**2. Literature Review**

The European Organization for Cooperation and Development (OECD) emphasizes a vision based on bio-technologies (reflected in the vision on bioecology-focused bioeconomy) [9]. From the perspective of the European Commission (2017), a significant variety of research and innovation priorities related to the bioeconomy at the level of the European regions/countries have been identified. Most countries/regions use a mix of thematic areas, from the perspective of both the focus on bioresources and the orientation towards energies obtained from bioresources. At the European level [10], several bioeconomic development initiatives, including with regional orientation, have

been identified. The need for each nation to build a competitive advantage (supported by a localized territorial process and allowing it to di fferentiate from the other nations) has allowed the emphasis on the competitiveness of a nation through its ability to innovate and on the ability to create and assimilate knowledge [11]. Other approaches [12] consider that the focus on the bioeconomy stems from the need to cover the food requirements of a growing population, related to lower yields of agricultural production, or from the need to ensure energy and food security as well as economic prosperity in the face of some new challenges—climate change. The transition to the bioeconomy involves concerted e fforts, both on the part of the authorities and on the part of the society, as such a transformation involves substantial changes in the market through the impact of technological development on industrial processes, ultimately a ffecting the production and consumption patterns. The success of the bioeconomy is dependent on the active involvement of the authorities in the creation of an adequate legislative framework, taking into account that the advanced bioeconomy will become a reality only if the intensification of the research and development e fforts will be reflected in the subsequent implementation of the technologies. The bioeconomy can reflect the direct link between innovation and economic growth [10], in the sense that increasing productivity by maximizing the efficiency of the resources used in counterpoint with limiting the impact on the environment can be achieved only through technological research and development. It is worth mentioning that innovation must be accepted by each participant in the economic chain as well as by the society as a whole. Identifying the stimulating factors of the transition to the bioeconomy is a di fficult and complex process, given their diversity. The analysis carried out in 2018 by the FAO (Food and Agriculture Organization, a specialized agency of the UN, with the aim to eliminate world hunger, as well as to improve the food, by coordinating the activities of the governments in the field of agriculture, forestry, and the fishing industry [13]) reflects the contribution of the bioeconomy to a country's economic growth. Although the implementation of this concept requires a harmonization with the particularities and priorities of each state, in a general framework, however, certain aspects essential to the development of the bioeconomy can be identified (see Figure 1).

**Figure 1.** Essential factors for the development of a sustainable bioeconomy. Source: [13].

An empirical study [14] on the EU component states has shown that Finland and Sweden have the lowest levels of environmental pollution due to the rigorous ecological awareness of the population; the focus in these countries is on education and vocational training, with a basis of solid knowledge. In addition, the two states are among the most innovative countries in the EU and are based on rich and diverse natural resources. Denmark, Ireland, the Netherlands, and the UK have similarities in terms of innovation capacity supported by a developed economy, but natural assets are narrower than in the two Nordic countries (much of the countries' areas are used for agriculture, as forests are restricted as a surface), and the quality of the environment is above average. At the opposite end, these states are noted: Bulgaria, Croatia, Cyprus, Czech Republic, Greece, Hungary, Italy, Latvia, Lithuania, Poland, Portugal, Romania, Slovakia, and Spain. Although they have the largest agricultural sectors, innovation activity is relative low, which results in a low employment rate in the technological field. In these

countries, the public authorities are less dedicated to education and training, and the population is not so concerned about the environment. According to the studies, the historical, geographical, and cultural factors influence the pro-bioeconomic behaviors adopted by the citizens. The size of the socio-economic context highlighted the most visible differences between countries, leading us to the conclusion that the countries of Central and Eastern Europe are in different stages of development [15].

The economic literature has developed progressively, encompassing the issues of bioeconomy and sustainability, as well as the determining elements that influence the economic growth.

