Acta Universitatis Danubius. Œconomica, Vol 13, No 1 (2017)
The Relationship between Spatial Interdependencies in the European Union and the Trade - II
Cătălin Angelo Ioan^{1}, Gina Ioan^{2}
Abstract: The article treats the links between exports of EU countries and relative distances between them. Mostly there are linear regressions equations that modeling the export relative to the spatial relations between states.
Keywords: graph; European Union; trade; export; import
JEL Classification: F21
1. Introduction
In the previous paper we analyzed the dependence of European Union countries imports on exports depending on their closeness.
Thus, after the construction of a graph of links between countries, we determined the minimum length between these roads, then we built a normalized matrix based on inverse distance (in the sense of graph theory and not actual distances). Considering the situation of global exports of those countries we multiplied (for each individual year) their values with the dependence degree of EU countries obtaining a virtual import value of each country. After this, we performed regression analysis in which we correlate these data with real data virtual obtaining in most cases, links expressing linear dependence of imports to exports of other countries. Finally, we compared the regression coefficients (with meanings of percentage) with actual percentages of UE-exports in each country commenting, finally, differences emerged.
In what follows, we will analyze the reverse dependence of exports on imports of other countries according to their closeness. All theoretical concepts and primary results on the degrees of connection matrix between countries are concretely explained in the first part of this article.
2. The Analysis of the Exports of EU Countries
In this section we shall analyze the relations between the import of EU countries and exports of each of them.
In Appendix A.1 and A.2 we have the tables of imports of European Union countries during 2004-2015.
Multiplying the matrix G with the values from tables A.1 and A.2, we find the tables A.3-A.6 in Appendix A.2.
Because not all imports from one country will be transferred to the EU reference country, we shall search if there is a linear dependence between real exports and computed exports (after the results from tables A.3-A.6).
In the case of Austria, from Appendix A.7 we can see that is a strong link between the two groups of indicators (R^{2}=0.9687), having finally:
EX_AT(t)=0.021IM_BE(t)+0.014IM_BG(t)+0.021IM_HR(t)+0.014IM_CY(t)+0.0419IM_CZ(t)+
0.021IM_DK(t)+0.0084IM_EE(t)+0.0105IM_FI(t)+0.021IM_FR(t)+0.0419IM_DE(t)+
0.021IM_EL(t)+0.0419IM_HU(t)+0.0105IM_IE(t)+0.0419IM_IT(t)+0.0105IM_LV(t)+
0.014IM_LT(t)+0.021IM_LU(t)+0.021IM_MT(t)+0.021IM_NL(t)+0.021IM_PL(t)+0.0105IM_PT(t)+0.021IM_RO(t)+0.0419IM_SK(t)+0.0419IM_SI(t)+0.014IM_ES(t)+0.014IM_SE(t)+0.014IM_UK(t)+15293.754
where
EX_ means real exports, IM_ means real imports, t – the
reference time and the abbreviations
for countries are the usual: Austria – AT, Belgium – BE,
Bulgaria – BG, Croatia – HR, Cyprus – CY, Czech
Republic – CZ, Denmark – DK, Estonia – EE, Finland
– FI, France – FR, Germany – DE, Greece – EL,
Hungary – HU, Ireland – IE, Italy – IT, Latvia –
LV, Lithuania – LT,
Luxembourg – LU, Malta –
MT, Netherlands – NL, Poland – PL, Portugal – PT,
Romania – RO, Slovakia – SK, Slovenia – SI, Spain –
ES, Sweden – SE, United Kingdom – UK.
A comparison of regression coefficients and percentages exports into studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 1) indicates that there are no large differences except Croatia (8.90% vs. 2.10%) and Slovenia (8.40% vs. 4.19%). Also, we can see that the real imports of EU-countries from Austria are in general below of those suggested by the regression equation which means that exports are below the potential offered by its geographic position.
The average distance between real data and those from the regression is: 1.32%.
