58EXPORT OF AGRICULTURAL PRODUCTS FROM SERBIA TO THE EU - PANEL GRAVITY MODEL
Vladimir Ristanovic1, Aleksandra Tosovic Stevanovic2 Corresponding author E-mail: vmristanovic@gmail.com
A R T I C L E I N F O
Review Article
Received: 28 January 2022
Accepted: 01 February 2022
doi:10.5937/ekoPolj2201257R
UDC 339.564:338.43(497.11) (4-672EU)
Keywords:
agro products, export, panel gravity model, Serbia, EU
JEL: C33, F13, F15, Q17
A B S T R A C T
This paper will analyse the export of agricultural products of Serbia to the EU during the period from 2001 to 2017. A panel gravity model was used to assess the effects of trade flows. The main advantage of the gravity model is the application of basic indicators of the economy and the ability to evaluate panel series. The obtained results show that the size of the economy, measured by gross domestic product, the size of the market of foreign trade partners, measured by population, and the distance between trading partner countries, have measurable effects on the export of agricultural products from Serbia to the EU. Using a simple econometric model, we analysed the effects of Serbia's international trade and noticed that there is significant room for improving the exchange between Serbia and EU members. A multi-variable model would provide more information to trade policy makers.
© 2022 EA. All rights reserved.
Introduction
Economic integration has been attracting public attention for decades. This is particularly evident at the regional level, with the escalation of regional integration agreements (RIAs) from free trade areas (FTAs) to customs unions (CUs), all the way to monetary union (MU) such as the EU. For small economies, this type of international cooperation can be of great importance.
For Serbia, a small and open economy, an important development component is foreign trade relations, especially those related to the export of agricultural and food products. Agriculture is the core of the economy and the engine of rural development, but also an important component of the EU accession process. Trade liberalization and the ever-widening global market for agricultural and food products are a chance to improve the export range of agricultural products from Serbia. At the same time, market opening
1 Vladimir Ristanovic, PhD, Associate Researcher, Institute of European Studies, Trg Nikole Pasica 11, Beograd, Phone: +381646454415, E-mail: vmristanovic@gmail.com, ORCID ID (https://orcid.org/0000-0002-2957-3465)
2 Aleksandra Tosovic Stevanovic, PhD, Associate Professor, Faculty of Business Economics and Entrepreneurship, Metropolita Petra 8, Belgrade, Phone: +381641106527, E-mail: aleksandra. tosovic.stevanovic@gmail.com, ORCID ID (https://orcid.org/0000-0003-0281-0579)
can pose a serious threat to certain branches of agriculture, so it is necessary to timely adjust the production structures of the agricultural sector.
Therefore, it is important to consider all relevant parameters and key factors related to agricultural products that are an important topic of international trade, whose characteristics in terms of price and quality are constantly being examined in an open economy.
In this analysis, a gravity model was used to estimate the effects oftrade flows of agricultural products. The gravity model was estimated on the basis of mutual data exchange between Serbia and 28 EU member states. The emphasis is not on the overall trade potentials, but on the regional orientation of the export of Serbian agricultural products.
Serbian agrarian products export
For decades now, Serbia has had the largest volume of exchange with the EU member states. In 2018, the EU was a key trade partner of Serbia with a share of 67% in its total exports and over 60% in total imports. Exports of Serbian products to the EU in 2018 amounted to 10.9 billion euros.
Among the EU member states, which are traditionally represented with the largest share in foreign trade with Serbia, in 2018 are Italy (12.2%), Germany (11.9%) and Romania (5.9%). In addition to EU members, Russia, China and Bosnia and Herzegovina are at the very top of Serbia's most important trade partners (SORS, 2019).
Figure 1. EU share in total export and agro-food export 2008 - 2017
Source: Authors' calculations
In the observed period, Serbia was mostly focused on the trade with the EU countries, both in total and for agricultural products. The EU market is of great importance for Serbia, especially for agricultural products, where it sells almost half of its agricultural products annually. Having in mind the growing openness of Serbia, high dependence on imports and the deficit in foreign trade, the agricultural sector is becoming a significant factor in balancing the overall trade balance (Bozic and Nikolic, 2016).
According to SORS (2019), the export of agricultural products from Serbia has more than doubled in the last decade, with almost 1.3 billion euros in 2018. Producers of cereals, industrial plants and fruits gained profit, and vegetable producers were at a loss, while cattle breeders had almost identical results as in 2017.
