroad and export was defined. Bajo-Rubio, Montero-Muñoz (1999) made analysis of relation be- tween Spanish direct investment abroad and export, so the conclusion was made that direct investment abroad has a positive impact on the growth of export. The authors claim, that direct investment abroad is very important as the way to promote export. Also, links of direct investment abroad and export was analysed by Fontagné (1999) and the conclusion was that direct investment abroad promotes the growth of export as well as supplement foreign trade. Analysis showed that each dollar invested directly abroad leads to 2 additional export dollars, “One Direct investment abroad dollar is related to 2.3 additional export dollars and 1.9 dollars of additional import”. Portugese direct investment abroad and foreign trade was analysed by Fonseca et al. (2010) and it was noticed that investment had a negative impact on the balance of trade with the exception trends in Angola and Spain. Gross Domestic Product (GDP) per capita, the distance from Lisbon to other countries, Portugese direct investment to 15 EU countries, USA, Brazil, Angola, Japanese and China was analysed by the study and the conclusion was made, that Portugese direct investment abroad is negatively related to export, the substitution effect was noticed. Falk and Hake (2008) analysed the links of export and direct investment abroad and claimed that direct investment abroad might be promoted by export but not vice versa. It was written by the authors that export is supplemented but not substituted by direct investment abroad. Seo and Suh (2006) estimated Korean direct investment to Asian countries, that is Indonesia, Malaysia, Philippines, Thailand applying econometric model. It was noticed that direct investment abroad doesn‘t have the impact of substitute neither on export nor on import, though it was clear that by that time the flows of direct investment abroad could give slight contributions to Korean reagional export. Kang (2012) made analysis on the impact of Korean direct investment abroad and export to the developed as well as developing countries. The conclusion made was, that Korean export to the developing 139 Business, Management and Education, 2016, 14(1): 136–151 countries might be enhanced by direct investment to those countries. Foreign trade with developing countries is supplemented by the flows of direct investment abroad, though it has no impact on the trade with develped countries. The analysis of the impact made on direct investment abroad in Asian countries was made by Ahmad et al. (2015) and the conclusion was, that direct investment abroad is a supplement to foreign trade. Significant positive factors of four Asian countries (Malaysia, Philippines, Singapore and Thailand) were obtained. Chow (2012) analyzed the effect of outward FDI, both country by country and host groups as a whole, on Tai- wan’s exports. It is concluded that outward FDI has a complementary effect on home country’s export in Taiwan. The scientific literature gives no unambiguous answer to the question: what the im- pact is of direct investment abroad on foreign trade. The final conclusion can’t be made following the research performed to find out whether the flows of direct investment abroad in a small developed country give some contribution to the growth of foreign trade and what the impact on trade balance is as the obtained results differ. 3. The research methoodology Lithuania is involved in the movement of international stock just most recently in com- parison to other countries, where they have long experience, though today the country involved in the process handles both, import and export. Nevertheless, Lithuania is rather small country but it is classified as developed country used by UNCTAD. The data range from 1997 till 2014 was used for the research work. There was ana- lysed direct investment abroad and foreign trade with such countries as Belorus (BY), Estonia (EE), United Kingdom (UK), Cyprus (CYP), Latvia (LV), Poland (PL), Russia (RU), Germany (DE), the Netherlands (NL), Sweeden (SE), Ukraine (UA), Italy (IT). One region of the country e.g. Kaliningrad region of Rusian Federation (KAL) was separated for analysis, because direct investment to this region is very important and consist a big part of all investment abroad. Matching the data presented by the Depart- ment of Statistics in 1997–2014 these countries are considered to be the most important partners of export and import. The statistic information of data basis owned by Lithu- anian Republic Department of Statistics (Official… 2016b) as well as the information presented on the website (Official… 2016a) was used. Furthermore, the research had the use of annual data of direct investment abroad flows in Lithuania as well as the extent of annual export and import in each country separately (Official… 2016c; UNCTAD 2014; Eurostat 2016). To identify, whether direct investment abroad changes do have the impact on import and export range, Engle-Granger Causality Test was applied. Engle-Granger Causality Test because of its versatility is applied in most of the spheres. In this case it was used to check whether a pair of indictors are cointegrated and if long-term relation exists among the indicators. 140 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... Engle-Granger Causality Test is performed in several stages (Kvedaras 2005; SAS/ ETS®9.2 2010): – Checking the stacionarity of indicators aplying expanded Dickey-Fuller test (ADF); – Regression is defined by the least squares method, X and Y; – Dickey-Fuller test is used to determine the margin of error of regression model having the aim to check the margin of error of stationarity; – In case the margin of error is stationary error correction model (ECM) is worked out. Otherwise it might be claimed the indicators are not cointegrated. The first stage is to check analyzed indicators (a pair of indicators) whether they meet assumption of stationarity. This study involves checking the time lines stationarity in the broadest sense. The process cosidered to be stationary in the broadest sense if: 1. 2 tEY <∞ , for any t T∈ ; 2. 2t 0EY =EY , for any t T∈ ; (1) 3. ( )cov t,s =cov(t+h,s+h) for any t,s,h T∈ . To check the stationarity extended Dickey-Fuller test is used. In case analyzed nomi- nal indicators don‘t meet the assumption of stationarity, the indicators are integrated by the first line (I(1)), that is, indicator changes to previous period are calculated. If the data doesn’t meet stationarity definition the indicators are integrated by the second line (I(2)). Usually indicators integrated by the first line are used for econometric research because of the interpretation simplicity. By using ADF it is defined whether the time line is matched to the integrated process in the first line involving stochastic trend component, also, if the time line is matched to the stationarity process involving determed trend component. Model (2) is used to perform Dickey-Fuller test: 1 1 1 ∆ = + β ⋅ + γ ⋅ + δ ⋅ ∆ + … + δ ⋅ ∆ + νt t t t p tY c t Y Y Yp- - -   , (2) Where: c and β – constants , and p – delay line. Hypothesis tested: ( ) ( ) ( ) ( ) 0 0 0 0 0 β γ = β γ ≠a H c c H c c , , , , ; , , ,: , : . (3) Having performed the stationarity test, it is moved to the next stage where regres- sion model is worked out. The regression equation is calculated by using the method of the least squares: t t ty c x e= + β ⋅ + , (4) Next, the square sums of both equation remains are compared: T 2 1 t 1 RSS = = ∑ tû , T 2 0 t t=1 RSS =∑ê . (5) 141 Business, Management and Education, 2016, 14(1): 136–151 The test statistics (S1) is calculated by formula 6. ( )0 1 21 1 T(RSS -RSS ) S = ~ p RSS χ , (6) If the test statistics exceeds critical value of theoretical statistics, then hypothesis H0 (claiming X is no Y Granger cause) is rejected. Models of Panel Data. There are formed 2 types of panel data regression models in order to define the impact of direct investmant abroad on import and export. Indicator changes (I(1)) and fixed effect regression is used in the study. DIA impact on import and export indicator factors are defined by these 2 