In this section of the research, relevant aspects of some studies and research were analyzed, which include the issue of the sustainability of economic growth, the analysis of its components, analysis of the developments of bioeconomy at the European level, research on the six macroeconomic indicators included in the empirical study, and their correlations. Foreign trade, both export and import, continued to be one of the fundamental factors of economic growth contributing to the growth of national economies. The value of foreign agri-food trade is relevant, considering that, in 2018, the EU maintained its position as a world leader in the global export of agri-food products, with EU exports reaching 138 billion EUR in 2018. The top five destinations for food products exported by the European Union continue to be the United States, China, Switzerland, Japan, and Russia, which account for 40% of EU exports. The EU's common agricultural policy has become increasingly market-oriented, thus contributing to the EU's success in agricultural trade [16]. In 2018, the EU became the second largest importer of agri-food products in the world, the value of its imports amounting to 116 billion EUR, bringing the EU trade balance for this sector to a net positive result of 22 billion EUR. The EU mainly imports three types of products: Products not produced in the EU (or are produced only to a small extent, such as tropical fruits, co ffee, and fresh or dried fruits), representing 23.4% of EU imports, products that are intended for animal feed (accounting for 10.8% of EU imports), and products used as ingredients in further processing [16]. Although agri-food trade is shown to benefit from a positive global climate assessment from 2019 [16], substantial future risks remain for trade developments [16]. The biggest threats to trade developments include protectionist political approaches (which are increasingly important for economies), more frequent trade disputes, and possible trade unrest linked to Britain's decision to leave the EU. On the positive side, global demand for food is likely to increase, correlated with population growth, income growth, middle-class expansion, and changes in consumer preferences [17].

Figures 2 and 3 respectively show the evolution of exports and food imports at the level of the 24 countries included in the empirical study, highlighted distinctly by the two groups (countries of Western and Northern Europe, considered countries with developed economies, and countries of Central and Eastern Europe, considered countries with emerging/developing economies).

**Figure 2.** The evolution of total food exports in the countries included in the empirical analysis. Source: Own processing, data are sourced from the World Bank database [18].

**Figure 3.** The evolution of total food imports in the countries included in the empirical analysis. Source: Own processing, data are sourced from the World Bank database [18].

Thus, it is observed that, in the case of countries with emerging/developing economies, the variations of exports and imports are more pronounced compared to those registered with the countries with developed economies, which can absorb the impacts of the influence factors. The economic literature analyzes the effects of imports and exports on private research and development expenditures in the food-processing sector. The empirical results [19] reflect that increasing the level of import intensity leads to reductions in private spending on research and development, while increasing the level of export intensity promotes higher private spending on research and development. These results imply that the effects of reducing the research and development activity of imports offset the effects of improving export research and development. Other studies examine the impact of EU enlargement on export performance of agri-food products in 12 new EU Member States and five new independent states on EU markets, covering the period 1999–2007 [20]. A longer duration for agri-food exports from the new EU member states was identified. The results confirm the gains from the eastern enlargement of the EU in terms of export growth and a longer duration for the export of specialized foods, with a higher added value for consumers and more competitive niche agri-food products [20].

### **3. Materials and Methods**

This research applies scientific tests, uses specific estimators and statistical–econometric techniques, investigates data sets and collections, and assesses the most appropriate methods of investigation in order to provide accurate results. The activity of foreign trade in foodstuffs, transposed in the external balance of a country, can make a significant contribution to the economic (sustainable) growth of the respective country. Especially in the context of the transition to the green economy, a strategic vision must include the factors that achieve a significant influence. It is also necessary to integrate and study the behavioral evolution, habits/preferences and attitudes of consumption, and the degree of adoption and use of technologies along the value chain from plant culture/animal growth to food processing/distribution.

As regards the selected countries (presented in Table 2), on the one hand, the founding countries of the European Union were included; on the other, countries in Central and Eastern Europe, representative in terms of structural changes in the economy, were also included. The countries in Central and Eastern Europe are affected by processes of transition from a centralized economy to a market economy, or are even in the early stages of reforms, such as in Macedonia, Former Yugoslav Republic (FYR), a candidate country for EU accession. The authors opted for a split into two groups of countries based on the criteria: Geographical and economic development. The division into two groups of countries based on their level of development was made taking into consideration similar approaches to be found in the field's literature, such as the ones cited in this article. Additionally, a division into three groups would make the groups very unequal with respect to the volume of the sample, with advanced economies having much more representation than the developing or the emerging ones; consequently, the representativeness of such results would be far lower than in the present situation (lower accuracy). Similar divisions of the European countries are to be found in [21,22]. With respect to the criteria of economic development, in practice, international bodies also operate with the same classifications: 1. economically advanced countries and developing and emerging countries [23]; 2. developed markets, emerging markets, and border markets [24]; 3. developed economies and economies in transition [25].


**Table 2.** List of countries according to their grouping by level of development.

Source: Authors' own processing.

In the empirical study, to determine the inter-correlation with the exogenous indicators of foreign food trade, six relevant endogenous indicators were selected, including from the perspective of the bioeconomy/sustainable development: Labor force, added value of agriculture, forestry and fisheries, research and development expenses, forest area, fuel consumption based on fossils, and renewable energy consumption:


spent) and data on patents for invention (granted for inventive technologies with marketing prospects) [18].