Table 1. The correlation between the coefficients of regression and the real imports of EU-countries in Austria (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
- |
- |
Italy |
4.19% |
2.50% |
Belgium+Luxembourg |
4.20% |
0.64% |
Latvia |
1.05% |
1.20% |
Bulgaria |
1.40% |
2.70% |
Lithuania |
1.40% |
0.85% |
Croatia |
2.10% |
8.90% |
Malta |
2.10% |
0.53% |
Czech Republic |
4.19% |
3.10% |
Netherlands |
2.10% |
0.48% |
Denmark |
2.10% |
1.00% |
Poland |
2.10% |
1.80% |
Estonia |
0.84% |
0.74% |
Portugal |
1.05% |
0.48% |
Finland |
1.05% |
0.87% |
Romania |
2.10% |
3.90% |
France |
2.10% |
1.10% |
Slovakia |
4.19% |
2.90% |
Germany |
4.19% |
3.90% |
Slovenia |
4.19% |
8.40% |
Greece |
2.10% |
0.97% |
Spain |
1.40% |
0.71% |
Hungary |
4.19% |
6.10% |
Sweden |
1.40% |
1.20% |
Ireland |
1.05% |
1.60% |
United Kingdom |
1.40% |
0.73% |
Figure 1. The relationship between imports based on distances and the real imports in 2013 in Austria (in percent)
Because in the upper analysis we have Durbin Watson statistic d=0.8443 therefore a positive autocorrelation of errors for the limits of autocorrelation: (0,0.97) and - the autocorrelation coefficient of errors has value = 0.528085453 we shall make another regression analysis for the set of data: Exports-computed-new(t)=Exports-computed(t)-Exports-computed(t-1) and Imports-real-new(t)= Imports-real(t)-Imports-real(t-1) (table A.8). Finally, we obtain the equation of regression:
EX_AT(t)=0.5281EX_AT(t-1)+0.IM_AT(t)+0.IM_AT(t-1)+0.0228IM_BE(t)-0.0121IM_BE(t-1)+ 0.0152IM_BG(t)-0.0081IM_BG(t-1)+0.0228IM_HR(t)-0.0121IM_HR(t-1)+0.0152IM_CY(t)-0.0081IM_CY(t-1)+0.0457IM_CZ(t)-0.0241IM_CZ(t-1)+0.0228IM_DK(t)-0.0121IM_DK(t-1)+ 0.0091IM_EE(t)-0.0048IM_EE(t-1)+0.0114IM_FI(t)-0.006IM_FI(t-1)+0.0228IM_FR(t)-0.0121IM_FR(t-1)+0.0457IM_DE(t)-0.0241IM_DE(t-1)+0.0228IM_EL(t)-0.0121IM_EL(t-1)+ 0.0457IM_HU(t)-0.0241IM_HU(t-1)+0.0114IM_IE(t)-0.006IM_IE(t-1)+0.0457IM_IT(t)-0.0241IM_IT(t-1)+0.0114IM_LV(t)-0.006IM_LV(t-1)+0.0152IM_LT(t)-0.0081IM_LT(t-1)+ 0.0228IM_LU(t)-0.0121IM_LU(t-1)+0.0228IM_MT(t)-0.0121IM_MT(t-1)+0.0228IM_NL(t)-0.0121IM_NL(t-1)+0.0228IM_PL(t)-0.0121IM_PL(t-1)+0.0114IM_PT(t)-0.006IM_PT(t-1)+ 0.0228IM_RO(t)-0.0121IM_RO(t-1)+0.0457IM_SK(t)-0.0241IM_SK(t-1)+0.0457IM_SI(t)-0.0241IM_SI(t-1)+0.0152IM_ES(t)-0.0081IM_ES(t-1)+0.0152IM_SE(t)-0.0081IM_SE(t-1)+ 0.0152IM_UK(t)-0.0081IM_UK(t-1)+2372.02
In the case of Belgium, from Appendix A.4 we can see that is a strong link between the two groups of indicators (R^{2}=0.9846), having finally:
EX_BE(t)=0.0497IM_AT(t)+0.0248IM_BG(t)+0.0248IM_HR(t)+0.0248IM_CY(t)
+0.0497IM_CZ(t)+0.0497IM_DK(t)+0.0198IM_EE(t)+0.0248IM_FI(t)+0.0992IM_FR(t)+
0.0992IM_DE(t)+0.0331IM_EL(t)+0.0331IM_HU(t)+0.0497IM_IE(t)+0.0497IM_IT(t)+
0.0248IM_LV(t)+0.0331IM_LT(t)+0.0992IM_LU(t)+0.0331IM_MT(t)+0.0992IM_NL(t)+
0.0497IM_PL(t)+0.0331IM_PT(t)+0.0248IM_RO(t)+0.0331IM_SK(t)+0.0331IM_SI(t)+
0.0497IM_ES(t)+0.0331IM_SE(t)+0.0992IM_UK(t)+33128.7758
Also, in the case of Luxembourg, from Appendix A.5 we can see that practically is not a link between the two groups of indicators (R^{2}=0.0018) therefore we will immerse the data into those of Belgium.
A comparison of regression coefficients and percentages exports into studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 2) indicates that there are no large differences except Germany (5% vs. 9.92% - figure 2) and United Kingdom (5.2% vs. 9.92%) for which the imports are much below the distance. Also, we can see that the real imports of EU-countries from Belgium and Luxembourg are below of those suggested by the regression equation which means that imports are below the potential offered by its geographic position.
The average distance between real data and those from the regression is: 1.86 %.