models. I(1) mathematical presentation of regression model: ∆ = β ⋅ ∆ + ∆εit it itY TUI , (7) where: ∆ itY – import/export annual changes in the i country; ∆ itTUI – DIA annual changes in the i country; β – DIA impact on import/export; εit – casual margin of error; t – time period; i – the country. Mathematical presentation of fixed effect model: -( ) ( - )iit i it itY Y TUI TUI= β × + ε , (8) where: Yit – import/export indicator in the i country; TUIit – DIA indicator in the i country; β – DIA impact on import/export; iY – DIA average in the i country; iTUI – import/export average in the i country; εit – casual margin of error; t – time period; i – the country. Rating of model parametres and testing of statistic significance. For regression of changes in indicator I(1) rating as well as rating of fixed effects and model parametres the least squares model (formula 4) is used. The main goal of this stage is to obtain ratings of matched parametres. It is also of great importance to have ratings of worked out model parametres being statistically significant. Statistically significance of model parametres rating is defined having tested hypothesis: 0 0 a 0 H : =0; H : 0. β β ≠ (9) For H0 hypothesis testing F-statistics is used. In case hypothesis H0 is rejected, the conclusion arises that the model is statistically significant so, DIA variation is justified by export (import) variation. 142 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... 4. Empirical test results 4.1. Import, export and direct investment trends in investigated countries In order to perform analysis of indicator relation, the survey of import, export and DIA variation in different countries during a certain time period is made. As it is shown in the Figure 1 (in horizontal axix years is represented and ranges from 1997 to 2014), the import of Poland is much bigger having compared with the other countries during the testing time period. Russia and Latvia are the leading coun- tries considering the import extent. The import extent of the other countries are similar. 1 = ‘BY’; 2 = ‘EE’; 3 = ‘UK’; 4 = ‘CYP’; 5 = ‘LV’; 6 = ‘PL’; 7 = ‘RU’; 8 = ‘KAL’; 9 = ‘DE’; 10 = ‘NL’; 11 = ‘SE’ Fig. 1. The import index variation in different countries (Eur mill.) (source: created by authors) While analysing export standard variation it‘s clearly seen that in 1999 Russian export was greately enlarged in comparison with the export extent in other countries. Though, the other years data shows that Poland was the leading country considering the export extent. While analysing the crisis impact on the export extent it‘s clear, that in 2009 the export extent was decreasing in all the countries (Fig. 2) (in horizontal axix years is represented and ranges from 1997 to 2014). Monitoring the rate variation of different DIA indicators it‘s clear, that significant growth of investment is fixed in Lithuania from 2003 (Fig. 3) (in horizontal axix years is represented and ranges from 1997 to 2014). 143 Business, Management and Education, 2016, 14(1): 136–151 1 = ‘BY’; 2 = ‘EE’; 3 = ‘UK’; 4 = ‘CYP’; 5 = ‘LV’; 6 = ‘PL’; 7 = ‘RU’; 8 = ‘KAL’; 9 = ‘DE’; 10 = ‘NL’; 11 = ‘SE’ Fig. 2. The export index variation in different countries (Eur mill.) (source: created by authors) 1 = ‘BY’; 2 = ‘EE’; 3 = ‘UK’; 4 = ‘CYP’; 5 = ‘LV’; 6 = ‘PL’; 7 = ‘RU’; 8 = ‘KAL’; 9 = ‘DE’; 10 = ‘NL’; 11 = ‘SE’; 12 = ‘IT’; 13 = ‘UA’ Fig. 3. DIA index variation in different countries (Eur mill.) (source: created by authors) 144 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... In Latvia DIA rates were steady during the testing time period. The biggest DIA rates were in the Netherlands starting from 2009 and it remains at the same position till 2014. 