Food production is an area in which Serbia has significant export potential, however, as the comparative advantage is not sufficient in itself, domestic export potential of food production is not adequately used due to low competitiveness of domestic food companies, but also the products themselves (Jovovic and Jovovic, 2018). In order to encourage and increase the level of export of agricultural products to the international market, a gravity model will be presented to assess the effects of trade flows so as to try to understand the decisive factors for improving Serbia's trade with EU member states.
Materials and methods
The gravity model in its basic form, in the literature, has a significant application in the process of explaining bilateral trade. There are numerous research papers (Smarzynska, 2001; Eita and Jordaan, 2007; Viorica, 2012; Waheed and Abbas, 2015; Bialynicka-Birula, 2015; Ristanovic et al, 2017; Ranilovic, 2017; Ristanovic et al, 2019) that explain the effects different economic determinants have (economy size, commodity prices, foreign direct investment, exports, foreign exchange reserves, population, exchange rate, etc.), but GDP, population and distance between countries are the ones which dominate (Oguledo and MacPhee 1994; Eita and Jordaan 2007; Bergstrand 1985 and 1989; Porojan, 2000; Chan-Hyun 2001; Smarzynska, 2001; Sevela, 2002). According to the theory of the gravitational model, exports are positively correlated with the degree of economic growth, negatively correlated with the population of the exporting country, and negatively correlated with the distance between the two economies.
Within the gravity model, the aggregate size of GDP reflects the size or economic strength of the economy, the number of inhabitants in the model represents the size of the market, while distance is a substitute for trade barriers (transport costs in international trade, export / import tariffs, dumping prices, etc.). Through the introduction of a set of three regional variables into the model, it is possible to simultaneously test the effects of grouping on intra-union, non-union, and export trade. Such estimates require the use of panel data to verify potentially inconspicuous factors that are specific to each pair of countries, which will have an impact on trade between them (Trotignon, 2010). In order to examine the individual characteristics of the countries, which participate in the analysis of trade flows, and through which we want to see the mutual trade relations, artificial variables are included in the model.
The influence of specific factors in the gravity model is examined by regression equations with the help of panel series. The advantage of panel series is that they provide the possibility of simultaneous analysis of both comparative data (N) and time series data (T). This increases the sample size (NT) and increases the amount of information from a limited number of observations (sample). The larger the sample, the greater the efficiency of model estimates, at the same time the greater the degree of variability and the greater the degree of freedom. Such a model has less correlation of explanatory variables. Also, the gravity panel model allows simultaneous analysis of both export structures and changes in exports over time. The evaluated results of the model should show the relationship between the size of the economy, the market and the distance, on the one hand, and the export capacities, on the other hand.
All data used to estimate the model parameters come from official sources. The definitions of the model variables and data sources are shown in Table 1. The analysis of trade exchange between Serbia and 28 EU member states covered the period from 2001 to 2017.
Table 1. Defining variables
Variable Definition Data Source
exp Export in current dollars Eurostat and Trade statistics for international business development
GDP Gross Domestic Product of the domestic country in current dollars Eurostat and Trade statistics for international business development
GDP* Gross Domestic Product of the foreign country in current dollars Eurostat and Trade statistics for international business development
POP Population of the domestic country World Bank annual statistics
POP* Population of the foreign country World Bank annual statistics
distance Distance in kilometres (represents the distance between the capitals) CEPII - le Centre d'études prospectives et d'informations internationales
border Dummy variables, 0 is the one that takes the value 1 if the countries i and j have a common boundary, otherwise it is 0
language Dummy variables, 0 is the one that takes the value 1 if the countries i and j have a common boundary, otherwise it is 0
We assessed the effects of economic development, market size and distance (distance between countries) on the volume of exports of Serbian products to the EU using the gravity model. In addition to these explanatory variables, artificial variables are included in the model. Thus, the extended gravity panel model was tested and evaluated through the impact on exports and from the aspect of common border and common language. To evaluate the model variables, we used two random effect models (RE) and a fixed effect model (FE). In the case of a random effect model, the regression parameters with explanatory variables are invariant, while the random error, i.e. the random variable of the model, reflects variations by units of observation and over time. In the fixed effect model,
the random error u has a normal distribution, with zero mean and constant variance,
ijt ' '
while the explanatory variables are non-stochastic and error-independent. The choice of the model by which the variables will be evaluated are given by the Hausman test. It shows which model, the random effect model or the fixed effect model, will give the best results when testing and evaluating the coefficients with the model variables. Descriptive statistics show that the model contains 280 observations [N = 28; T = 10]. We evaluated the regression model through the statistical software Stata S / E, version 13.0.