Table 2.The correlation between the coefficients of regression and the real imports of EU-countries in Belgium+Luxembourg (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
4.97% |
1.90% |
Italy |
4.97% |
4.50% |
Belgium+Luxembourg |
- |
- |
Latvia |
2.48% |
1.70% |
Bulgaria |
2.48% |
2.00% |
Lithuania |
3.31% |
3.30% |
Croatia |
2.48% |
1.70% |
Malta |
3.31% |
1.30% |
Czech Republic |
4.97% |
2.10% |
Netherlands |
9.92% |
9.70% |
Denmark |
4.97% |
3.30% |
Poland |
4.97% |
2.70% |
Estonia |
1.98% |
1.50% |
Portugal |
3.31% |
2.20% |
Finland |
2.48% |
2.70% |
Romania |
2.48% |
2.30% |
France |
9.92% |
8.40% |
Slovakia |
3.31% |
1.20% |
Germany |
9.92% |
5.00% |
Slovenia |
3.31% |
1.70% |
Greece |
3.31% |
3.20% |
Spain |
4.97% |
2.90% |
Hungary |
3.31% |
2.30% |
Sweden |
3.31% |
4.30% |
Ireland |
4.97% |
2.20% |
United Kingdom |
9.92% |
5.20% |
Figure 2. The relationship between imports based on distances and the real imports in 2013 in Austria (in percent)
In the case of Bulgaria, from Appendix A.6 we can see that is a strong link between the two groups of indicators (R^{2}=0.8730), having finally:
EX_BG(t)=0.0108IM_AT(t)+0.0081IM_BE(t)+0.0108IM_HR(t)+0.0162IM_CY(t)+
0.0081IM_CZ(t)+0.0065IM_DK(t)+0.0046IM_EE(t)+0.0046IM_FI(t)+0.0108IM_FR(t)+
0.0081IM_DE(t)+0.0325IM_EL(t)+0.0162IM_HU(t)+0.0065IM_IE(t)+0.0162IM_IT(t)+
0.0054IM_LV(t)+0.0065IM_LT(t)+0.0081IM_LU(t)+0.0108IM_MT(t)+0.0065IM_NL(t)+
0.0081IM_PL(t)+0.0065IM_PT(t)+0.0325IM_RO(t)+0.0108IM_SK(t)+0.0108IM_SI(t)+
0.0081IM_ES(t)+0.0054IM_SE(t)+0.0081IM_UK(t)-22905.4187
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 3) indicates that there are no large differences, therefore we can see that the real imports of EU-countries from Bulgaria are closer to those suggested by the regression equation which means that imports depend preferential from the potential offered by its geographic position.
The average distance between real data and those from the regression is: 0.62%.
Table 3. The correlation between the coefficients of regression and the real imports of EU-countries in Bulgaria (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
1.08% |
0.32% |
Italy |
1.62% |
0.64% |
Belgium+Luxembourg |
1.62% |
0.21% |
Latvia |
0.54% |
0.19% |
Bulgaria |
- |
- |
Lithuania |
0.65% |
0.21% |
Croatia |
1.08% |
0.33% |
Malta |
1.08% |
0.19% |
Czech Republic |
0.81% |
0.23% |
Netherlands |
0.65% |
0.13% |
Denmark |
0.65% |
0.14% |
Poland |
0.81% |
0.26% |
Estonia |
0.46% |
0.15% |
Portugal |
0.65% |
0.28% |
Finland |
0.46% |
0.10% |
Romania |
3.25% |
2.70% |
France |
1.08% |
0.20% |
Slovakia |
1.08% |
0.25% |
Germany |
0.81% |
0.31% |
Slovenia |
1.08% |
0.42% |
Greece |
3.25% |
3.00% |
Spain |
0.81% |
0.21% |
Hungary |
1.62% |
0.36% |
Sweden |
0.54% |
0.09% |
Ireland |
0.65% |
0.07% |
United Kingdom |
0.81% |
0.10% |
Figure 3. The relationship between imports based on distances and the real imports in 2013 in Bulgaria (in percent)
In the case of Croatia, from Appendix A.7 we can see that is a strong link between the two groups of indicators (R^{2}=0.9170), having finally:
EX_HR(t)=0.0039IM_AT(t)+0.002IM_BE(t)+0.0026IM_BG(t)+0.002IM_CY(t)+0.0026IM_CZ(t)+
0.002IM_DK(t)+0.0013IM_EE(t)+0.0013IM_FI(t)+0.0026IM_FR(t)+0.0026IM_DE(t)+
0.0026IM_EL(t)+0.0078IM_HU(t)+0.0016IM_IE(t)+0.0039IM_IT(t)+0.0016IM_LV(t)+
0.002IM_LT(t)+0.002IM_LU(t)+0.0026IM_MT(t)+0.002IM_NL(t)+0.0026IM_PL(t)+
0.0016IM_PT(t)+0.0039IM_RO(t)+0.0039IM_SK(t)+0.0078IM_SI(t)+0.002IM_ES(t)+
0.0016IM_SE(t)+0.002IM_UK(t)-1510.5281
Let note that we have a small autoregression (d=0.7535) and P-Value for the Intercept is 0.16. If we shall try to eliminate the autoregression we shall find again d=0.5860 (much worth) and a P-Value for the Intercept 0.61. Therefore, we shall let the first regression which is much better than the second.