4.2. Granger causality analysis Before analysis started, stationarity assumption of analysed indicators was tested, so as the assumption of indicator cointegration is such that indicator data must be station- ary. Dickey-Fuller expanded criteria was applied to test stationarity data. The results obtained showed, that all the indicators (DIA, export and import) meet the stationarity assumption because p < 0.05 (see Table 1). Table 1. Static results (Dickey-Fuller test) of export and import index and DIA (source: cre- ated by authors) DIA EXP IMP test statistic: tau_c(1) –4.13658 –5.19251 –3.50822 asymptotic p-value 0.0008412 8.107e-006 0.007806 In the other stage of study Granger Causality test was applied. The goal of using this test is to define whether the DIA indicator is the cause of Granger import/export, that is if DIA extent variation impacts the changes of import/export. During the analysis both, the test statistics and error assumption is calculated. The obtained results showed, that DIA is the cause of export Granger indicator, as tau_c(1) = –4.159 (p = 0.004 < 0.05). And besides DIA is the cause of import Granger indicator, as tau_c(1) = 4.387 (p = 0.0018 < 0.05). It means, that the margin of error of both worked out regression models (having not depending variable export/import extent) meet the stationarity assumption and it indicates accordingly the rate of cointegration. While analysing relation of export/import indicator variation it was defined that, these indicators have specific bilateral links as Granger causality criterion results are statistically significant, when not depending regression variable standard is both, for export and import (respectively, p < 0.05) (Table 2). Table 2. Granger causality test results (source: created by authors) test statistic: tau_c(1) asymptotic p-value DIA – EXP –4.15916 0.004155 EXP – DIA –1.00775 0.9002 DIA – IMP 4.38715 0.001817 IMP – DIA –0.8893 0.9198 EKSP – IMP –6.47986 7.614e-008 IMP – EXP –4.54145 0.001002 145 Business, Management and Education, 2016, 14(1): 136–151 Summing up Granger casualty results, it must be said, that DIA variation is related to export/import extent variation. This study deals with analysis of indicator relation in the sense of cointegration, so the discussed relations occur during the long time period. 4.3. DIA impact on import and export In further stage of study using Granger causality test defined relation trends were ana- lysed, so as it is important to know not only if there are links related to the indica- tors, but also tendency of DIA impact on export/import ratings. Cosequently, there was worked out two types of models having one regression factor: (a) I(1) regression and (b) fixed effect models. The first model is specific as there rating variation considering the previous time period were analysed but not the nominal value of index. The obtained results shows, that having integrated the indexes in the first line the DIA impact on import extent is not important statisticly, because d_DIA = 0.266 (p value = 0.742 > 0.05) (Table 3). Besides, determination coefficient of the worked out model is rather low (R2 < 0.1), that’s why the model results are not acceptable and it‘s necessary to modify the model of regression. Table 3. I(1) regression IMP-DIA results (source: created by authors) Coefficient Std. error t-ratio p-value Const 81.3813 34.6743 2.3470 0.0201 ** d_DIA 0.266265 0.80745 0.3298 0.7420 R-squared: 0.000636. ** The value of parameter is statistically significant when statistical significance level is 0.05. Analogous situation is monitored and the impact of DIA integrated index on export extent is analysed in the first line. The impact of investment is not significant statisti- cally, so as d_DIA = 0.709, p value = 0.1089 > 0.05 (Table 4). It shows, that DIA an- nual growth in the countries doesn‘t have statistically significant impact on the growth of export. The determination coefficient of the worked out model is also low enough (R2 < 0.1), so the results obtained by the model are not acceptable and it is necessary to modify the model of regression. Table 4. I(1) regression EKSP-DIA results (source: created by authors) Coefficient Std. error t-ratio p-value Const 77.7222 18.8999 4.1123 <0.0001 *** d_DIA 0.709218 0.440115 1.6114 0.1089 R-squared: 0.01. ***The value of parameter is statistically significant when statistical significance level is 0.01. 