Method of Research - Gravity Model
The gravity model was first introduced to economic analysis during the 1970s. It all started with the modeling of trade flows by Tinbergen (1962) and Poyhonen (1963), and the increasing application to the field of international trade by Linnemman (1966). Aitken (1973) shows that trade between EEC members increased significantly with their integration. In the following decades, the application of the gravity model spread, especially during the 1990s, when the exchange potential of Central and Eastern European countries, on the one hand, and developed Western European countries, on the other hand, was assessed (eg Hamilton and Winters, 1992; Baldwin, 1994; Egger and Pfaffermayr, 2003; Bussiere et al. 2005). The differences in results between authors arise not only due to different periods of analysis, but also due to the presence or absence of explanatory variables of the model.
According to the gravity model, potential foreign trade between the two countries is determined by the following determinants of bilateral flows:
1. Demand for imports from the importing country (as a rule, is directly proportional to GDP, and inversely proportional to population),
2. Supply of the exporting country (is represented by the size of GDP and the size of the country - the degree of openness of the economy),
3. Trade barriers: natural constraints, such as transaction costs and transportation costs, other constraints such as, for example, customs and the like.
The Gravity Model has the following form:
% = ft Yj^POP^POP^DI^^F^^ (1)
where generally speaking Xijt shows the total export of the economy i to the economy j in year t; Yi (Yj) reflects the GDP of the economy i and the economy j in year t; POPi (POPj) is the population of economy i and economy j in year t; DISTij is a measure of the distance between major economic centers; Fij represents any other factors (variables) within the model. The random error of the model is marked with a v In the general case, a random error consists of three components: individual effects fa ), time effects (X,,) and the remainder of the random error (u ).
% = ft y^^MV*™^ (2)
where two artificial variables - common border (Bord) and common language (Lang) -
were introduced. This model basically shows the existence of a linear relationship. The values of some variables vary by country and time, while the values of others vary by country, but are constant over time.
The gravity model uses cross section data, which is a limiting factor because there is a problem of choosing a representative year. Using a panel model solves this limitation. Namely, the gravity panel model can evaluate bilateral exchange in a certain period of time.
As a rule, when using the gravity model to estimate export flows in bilateral trade, the estimated coefficients with variable GDP of the exporting country, i, (importer, j) are negative, while the coefficients with variable population of the exporting country, i, (importer, j) are positive. In other words, the growth of the GDP of the importing country, j, reflects the growth of the demand for imports, and also with the growth of the GDP of the exporting country, i, the export supply is higher. In the model, the variable population is an approximation of market size. The larger the population in the importing country, j, the higher the demand for imported products, so the effects on the exporting country, i, are positive. However, population growth in the exporting country, i, has an ambiguous impact. This impact depends on the effect of absorption, population structure and economies of scale.
The estimated gravity model in a logarithmic form is presented in Equation 3: % = ft + MOPit + PzBDPfi + tkPOPit + ^POPjt + M^ij + PsBardij +§5Langij + Eit (3)
The model was evaluated by the method of least squares (two-step procedure). Using the Breusch-Pagan test, a standard test to examine the existence of individual effects (variations by country) and time effects (variations over time), their existence was confirmed. Housman's specification test shows that there is no correlation between regressors and individual effects (as components of random error). In other words, the estimates of the regression parameters of the random error component model are unbiased and consistent. This leads to the conclusion that the specification of the panel model of the random error component is more suitable for the model, i.e. model with stochastic effects.
We used the above model to estimate export flows between different countries. In this paper, an analysis of the export potential of agricultural products in regards to the EU economies was conducted based on the results of econometric evaluation of the panel gravity model, for the period from 2001 to 2017. Specifically, the main goal of the analysis is to examine the directions of exchange of Serbian agricultural products with EU countries.