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 4) indicates that there are no large differences except Slovenia (figure 4) which is absolutely normal because of their former membership to Yugoslavia. Also, we can see that the real imports of EU-countries from Croatia are closer to those suggested by the regression equation which means that imports depend preferential from the potential offered by its geographic position.
The average distance between real data and those from the regression is: 0.30 %.
Table 4. The correlation between the coefficients of regression and the real imports of EU-countries in Croatia (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
0.39% |
0.44% |
Italy |
0.39% |
0.36% |
Belgium+Luxembourg |
0.40% |
0.05% |
Latvia |
0.16% |
0.04% |
Bulgaria |
0.26% |
0.15% |
Lithuania |
0.20% |
0.03% |
Croatia |
- |
- |
Malta |
0.26% |
1.10% |
Czech Republic |
0.26% |
0.09% |
Netherlands |
0.20% |
0.04% |
Denmark |
0.20% |
0.04% |
Poland |
0.26% |
0.07% |
Estonia |
0.13% |
0.22% |
Portugal |
0.16% |
0.02% |
Finland |
0.13% |
0.04% |
Romania |
0.39% |
0.16% |
France |
0.26% |
0.03% |
Slovakia |
0.39% |
0.19% |
Germany |
0.26% |
0.11% |
Slovenia |
0.78% |
4.00% |
Greece |
0.26% |
0.20% |
Spain |
0.20% |
0.02% |
Hungary |
0.78% |
0.28% |
Sweden |
0.16% |
0.04% |
Ireland |
0.16% |
0.01% |
United Kingdom |
0.20% |
0.04% |
Figure 4. The relationship between imports based on distances and the real imports in 2013 in Croatia (in percent)
In the case of Cyprus, from Appendix A.8 we can see that is a weak link between the two groups of indicators (R^{2}=0.6655), having finally:
EX_CY(t)=0.0005IM_AT(t)+0.0004IM_BE(t)+0.0007IM_BG(t)+0.0004IM_HR(t)+
0.0004IM_CZ(t)+0.0003IM_DK(t)+0.0002IM_EE(t)+0.0002IM_FI(t)+0.0005IM_FR(t)+
0.0004IM_DE(t)+0.0014IM_EL(t)+0.0004IM_HU(t)+0.0003IM_IE(t)+0.0007IM_IT(t)+
0.0002IM_LV(t)+0.0002IM_LT(t)+0.0004IM_LU(t)+0.0005IM_MT(t)+0.0003IM_NL(t)+
0.0003IM_PL(t)+0.0003IM_PT(t)+0.0005IM_RO(t)+0.0004IM_SK(t)+0.0005IM_SI(t)+
0.0004IM_ES(t)+0.0002IM_SE(t)+0.0004IM_UK(t)-457.8204
Let note that we have a P-Value for the Intercept 0.25 therefore we will reject the null hypothesis with a probability almost 0.75.
In the case of Czech Republic, from Appendix A.9 we can see that is a strong link between the two groups of indicators (R^{2}=0.9308), having finally:
EX_CZ(t)=0.0804IM_AT(t)+0.0402IM_BE(t)+0.02IM_BG(t)+0.0268IM_HR(t)+0.02IM_CY(t)+
0.0402IM_DK(t)+0.02IM_EE(t)+0.02IM_FI(t)+0.0402IM_FR(t)+0.0804IM_DE(t)+0.0268IM_EL(t)+0.0402IM_HU(t)+0.02IM_IE(t)+0.0402IM_IT(t)+0.0268IM_LV(t)+0.0402IM_LT(t)+
0.0402IM_LU(t)+0.0268IM_MT(t)+0.0402IM_NL(t)+0.0804IM_PL(t)+0.02IM_PT(t)+
0.0268IM_RO(t)+0.0804IM_SK(t)+0.0402IM_SI(t)+0.0268IM_ES(t)+0.0268IM_SE(t)+
0.0268IM_UK(t)-86039.0944
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 5) indicates that there are many differences (real vs. predicted imports) like Austria (3.90% vs. 8.04%), Belgium+Luxembourg (0.98% vs. 8.04%), Germany (3.90% vs. 8.04%), Poland (3.90% vs. 8.04%) and Slovakia (14% vs. 8.04%) in the last case being absolutely normal because of their former membership to Czechoslovakia.
Also, we can see that the real imports of EU-countries from Czech Republic are under to those suggested by the regression equation which means that imports not use the potential offered by its geographic position.
The average distance between real data and those from the regression is: 2.18%.