146 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... Summing up I(1) the results of regression models it can be said, that using the gener- al sample data (pooled OLS), modelling DIA impact on import and export is not accept- able in different countries, as it‘s not taken in consideration the range of indicators of each country separately what causes error while calculating the parametre estimations. In the other stage of study the alternative strategy of modelling is used, applying the panel model of fixed effects which enables to estimate rating changes in each country separartely. While working out the model, the average of DIA and import/export ratings are calculated in each country and making further calculations using regression equa- tion, differentiated values matching the general rating of each country are applied but not nominal DIA and import/export rating values. The modelling process mentioned above gives us the possibility to eliminate wrong relation between analysed ratings in different countries. The results of fixed effect model show that DIA impact on import extent is significant statistically, because DIA = 4.26 (p value = 0.0001 < 0.05) (Table 5). General determination coefficient of created model is 0.699 (least-squares dummy variables (LSDV)) and it means that average data of created model is appropriate. More important is within determination coefficient which equals 0.2496. It shows that created panel regression justifies 25% on average the dissemination of import index in each country, according to DIA changes. Table 5. Fixed effects regression IMP- DIA results (source: created by authors) Coefficient Std. error t-ratio p-value Const 669.743 66.5765 10.0597 <0.0001 *** DIA 4.26214 0.561813 7.5864 <0.0001 *** LSDV R-squared: 0.699311. Within R-squared: 0.249632. *** The value of parameter is statistically significant when statistical significance level is 0.05. In case of estimated regression in export range it was defined that DIA impact on export standard is significant statistically, because DIA = 3.53 (p value = 0.0001 < 0.05) (Table 6). LSDV determination coefficient of created model equals 0.52, it shows the model average data is appropriate. Within determination coefficient equals 0.28, so the panel regression justifies 28% on average the dissemination of export index in each country, according to DIA changes. Table 6. Fixed effects regression EKSP- DIA results (source: created by authors) Coefficient Std. error t-ratio p-value Const 522.154 50.8122 10.2762 <0.0001 *** DIA 3.53282 0.428784 8.2392 <0.0001 *** LSDV R-squared: 0.516438. Within R-squared: 0.28181. *** The value of parameter is statistically significant when statistical significance level is 0.01. 147 Business, Management and Education, 2016, 14(1): 136–151 Summing up the panel regression models it might be said that better explanation of import and export indicators range in different countries according to DIA range is given by the panel fixed effect model than by I(1) regression models. It says, it’s very important to estimate the economical impact of individual country applying fixed ef- fect regression. Nevertheless, it‘s possible to develop worked out fixed effect models, eliminating monitoring which gives rather big divergency from the dominant tendency. Thus, the import and export extent of Poland is much more bigger comparing to other countries during the test period, so, when the Polish data is removed, developed fixed effect models are worked out and the estimation of DIA impact on import and export might be given (not including Polish data). The obtained results show, that DIA im- pact on import extent remains positive and statistically significant, because DIA = 2.62 (p value = 0.0001 < 0.05) (Table 7). LSDV determination coefficient of the worked out model equals 0.77, what is the evidence of high coherence of data. Within determination coefficient equals 0.39. In comparison to previous worked out models, the new model describes by 15% more data dissemination, so the removing the Polish data had the positive impact on general coordination of the model. Table 7. Fixed effects regression IMP- DIA results (having Polish indicators removed) (source: created by authors) Coefficient Std. error t-ratio p-value Const 494.05 30.7291 16.0776 <0.0001 *** DIA 2.623 0.261405 10.0342 <0.0001 *** LSDV R-squared: 0.770154. Within R-squared: 0.392253. *** -The value of parameter is statistically significant when statistical significance level is 0.01. Figure 4 shows (in horizontal axix years is represented and ranges from 1997 to 2014), the model gives rather exact range description of import index in different coun- tries. The biggest variations of import index from modelled values are monitored in Latvia, Russia and Germany, that is the countries where import and DIA extent had the biggest range. The obtained results show, that DIA impact on export extent is also positive and statistically significant, because DIA = 2.26 (p value = 0.0001 < 0.05) (Table 8). LSDV determination coefficient of the worked out model equals 0.50, it shows me- dium coherence of data. Within determination coefficient equals 0.25. In comparison to initial panel model (full taken), the quality of this model became slightly worse, so, removal of Polish data didn‘t have the positive impact on general coordination of the model. However, initial and new export regression models are equivalents considering predicted accuracy. 148 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... 1 = ‘BY’; 2 = ‘EE’; 3 = ‘UK’; 4 = ‘KIP’; 5 = ‘LV’; 6 = ‘RU’; 7 = ‘KAL’; 8 = ‘DE’; 9 = ‘NL’; 10 = ‘SE’; 11 = ‘IT’; 12 = ‘UA’ Fig. 4. Fitted and actual values of the import regression model (source: created by authors) Table 8. Fixed effects regression EKSP- DIA results (having Polish indicators removed) (source: created by authors) Coefficient Std. error t-ratio p-value Const 500.499 36.7683 13.6123 <0.0001 *** DIA 2.26132 0.312778 7.2298 <0.0001 *** LSDV R-squared: 0.503417. Within R-squared: 0.250971. *** The value of parameter is statistically significant when statistical significance level is 0.01. Figure 5 shows (in horizontal axix year is represented an range are from 1997 to 2014) that export index range in different countries is described with high precision by the model. The data of Latvian export extent is described poorly so, as in Latvian sample of data outliers is observed (export extent > 4000). Summing up fixed effects of import and export regression models results it might be said, these are the best alternatives for panel data modelling in comparison with the regression models of general sample. Fixed effect regression results show, that having DIA increase of 1 Eur mill. the import extent of the country has an average increase 149 Business, Management and Education, 2016, 14(1): 136–151 of 2.62 Eur mill. While analysing DIA impact on the export extent, it might be said, that having DIA increase of 1 Eur mill., export extent in the countries has an average increase of 2.26 Eur mill. So it is seen from the research data that foreign trade and Lithuanian direct investment are supplements. 5. Conclusions The question of relation between the direct investment abroad and trade is under the discussion in the scientific literature, so as there is no unambiguous answer whether the direct investment abroad is a supplement or substitution of the trade. Obtained conclu- sions of empirical studies are quite different: there is a positive bilateral link between direct investment abroad and foreign trade, direct investment abroad supplements trade (motivates foreign trade) or may add its slight contribution to the export growth, direct investment abroad substitutes the trade. Just recently Lithuania takes part in the interna- tional stock movement. Since 1997 the flows of the direct investment abroad increased greately in Lithuania, today its bigger part has EU countries. The research analysis in- volves Lithuanian direct investments in the developed countries except some countries having economy in transition, such as Belorus, Russia and Ukraine. It is determined by 1 = ‘BY’; 2 = ‘EE’; 3 = ‘UK’; 4 = ‘KIP’; 5 = ‘LV’; 6 = ‘RU’; 7 = ‘KAL’; 8 = ‘DE’; 9 = ‘NL’; 10 = ‘SE’; 11 = ‘IT’; 12 = ‘UA’ Fig. 5. Fitted and actual values of the export regression model (source: created by authors) 150 A. Kozlova, A. Miečinskienė. The research on interface between Lithuanian direct investment abroad... the study, that the range of Lithuanian direct investment abroad is related to the range of export and import extent and also there is a positive bilateral link. It is also observed the impact of general development of Lithuanian direct investment abroad (considering certain countries) on foreign trade and DIA supplement foreign trade. References Ahmad, F.; Draz, M. U.; Yang, S.