Results and Discussion
We tested the nature of individual effects (fixed and random) using the F-test. The obtained values of the F-test statistics indicate that we accept the null hypothesis - all coefficients in the model are different from zero, the difference that occurs between standard deviations is a consequence of random errors. In other words, the set model is correct. The rule is that if the test shows that we should not reject the zero hypothesis, then the use of a model with a fixed effect is not recommended. In the model, the
dependent variable (Y) is explained with the independent variables of the model (Xi), which reflects a high level of the determination coefficient (R2). According to other results, the choice is on the model of random effects. Also, according to the Breusch and Pagan Lagrangian multiplication test for the random effect model, we obtained values that are significant. This means that we reject the hypothesis - there are no individual effects, (Chi-square = 16.77; Prob = 0.0000). In this way, the right choice was additionally confirmed - a panel regression model of random effects. In addition, diagnostic tests were performed, which eased the assumptions of the random effects model (more details in the Appendix).
The results of Table 2 show the regression estimates of the gravity model obtained on the basis of Equation 4. First, the least squares estimation (OLS model) was performed, followed by the estimation of the fixed and random effects models. The evaluation was performed for all variables during the entire observed period. In Table 2, the estimates obtained by the basic least squares model (OLS model) are effective, but the estimates are biased (individual heterogeneities are neglected). The fixed effect model, as a rule, does not estimate coefficients that are time invariant, i.e. variables that do not change over time -distance, common language. In contrast to the fixed effect model, the random effect model encompasses the heterogeneity of all explanatory variables. This allows us to simultaneously estimate the parameters of all model variables, both time-varying and time-invariant.
Table 2. Estimated results for Serbia
Dependent variable: X
Variables OLS model Random effects model Fiksed effects model
gdP -5.271971** -3.646479*** -3.756649***
gdp* .6336992** .4144392 1.246826
P°P -23.77332** -26.66919*** -27.01553***
pop* .6816342** .8711883* -6.877642**
dis -2.918302*** -3.732251*** 0
bord .1446244** .082204 0
lang 1.234424*** .9606135 0
cons 432.1311** 466.3214*** 562.1839***
obs 227 227 227
R2 69.48 70.38 30.13
Note: ***; **; * are statisticaly significant at the level of 1%; 5%; 10%.
Table 2 presents the estimated coefficients from the model. In part, certain signs have been realized, while others are contrary to theory and expectations. The coefficient in front of the variable GDP of Serbia (gdp) is statistically significant, and shows a negative impact on the export of agricultural products from Serbia, in all three models. This is contrary to expectations and theory. How to describe the existence of the opposite of the expected impact of Serbia's GDP on the export of agricultural products. In the observed period, the volume of investments in the agricultural sector decreased. The second reason is the structure of household income in Serbia that comes from agriculture and
is distributed within the farm. This is especially the case with small farms, which are represented in a high percentage in Serbia (Tosovic-Stevanovic et al, 2020). At the same time, we should not overlook the fact that the conditions and standardization for access to the EU market are rigorous and almost unattainable for certain agricultural products. The impact of foreign GDP (gdp*) on exports of Serbian agricultural products is positive (statistically significant), low and reflects the stated expectations. This is understandable if we know that there are high subsidies for agriculture within the EU and that they produce enough food for their own needs. Thus, an increase in the GDP of EU member states of 1% would affect the growth of exports of agricultural products from Serbia by only 0.41%, according to the random effect model. Therefore, by directing exports to the EU, high positive results cannot be expected. Here we are talking about the export of raw agricultural products, not agricultural products of higher stages of processing. Investing in agricultural products of higher stages of processing gives a better export result and highlights the growing contribution of GDP (Dimitrijevic et al., 2020). This does not mean that it is necessary to further intensify agricultural production at all costs, as it is often of a devastating nature for the land. Unless there is an intention to introduce innovative, more resource-saving technologies.
The size of the market, which we determined in the model according to the population, has the expected results. Population coefficient for Serbia (pop) is statistically significant and has a negative impact on the export of agricultural products from Serbia. This is due to the stronger absorption effect. On the other hand, the impact of the importer population (pop*) is significant but too low. As with the level of development, we have low elasticity of demand from the EU for these agricultural products from Serbia. The signs of the geographical distance coefficients are statistically significant for Serbia, the exporting country and the EU countries, the countries importing agricultural products (dis and dis*), and they are in line with the expectations. As the distance between the two countries expands, the impact on the export of agricultural products decreases. The common border (bord) and the common language (lang) do not show significant influence.
Conclusions
Through this paper, we have analysed bilateral trade between Serbia and the EU member states. In addition to assessing the effects of agricultural exports from Serbia to the EU, we used the gravity panel model, a dynamic econometric model that has been presented in practice as a convenient approach to examining multilateral trade flows. We assessed the export-import trade for 28 EU economies and Serbia, in the period 2001-2017.