Table 5. The correlation between the coefficients of regression and the real imports of EU-countries in Czech Republic (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
8.04% |
3.90% |
Italy |
4.02% |
1.20% |
Belgium+Luxembourg |
8.04% |
0.98% |
Latvia |
2.68% |
1.50% |
Bulgaria |
2.00% |
2.10% |
Lithuania |
4.02% |
1.60% |
Croatia |
2.68% |
1.90% |
Malta |
2.68% |
0.21% |
Czech Republic |
- |
- |
Netherlands |
4.02% |
0.99% |
Denmark |
4.02% |
1.40% |
Poland |
8.04% |
3.90% |
Estonia |
2.00% |
1.10% |
Portugal |
2.00% |
0.61% |
Finland |
2.00% |
1.10% |
Romania |
2.68% |
2.80% |
France |
4.02% |
1.20% |
Slovakia |
8.04% |
14.00% |
Germany |
8.04% |
3.90% |
Slovenia |
4.02% |
2.30% |
Greece |
2.68% |
0.47% |
Spain |
2.68% |
1.10% |
Hungary |
4.02% |
4.00% |
Sweden |
2.68% |
1.30% |
Ireland |
2.00% |
0.72% |
United Kingdom |
2.68% |
1.20% |
Figure 5. The relationship between imports based on distances and the real imports in 2013 in Czech Republic (in percent)
In the case of Denmark, from Appendix A.10 we can see that is a strong link between the two groups of indicators (R^{2}=0.9581), having:
EX_DK(t)=0.0117IM_AT(t)+0.0117IM_BE(t)+0.0047IM_BG(t)+0.0059IM_HR(t)+
0.0047IM_CY(t)+0.0117IM_CZ(t)+0.0078IM_EE(t)+0.0117IM_FI(t)+0.0117IM_FR(t)+
0.0235IM_DE(t)+0.0059IM_EL(t)+0.0078IM_HU(t)+0.0059IM_IE(t)+0.0078IM_IT(t)+
0.0059IM_LV(t)+0.0078IM_LT(t)+0.0117IM_LU(t)+0.0059IM_MT(t)+0.0117IM_NL(t)+
0.0117IM_PL(t)+0.0059IM_PT(t)+0.0059IM_RO(t)+0.0078IM_SK(t)+0.0078IM_SI(t)+
0.0078IM_ES(t)+0.0235IM_SE(t)+0.0078IM_UK(t)+25237.4467
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 6) indicates that there are no large differences (real vs. predicted imports) except Sweden (8% vs. 0.78% - figure 6) which is absolutely normal as a consequence of commercial traditions that have bound these countries.
Unlike the other countries analyzed so far, one can see that in general, real imports are close to those provided by regression analysis, which shows a strong trade policy, taking into account the dependence on proximity.
The average distance between real data and those from the regression is: 0.78%.
Table 6. The correlation between the coefficients of regression and the real imports of EU-countries in Denmark (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
1.17% |
0.41% |
Italy |
0.78% |
0.57% |
Belgium+Luxembourg |
2.34% |
0.36% |
Latvia |
0.59% |
2.00% |
Bulgaria |
0.47% |
0.36% |
Lithuania |
0.78% |
1.70% |
Croatia |
0.59% |
1.20% |
Malta |
0.59% |
0.60% |
Czech Republic |
1.17% |
0.61% |
Netherlands |
1.17% |
0.94% |
Denmark |
- |
- |
Poland |
1.17% |
1.20% |
Estonia |
0.78% |
1.20% |
Portugal |
0.59% |
0.43% |
Finland |
1.17% |
3.20% |
Romania |
0.59% |
0.83% |
France |
1.17% |
0.50% |
Slovakia |
0.78% |
0.36% |
Germany |
2.35% |
1.20% |
Slovenia |
0.78% |
0.31% |
Greece |
0.59% |
0.94% |
Spain |
0.78% |
0.54% |
Hungary |
0.78% |
0.68% |
Sweden |
2.35% |
8.00% |
Ireland |
0.59% |
1.40% |
United Kingdom |
0.78% |
1.40% |
Figure 6. The relationship between imports based on distances and the real imports in 2013 in Denmark (in percent)
In the case of Estonia, from Appendix A.11 we can see that is a strong link between the two groups of indicators (R^{2}=0.9040), having:
EX_EE(t)=0.004IM_AT(t)+0.004IM_BE(t)+0.0028IM_BG(t)+0.0033IM_HR(t)+0.0025IM_CY(t)+
0.0049IM_CZ(t)+0.0066IM_DK(t)+0.0198IM_FI(t)+0.004IM_FR(t)+0.0049IM_DE(t)+
0.0028IM_EL(t)+0.004IM_HU(t)+0.0028IM_IE(t)+0.0033IM_IT(t)+0.0198IM_LV(t)+
0.0099IM_LT(t)+0.004IM_LU(t)+0.0028IM_MT(t)+0.004IM_NL(t)+0.0066IM_PL(t)+
0.0028IM_PT(t)+0.0033IM_RO(t)+0.0049IM_SK(t)+0.0033IM_SI(t)+0.0033IM_ES(t)+
0.0099IM_SE(t)+0.0033IM_UK(t)-9027.2563
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 7) indicates that there are no large differences (real vs. predicted imports) except former Soviet Union countries – Latvia (6.70% vs. 1.98%) and Lithuania (2.50% vs. 0.99%) which is absolutely normal as a consequence of commercial traditions that have bound these countries.
Let note that in general, real imports were close, but under to those provided by regression analysis, which shows a trade policy depending on proximity of the EU-countries but not exploring all the possibilities of the minimal distances recovery.