-C. 2015. Outward FDI and home country exports: empirical evi- dence from ASEAN Region. Social Science Electronic Publishing, 1–16. http://dx.doi.org/10.2139/ssrn.2649493 Bajo-Rubio, O.; Montero-Muñoz, M. 1999. Foreign direct investment and trade: a causality analysis, Open Economies Review 12(3): 305–323. http://dx.doi.org/10.1023/A:1011185507169 Chow, P. C. Y. 2012. The effect of outward foreign direct investment on home country’s export: a case study on Taiwan, 1989–2006, The Journal of International Trade & Economic Development: An Inter- national and Comparative Review 21(5): 725–754. http://dx.doi.org/10.1080/09638199.2010.493616 Ekholm, K.; Forslid, R.; Markusen, J. R. 2003. Export – platform foreign direct investment. The Na- tional Bureau of Economic Research. NBER Working Paper No. 9517 [online], [cited 5 May 2016]. Available from Internet: http://www.nber.org/papers/w9517 Eurostat [online]. 2016 [cited 6 May 2016]. Available from Internet: http://ec.europa.eu/eurostat/data/ database Falk, M.; Hake, M. 2008. A panel data analysis on FDI and exort: FIW Research Report No. 012. Austrian Institute of Economic Research [online], [cited 5 May 2016]. Available from Internet: http://www.wifo. ac.at/jart/prj3/wifo/resources/person_dokument/person_dokument.jart?publikationsid=34229&mime_ type=application/pdf Fonseca, M.; Mendonça, A.; Passos, J. 2010. Home country trade effects of outward FDI: an analy- sis of the Portuguese case, 1996–2007. FEP working papers. Research Work in Progress Nr. 365 [online], [cited 6 May 2016]. Available from Internet: http://www.fep.up.pt/investigacao/workingpa- pers/10.03.01_wp365.pdf Fontagné, L. 1999. Foreign direct investment and international trade: complements or substitutes? OECD Science, Technology and Industry Working Papers 1999/03 [online], [cited 6 May 2016]. Avail- able from Internet: http://www.oecd-ilibrary.org/docserver/download/5lgsjhvj7n0r.pdf?expires=146470 1513&id=id&accname=guest&checksum=6A51977438A2CE5A96DCB5D826F605AF Helpman, E. 1984. A simple theory of international trade with multinational corporations, Journal of Political Economy 92(3): 451–471. http://dx.doi.org/10.1086/261236 Kvedaras, V. 2005. Ekonometrijos įvadas. Vilnius, 3–48. Kang, K. 2012. Is the relationship between foreign direct investment and trade different across de- veloped and developing countries? Evidence from Korea, Asian-Pacific Economic Literature 26(2): 144–154. http://dx.doi.org/10.1111/j.1467-8411.2012.01343.x Markusen, J. R. 1984. Multinational, multi-plant economies, and the gains from trade, Journal of International Economics 16(3-4): 205–266. http://dx.doi.org/10.1016/S0022-1996(84)80001-X Markusen, J. R.; Venables, A. J.; Konan, D. E.; Zhang, K. H. 1996. A unified treatment of horizontal direct investment, vertical direct Investment, and the pattern of trade in goods and services. The na- tional bureau of economic research. NBER Working Paper No. 5696 [online], [cited 15 May 2016]. Available from Internet: http://www.nber.org/papers/w5696 http://dx.doi.org/10.2139/ssrn.2649493 http://link.springer.com/journal/11079 http://ec.europa.eu/eurostat/data/database http://ec.europa.eu/eurostat/data/database http://www.fep.up.pt/investigacao/workingpapers/10.03.01_wp365.pdf http://www.fep.up.pt/investigacao/workingpapers/10.03.01_wp365.pdf http://dx.doi.org/10.1086/261236 http://dx.doi.org/10.1111/j.1467-8411.2012.01343.x http://dx.doi.org/10.1016/S0022-1996(84)80001-X 151 Business, Management and Education, 2016, 14(1): 136–151 Mundell, R. 1957. International trade and factor mobility, The American Economic Review 47(3): 321– 335. Official statistics portal. 