The presented results of the regression model show that they are statistically significant and explain about 70% of Serbian agricultural exports to the EU using three variable explanations: economic development, market size and distance between countries i and j. Specifically, based on the gravity model of exports of agricultural products from Serbia to the EU member states, it can be stated that there is a statistically significant positive correlation between economic development and market size, and a statistically significant negative correlation when it comes to the distance. Taking into account the
defined gravity model, it must be pointed out that the potential inclusion of additional variables in the model could lead to new conclusions important for trade flows.
The model results are expected. All the conclusions derived from the above analysis can be valuable for the makers of economic decisions in finding the best model for improving trade flows between Serbia and the EU members in the future.
Conflict of interests
The authors declare no conflict of interest.
References
1. Aitken, N. (1973), "The Effect of the EEC and EFTA on European Trade: A Temporal Cross-section Analysis", American Economic Review, 63, 881-92.
2. Bergstrand J. H. (1985), The gravity equation in international trade: some microeconomic foundation and empirical evidence, Review of Economic and Statistics, 67, 474-481.
3. Bialynicka-Birula, J. (2015). Modelling International Trade in Art - Modified Gravity Approach, Procedia Economics and Finance 30, 91 - 99.
4. Bozic, D. & Nikolic, M. (2016). Obelezja spoljnotrgovinske razmene poljoprivredno-prehrambenih proizvoda Srbije. Marketing, 47(4), 293-304, DOI: 10.5937/markt1604293B
5. Bussiere, Matthieu; Fidrmuc, Jarko & Schnatz, Bernd (2005). Trade integration of Central and Eastern European countries: lessons from a gravity model, ECB Working Paper, No. 545, European Central Bank.
6. Chan-Hyun, S. (2001). A gravity Model Analysis of Korea's Trade Patterns and the effects of a Pjegional Trading Agreement. The International Centre for the study of East Asian Development, Kitakyushu Working Paper Series, 09. Available on www.agi.or.jp
7. Dimitrijevic, M., Vrzina, S., & Lekovic, M., (2020). Agricultural Enterprises and Economic Growth: A Regional Analysis in the Republic of Serbia, Economics of Agriculture, 67(2), 585-600, ISSN 2334-8453. doi:10.5937/ekoPolj2002585D.
8. Egger, P. & M. Pfaffermayr (2003). The proper panel econometric specification of the gravity equation: A three-way model with bilateral interaction effects. Empirical Economics, 28, 571-580.
9. Eita, H. J, & Jordaan, A. C. (2007). South Africa's Wood Export Potential Using a Gravity Model Approach University of Pretoria. Working Paper 23.
10. Hamilton, C. B. and A. L. Winters (1992). Opening up International Trade with Eastern Europe. Economic Policy, April, 78-115.
11. Jovovic, D. & Jovovic, D. (2018). Competitiveness of food manufacturing of Republic of Serbia, Economics of Agriculture 1/2018, 49-64, doi:10.5937/ ekoPolj1801049J
12. Linnemann, H. (1966), An Econometric Study of International Trade Flows, Amsterdam: North-Holland.
13. Oguledo, V. I. & Macphee, C. R (1994). Gravity Models: A Reformulation and an Application to Discriminatory Trade Arrangements, Applied Economics. 26, 107120. doi/abs/10.1080/00036849400000066
14. Porojan, A. (2000), Trade flows andd spatial effects: the gravity modei revised, Review of University College London, 4, 168-173.
15. Pöyhönen, P. (1963), "A Tentative Model for the Volume of Trade between Countries", Weltwirtschaftliches Archiv, 90, 93-99.
16. Ranilovic, N. (2017). The Effects of Economic Integration on Croatian Merchandise Trade: A Gravity Model Study, Working Papers W-50, Croatian National Bank.
17. Ristanovic, V. (2015). Struktura industrijske proizvodnje u Srbiji nakon 2000-te, Ekonomski pogledi, 17(4); 159-172; ISSN 1450-7951 Online ISSN 2334-7570. Available on www.ekonomskipogledi.pr.ac.rs > Ekonomski pogledi.
18. Ristanovic, V., Barjaktarevic S., & Cogoljevic D., (2017). Direction of Serbian Trade: Gravity Model Based on Pool Data, EuroEconomica, 36(1). Available on journals.univ-danubius.ro > index.php > article > view
19. Ristanovic, V., Miljkovic, D. & Barjaktarevic S., (2019). Izvozni potencijal Srbije u EU, Megatrend revija, 16 (1), ISSN 1820-3159, 1-25. doi:10.5937/MegRev1901001R.