The average distance between real data and those from the regression is: 0.57%.
Table 7. The correlation between the coefficients of regression andthe real imports of EU-countries in Estonia (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
0.40% |
0.03% |
Italy |
0.33% |
0.03% |
Belgium+Luxembourg |
0.80% |
0.08% |
Latvia |
1.98% |
6.70% |
Bulgaria |
0.28% |
0.09% |
Lithuania |
0.99% |
2.50% |
Croatia |
0.33% |
0.07% |
Malta |
0.28% |
0.06% |
Czech Republic |
0.49% |
0.05% |
Netherlands |
0.40% |
0.07% |
Denmark |
0.66% |
0.43% |
Poland |
0.66% |
0.10% |
Estonia |
- |
- |
Portugal |
0.28% |
0.04% |
Finland |
1.98% |
2.80% |
Romania |
0.33% |
0.02% |
France |
0.40% |
0.05% |
Slovakia |
0.49% |
0.07% |
Germany |
0.49% |
0.06% |
Slovenia |
0.33% |
0.04% |
Greece |
0.28% |
0.02% |
Spain |
0.33% |
0.03% |
Hungary |
0.40% |
0.03% |
Sweden |
0.99% |
1.70% |
Ireland |
0.28% |
0.04% |
United Kingdom |
0.33% |
0.07% |
Figure 7. The relationship between imports based on distances and the real imports in 2013 in Estonia (in percent)
In the case of Finland, from Appendix A.12 we can see that is a very weak link between the two groups of indicators (R^{2}=0.1840), having:
EX_FI(t)=0.0042IM_AT(t)+0.0042IM_BE(t)+0.0024IM_BG(t)+0.0028IM_HR(t)+0.0024IM_CY(t)+0.0042IM_CZ(t)+0.0084IM_DK(t)+0.0169IM_EE(t)+0.0042IM_FR(t)+0.0056IM_DE(t)+
0.0028IM_EL(t)+0.0034IM_HU(t)+0.0028IM_IE(t)+0.0034IM_IT(t)+0.0084IM_LV(t)+
0.0056IM_LT(t)+0.0042IM_LU(t)+0.0028IM_MT(t)+0.0042IM_NL(t)+0.0042IM_PL(t)+
0.0028IM_PT(t)+0.0028IM_RO(t)+0.0034IM_SK(t)+0.0034IM_SI(t)+0.0034IM_ES(t)+
0.0169IM_SE(t)+0.0034IM_UK(t)+37525.6209
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 8) indicates that there are no large differences (real vs. predicted imports) except Estonia (9.60% vs. 1.69%), Latvia (4.30% vs. 0.84%), Lithuania (1.90% vs. 0.56%) and Sweden (5.60% vs. 1.69%).
In general, real imports were close which shows a trade policy depending on proximity of the EU-countries.
The average distance between real data and those from the regression is: 0.76%.
Table 8. The correlation between the coefficients of regression and the real imports of EU-countries in Finland (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
0.42% |
0.32% |
Italy |
0.34% |
0.39% |
Belgium+Luxembourg |
0.84% |
0.59% |
Latvia |
0.84% |
4.30% |
Bulgaria |
0.24% |
0.19% |
Lithuania |
0.56% |
1.90% |
Croatia |
0.28% |
0.22% |
Malta |
0.28% |
0.06% |
Czech Republic |
0.42% |
0.27% |
Netherlands |
0.42% |
0.82% |
Denmark |
0.84% |
1.60% |
Poland |
0.42% |
0.86% |
Estonia |
1.69% |
9.60% |
Portugal |
0.28% |
0.28% |
Finland |
- |
- |
Romania |
0.28% |
0.28% |
France |
0.42% |
0.41% |
Slovakia |
0.34% |
0.18% |
Germany |
0.56% |
0.64% |
Slovenia |
0.34% |
0.27% |
Greece |
0.28% |
0.29% |
Spain |
0.34% |
0.34% |
Hungary |
0.34% |
0.29% |
Sweden |
1.69% |
5.60% |
Ireland |
0.28% |
0.22% |
United Kingdom |
0.34% |
0.62% |
Figure 8. The relationship between imports based on distances and the real imports in 2013 in Finland (in percent)
In the case of France, from Appendix A.13 we can see that is a strong link between the two groups of indicators (R^{2}=0.9311), having:
EX_FR(t)=0.0444IM_AT(t)+0.0889IM_BE(t)+0.0296IM_BG(t)+0.0296IM_HR(t)+
0.0296IM_CY(t)+0.0444IM_CZ(t)+0.0444IM_DK(t)+0.0178IM_EE(t)+0.0222IM_FI(t)+
0.0889IM_DE(t)+0.0444IM_EL(t)+0.0296IM_HU(t)+0.0444IM_IE(t)+0.0889IM_IT(t)+
0.0222IM_LV(t)+0.0296IM_LT(t)+0.0889IM_LU(t)+0.0444IM_MT(t)+0.0444IM_NL(t)+
0.0444IM_PL(t)+0.0444IM_PT(t)+0.0222IM_RO(t)+0.0296IM_SK(t)+0.0444IM_SI(t)+
0.0889IM_ES(t)+0.0296IM_SE(t)+0.0889IM_UK(t)+158856.3841
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 9) indicates that there are no large differences (real vs. predicted imports) except Belgium+Luxembourg – under the distance between them (11% vs. 17.78%) and, on the other side, Romania (5.80% vs. 2.22%) and Portugal (6.30% vs. 4.44%) over the coefficients of regression.