2016a. Foreign trade [online], [cited 16 May 2016]. Available from Internet: http://osp.stat.gov.lt/temines-lenteles43 Official statistics portal. 2016b. Database of indicators [online], [cited 16 May 2016]. Available from Internet: http://osp.stat.gov.lt/statistiniu-rodikliu-analize1 Official statistics portal. 2016c. Lithuanian direct investors at the end of the year, Lithuanian direct investment abroad at the end of the year [online], [cited 16 May 2016]. Available from Internet: http:// osp.stat.gov.lt/web/guest/statistiniu-rodikliu-analize?portletFormName=visualization&hash=6d9d5cee -014b-4e1c-9e41-bc40d749919d Pfaffermayr, M. 1996. Foreign outward direct investment and exports in Austrian manufacturing: substitutes or complements, Weltwirtschaftliches Archiv 132(3): 501–522. http://dx.doi.org/10.1007/BF02707510 SAS/ETS®9.2 user’s guide. 2010. SAS Institute Inc., Cary, NC, USA. Schmitz, A.; Helmberger, P. 1970. Factor mobility and international trade: the case of complementarity, The American Economic Review 60(4): 761–767. Seo, J. S.; Suh, C.-S. 2006. An analysis of home country trade effects of outward foreign direct invest- ment: the Korean experience with ASEAN, 1987–2002, ASEAN Economic Bulletin 23(2): 160–170. http://dx.doi.org/10.1355/AE23-2B UNCTAD. 2014. The least developed countries report [online], [cited 01 May 2016]. Available from Internet: http://unctad.org/en/PublicationsLibrary/ldc2014_en.pdf Anžela KOZLOVA. She is PhD student in Economics (Vilnius Gediminas Technical University, Lithuania). Her research interests are outward foreign direct investment. Algita MIEČINSKIENĖ. PhD in Economics (Vilnius Gediminas Technical University, Lithuania), an Associate Professor at the Department of Finance Engineering, a head of Finance Engineering Department at Vilnius Gediminas Technical University (Lithuania). She is an author of more than 50 research papers. Her research interests are business decisions in uncertain environment, multinational corporate economics and finance, effective pricing, foreign direct investment (greenfield investment, mergers and acquisitions) and finance engineering. http://osp.stat.gov.lt/temines-lenteles43 http://osp.stat.gov.lt/statistiniu-rodikliu-analize1 http://osp.stat.gov.lt/web/guest/statistiniu-rodikliu-analize?portletFormName=visualization&hash=6d9d5cee-014b-4e1c-9e41-bc40d749919d http://osp.stat.gov.lt/web/guest/statistiniu-rodikliu-analize?portletFormName=visualization&hash=6d9d5cee-014b-4e1c-9e41-bc40d749919d http://osp.stat.gov.lt/web/guest/statistiniu-rodikliu-analize?portletFormName=visualization&hash=6d9d5cee-014b-4e1c-9e41-bc40d749919d http://dx.doi.org/10.1007/BF02707510 http://muse.jhu.edu/journals/asean_economic_bulletin http://muse.jhu.edu/journals/asean_economic_bulletin/toc/ase23.2.html http://dx.doi.org/10.1355/AE23-2B http://unctad.org/en/PublicationsLibrary/ldc2014_en.pdf significance of demographic variables for targeting of internet advertisements Václav STŘÍTESKÝ1, Marek STŘÍTESKÝ2, Martin John David QUIGLEY3, David ŘÍHA4 The Demand for Professional Knowledge as a Key Factor of the Development of Outsourcing of Financial and Accounting Services in Poland Anna BAGIEŃSKA THE CONCEPT OF A SMART CITY IN URBAN MANAGEMENT Sławomira HAJDUK The concept of basic income: global experience and implementation possibilities in Lithuania Algimantas LAURINAVIČIUS1, Antanas LAURINAVIČIUS2 Julia SIDERSKA evaluation of the impact of the eu structural support on the competitiveness of lithuanian economics Ligita GASPARĖNIENĖ1, Rita REMEIKIENĖ2 The Concept of Products and Services Integration – Analysis of Scientific Publications Justyna KOZŁOWSKA Categories of supply chain performance indicators: an overview of approaches Dorota LEOŃCZUK Multivariate analysis of short and long-impact indicators for corporate bond market development Ieva ASTRAUSKAITĖ THE RESEARCH ON INTERFACE BETWEEN LITHUANIAN DIRECT INVESTMENT ABROAD AND FOREIGN TRADE FLOWS Anžela KOZLOVA1, Algita MIEČINSKIENĖ2