20. Smarzynska, B. K. (2001). Does relative location matter for bilateral trade flows? An extension of the gravity model, Journal of Economic Integration, 16, 379-398. doi:10.11130/jei.2001.16.3.379
21. SORS (2018). Foreign merchandise trade, Statistical Office of the Republic of Serbia, Belgrade. http://webrzs.stat.gov. rs (03.05.2021.)
22. Tinbergen, J. (1962), Shaping the World Economy: Suggestions for an International Economic Policy, New York, NY: Twentieth Century Fund.
23. Tosovic-Stevanovic, A., Ristanovic, V., Calovic, D., Lalic, G., Zuza, M., & Cvijanovic, G. (2020). Small Farm BusinessAnalysis Using theAHP Model for EfficientAssessment of Distribution Channels, Sustainability, 12, 10479; doi:10.3390/su122410479
24. Trotignon, J. (2010), Does Regional Integration Promote the Multilateralization of Trade Flows? a Gravity Model Using Panel Data, Journal of Economic Integration, 25(2), 223-251.
25. Vioricä, E.D. (2012). Econometric Estimation of a Gravity Model for the External Trade of Romania, Journal of Eastern Europe Research in Business & Economics, Vol. 2012, Article ID 854058. doi: 10.5171/2012.854058,
26. Waheed, A., & Abbas, S. (2015). Potential export markets for Bahrain: A panel data analysis. International Journal of Trade, Economics and Finance, 6(3), 165.
Appendix
Table A1. Descriptive statistics of Serbian export of the agrarian products in the EU, 2001-2017
Panel A: Descriptive statistics
stats exp GDP GDP* POP POP* distance border language
mean 13.55505 33171.69 491187.8 7189280 1.81e+07 1534.393 .1428571 .0714286
max 385.566 36795.4 3263350 7350220 8.27e+07 3283 1 1
min .0 29766.3 6128.7 7040270 409370 389 0 0
sd 44.88181 1927.727 744402.2 104115.3 2.30e+07 826.1124 .3505537 .2580005
N 280 280 280 280 280 280 280 280
Panel B: Correlation
exp gdp gdp* p°p pop* dis bord lang
exp 1.0000
gdp -0.0860 1.0000
0.1969
gdp* 0.2685* 0.0404 1.0000
0.0000 0.5009
P°P 0.0035 -0.6943* -0.0466 1.0000
0.9577 0.0000 0.4570
pop* 0.4495* 0.0025 0.9125* -0.0037 1.0000
0.0000 0.9674 0.0000 0.9505
dis -0.5897* -0.0000 0.2452* 0.0000 0.0349 1.0000
0.0000 1.0000 0.0000 1.0000 0.5614
bord 0.4493* 0.0000 -0.2106* 0.0000 0.0365 -0.6584* 1.0000
0.0000 1.0000 0.0004 1.0000 0.5428 0.0000
lang 0.1715* -0.0000 -0.2450* -0.0000 -0.1917* -0.3678* 0.2831* 1.0000
0.0096 1.0000 0.0000 1.0000 0.0013 0.0000 0.0000
Table A2 Breusch-Pagan/Cook-Weisberg test heteroscedasticity
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity
Ho: Constant variance_
Variables: fitted values of exp
chi2(1) = 16.77 Prob > chi2 = 0.0000
Table A3. The variance inflation factor of independent variables
Variable VIF 1/VIF
gdp* 12.40 0.080644
pop* 10.98 0.091084
bord 2.29 0.437370
dis 2.14 0.467006
P°P 1.96 0.509658
gdP 1.96 0.510793
lang 1.28 0.780487
Mean VIF 4.71
Note: A vif > 10 or a 1/vif < 0.10 indicates trouble.
Table A4. Hausman test
Coefficients (b-B) Difference sqrt(diag(V_b-V_B)) S.E.
(b) fixed (B) random
gdP -3.756649 -3.646479 -.1101695
gdp* 1.246826 .4144392 .832386 .6939323
P°P -27.01553 -26.66919 -.3463429 .9249085
pop* -6.877642 .8711883 -7.74883 3.100887
b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(4) = (b-B)'[(V_b-V_B)A(-1)](b-B) = 6.33 Prob>chi2 = 0.1761 (V_b-V_B is not positive definite)