Let note that in general, real imports were close to those provided by regression analysis.
The average distance between real data and those from the regression is: 1.22 %.
Table 9. The correlation between the coefficients of regression and the real imports of EU-countries in France (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
4.44% |
2.80% |
Italy |
8.89% |
8.30% |
Belgium+Luxembourg |
17.78% |
11.00% |
Latvia |
2.22% |
1.80% |
Bulgaria |
2.96% |
2.90% |
Lithuania |
2.96% |
2.70% |
Croatia |
2.96% |
2.20% |
Malta |
4.44% |
7.10% |
Czech Republic |
4.44% |
3.30% |
Netherlands |
4.44% |
4.20% |
Denmark |
4.44% |
3.20% |
Poland |
4.44% |
3.90% |
Estonia |
1.78% |
2.10% |
Portugal |
4.44% |
6.30% |
Finland |
2.22% |
3.20% |
Romania |
2.22% |
5.80% |
France |
- |
- |
Slovakia |
2.96% |
3.00% |
Germany |
8.89% |
7.20% |
Slovenia |
4.44% |
4.20% |
Greece |
4.44% |
4.90% |
Spain |
8.89% |
10.00% |
Hungary |
2.96% |
4.00% |
Sweden |
2.96% |
4.20% |
Ireland |
4.44% |
4.20% |
United Kingdom |
8.89% |
6.20% |
Figure 9
In the case of Germany, from Appendix A.14 we can see that is a strong link between the two groups of indicators (R^{2}=0.9681). The P-Value Analysis reveals for Intercept a great value (0.2002) which indicates a weak evidence against the null hypothesis. In fact, assuming the threshold of 79% we obtain the regression in the table A.19. Finally, we have:
EX_DE(t)=0.4463IM_AT(t)+0.4463IM_BE(t)+0.1114IM_BG(t)+0.1486IM_HR(t)+
0.1114IM_CY(t)+0.4463IM_CZ(t)+0.4463IM_DK(t)+0.1114IM_EE(t)+0.1486IM_FI(t)+
0.4463IM_FR(t)+0.1486IM_EL(t)+0.2228IM_HU(t)+0.1486IM_IE(t)+0.2228IM_IT(t)+
0.1486IM_LV(t)+0.2228IM_LT(t)+0.4463IM_LU(t)+0.1486IM_MT(t)+0.4463IM_NL(t)+
0.4463IM_PL(t)+0.1486IM_PT(t)+0.1486IM_RO(t)+0.2228IM_SK(t)+0.2228IM_SI(t)+
0.2228IM_ES(t)+0.2228IM_SE(t)+0.2228IM_UK(t)-83740.245
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 10) indicates that there are many differences (real vs. predicted imports) between countries - Belgium+Luxembourg with a real percent of imports of 14% instead 89.26% (after regression), Czech Republic (26% vs. 44.63%), Denmark with 20% vs. 44.63%, France – 18% vs. 44.63%, Netherlands – 15% vs. 44.63%, Poland – 23% vs. 44.63%. We can easily see that these difference, maybe except Poland, are encountered in the case of the very developed countries from the European Union, which have themselves a strong import.
Let note that in general, real imports were strong under to those provided by regression analysis, even Germany is the main engine of UE.
The average distance between real data and those from the regression is very high: 11.44 %.
Table 10. The correlation between the coefficients of regression and the real imports of EU-countries in Germany (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
44.63% |
38.00% |
Italy |
22.28% |
15.00% |
Belgium+Luxembourg |
89.26% |
14.00% |
Latvia |
14.86% |
10.00% |
Bulgaria |
11.14% |
10.00% |
Lithuania |
22.28% |
10.00% |
Croatia |
14.86% |
14.00% |
Malta |
14.86% |
4.20% |
Czech Republic |
44.63% |
26.00% |
Netherlands |
44.63% |
15.00% |
Denmark |
44.63% |
20.00% |
Poland |
44.63% |
23.00% |
Estonia |
11.14% |
9.30% |
Portugal |
14.86% |
11.00% |
Finland |
14.86% |
13.00% |
Romania |
14.86% |
18.00% |
France |
44.63% |
18.00% |
Slovakia |
22.28% |
16.00% |
Germany |
- |
- |
Slovenia |
22.28% |
17.00% |
Greece |
14.86% |
10.00% |
Spain |
22.28% |
12.00% |
Hungary |
22.28% |
24.00% |
Sweden |
22.28% |
18.00% |
Ireland |
14.86% |
9.20% |
United Kingdom |
22.28% |
14.00% |
Figure 10
In the case of Greece, from Appendix A.15 we can see that is a strong link between the two groups of indicators (R^{2}=0.8716). We have:
EX_EL(t)=0.0114IM_AT(t)+0.0076IM_BE(t)+0.0228IM_BG(t)+0.0076IM_HR(t)+
0.0228IM_CY(t)+0.0076IM_CZ(t)+0.0057IM_DK(t)+0.0033IM_EE(t)+0.0038IM_FI(t)+
0.0114IM_FR(t)+0.0076IM_DE(t)+0.0076IM_HU(t)+0.0057IM_IE(t)+0.0228IM_IT(t)+
0.0038IM_LV(t)+0.0046IM_LT(t)+0.0076IM_LU(t)+0.0114IM_MT(t)+0.0057IM_NL(t)+
0.0057IM_PL(t)+0.0057IM_PT(t)+0.0114IM_RO(t)+0.0076IM_SK(t)+0.0114IM_SI(t)+
0.0076IM_ES(t)+0.0046IM_SE(t)+0.0076IM_UK(t)-15317.9389
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 11) indicates that there are only one major difference (real vs. predicted imports) between countries – Bulgaria were real imports are 5.20% versus 2.28% from the regression.
Let note that in general, real imports were under to those provided by regression analysis, therefore the export of Greece not exploit all the opportunities generated by the distances.
The average distance between real data and those from the regression is low: 0.64 %.
Table 11. The correlation between the coefficients of regression and the real imports of EU-countries in Greece (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
1.14% |
0.14% |
Italy |
2.28% |
0.64% |
Belgium+Luxembourg |
1.52% |
0.09% |
Latvia |
0.38% |
0.11% |
Bulgaria |
2.28% |
5.20% |
Lithuania |
0.46% |
0.12% |
Croatia |
0.76% |
0.41% |
Malta |
1.14% |
1.90% |
Czech Republic |
0.76% |
0.16% |
Netherlands |
0.57% |
0.12% |
Denmark |
0.57% |
0.18% |
Poland |
0.57% |
0.20% |
Estonia |
0.33% |
0.11% |
Portugal |
0.57% |
0.24% |
Finland |
0.38% |
0.12% |
Romania |
1.14% |
1.10% |
France |
1.14% |
0.13% |
Slovakia |
0.76% |
0.12% |
Germany |
0.76% |
0.20% |
Slovenia |
1.14% |
0.65% |
Greece |
- |
- |
Spain |
0.76% |
0.34% |
Hungary |
0.76% |
0.10% |
Sweden |
0.46% |
0.16% |
Ireland |
0.57% |
0.08% |
United Kingdom |
0.76% |
0.19% |
Figure 11
In the case of Hungary, from Appendix A.16 we can see that is a strong link between the two groups of indicators (R^{2}=0.9758). The P-Value Analysis reveals low values under 0.0003 which indicates a very strong evidence against the null hypothesis. Therefore, finally, we have:
EX_HU(t)=0.0583IM_AT(t)+0.0194IM_BE(t)+0.0291IM_BG(t)+0.0583IM_HR(t)+
0.0146IM_CY(t)+0.0291IM_CZ(t)+0.0194IM_DK(t)+0.0117IM_EE(t)+0.0117IM_FI(t)+
0.0194IM_FR(t)+0.0291IM_DE(t)+0.0194IM_EL(t)+0.0117IM_IE(t)+0.0291IM_IT(t)+
0.0146IM_LV(t)+0.0194IM_LT(t)+0.0194IM_LU(t)+0.0194IM_MT(t)+0.0194IM_NL(t)+
0.0291IM_PL(t)+0.0117IM_PT(t)+0.0583IM_RO(t)+0.0583IM_SK(t)+0.0583IM_SI(t)+
0.0146IM_ES(t)+0.0146IM_SE(t)+0.0146IM_UK(t)-25082.8642
A comparison of regression coefficients and percentages imports from studied countries (Source: http://atlas.media.mit.edu/en/profile/country - column Real in Table 12) indicates that there are not great differences (real vs. predicted imports) between countries, except Romania with real imports – 8.10% versus 5.83% after regression analysis. We can conclude that exports of Hungary are directed by territorial proximity criterion.
The average distance between real data and those from the regression is: 1.13%.
Table 12. The correlation between the coefficients of regression and the real imports of EU-countries in Hungary (in percent) in 2013
Country |
Regression |
Real |
Country |
Regression |
Real |
Austria |
5.83% |
2.90% |
Italy |
2.91% |
1.10% |
Belgium+Luxembourg |
3.88% |
0.40% |
Latvia |
1.46% |
1.10% |
Bulgaria |
2.91% |
2.90% |
Lithuania |
1.94% |
0.79% |
Croatia |
5.83% |
6.00% |
Malta |
1.94% |
0.12% |
Czech Republic |
2.91%</ |