Through a quantitative mix of instruments, the nature of the inter-correlations between these indicators was studied in order to provide certain answers to the fundamental question of this research: Which of the indicators analyzed at the level of the 24 European economies, over a period of 22 years, has a positive impact on the determination/influence, in a relevant way, of the evolution of food exports and imports? To answer this question for the present analysis, a series of six working hypotheses has been constructed, which will be tested using the multiple regression model; the first of these is methodological in nature:


• *H6. Renewable energy consumption is inversely correlated with food exports and directly correlated with imports.* Countries with important renewable energy sectors are more developed countries, which export products other than food, but mainly import this type of product because they do not have a highly developed agricultural sector.

The purpose of the present research is to link the foreign trade of foodstu ffs, estimated both by exports (EXP) and by imports (IMP), with the following factors:


In the preliminary stage of the actual modeling of the relation between the dependent variables and the main determinants considered in the present analysis, it is necessary to investigate the statistical properties of the series of variables. Following this examination, the most appropriate statistical–econometric techniques are decided to model the link between the variables included in the study. Moreover, before the start of the statistical analysis, all of the variables considered were respectively logarithmized for a possible normalization of their distribution for an easier interpretation of the associated coe fficients in the form of elasticities. To determine whether the series of variables are stationary in the level or first di fference, the Fisher–Phillips Perron unit root test developed by Choi [28] was applied. The main advantage of this test is that it can be applied to both balanced and unbalanced panel data. Thus, considering the series of our variables that sometimes have missing values, it was considered that the application of this test is the most appropriate. First, in the analysis of stationarity for the level of variables, it was included in the equation for both the constant and the trend. Considering that series of macroeconomic variables most often have a certain tendency, including the trend in the equation increases the accuracy of the results. Secondly, for the first di fference of the series, only the constant in the equation was included, since the di fferentiation of the series leads, in most cases, to the elimination of a possible tendency. Moreover, to correct for potential data persistence, both equations are aggregated with a lag.

The results of the stationarity test presented in Table 3 sugges<sup>t</sup> that both dependent variables, i.e., exports and imports, have a unit root (they are integrated with an order of one - I (1)); the p-value associated with the statistics calculated for the level of the variables is higher than the significance thresholds of 1% and 5%, thus leading to the acceptance of the null hypothesis that the series are characterized by the unit root. In contrast, for the first di fference of the variables, the p-value associated with the calculated statistics is lower even than the significance thresholds of 1%, leading to the rejection of the null hypothesis and the acceptance of the alternative one, according to which the series are stationary.

Regarding the exogenous (independent) variables, the results are mixed, as the variables are both integrated by the first order and stationary at the level. Taking into account the characteristics of the series of variables included in this analysis—namely that the dependent variables are I(1), and the independent variables are both I(1) and I(0)—for modeling the relation between them, a dynamic model was considered, namely an ARDL (Autoregressive Distributed Lag) data panel. The mathematical form of the dynamic model for the ARDL panel data (p, q\_1 ... q\_k) [29] is as follows:

$$Y\_{it} = \sum\_{j=1}^{p} \partial\_{ij} Y\_{i, t-j} + \sum\_{j=0}^{q} \gamma'\_{ij} X\_{i, t-j} + \mu\_i + \varepsilon\_{it\prime} \tag{1}$$

where *i* = 1, *N* represents the countries analyzed, and *t* = 1, *T* denotes the number of years included in the study (the period analyzed). *Yit* is the dependent variable, and *Xit* (*k* × 1) represents the vector of explanatory variables with the vector of associated coe fficients <sup>γ</sup>*ij*(*<sup>k</sup>* × <sup>1</sup>).


**Table 3.** Test of stationarity (unit root).

Source: own processing. Note: The null hypothesis (H0): All panels contain unit roots, and the alternative hypothesis (H1): At least one panel is stationary. For the stationary variables at a level of significance of 1% and 5%, the first difference of the series was not analyzed. \*\*\* and \*\* indicate statistical significance at a threshold of 1% and 5%. In the above table, the following abbreviations were used: Exports (EXP), imports (IMP), labor force (LABOR), gross added value of agriculture, forestry, and fisheries (AGRI), expenditure on research and development (RD), forest area (FOREST), fossil fuel energy consumption (ENG), and renewable energy consumption (RENEW).

In our case, the dependent variable is represented by the exports; the imports of food products, and as their main determinants, the labor force, the added value of agriculture, forestry and fisheries, and the expenditure for research and development were respectively considered. Moreover, μ\_i and ε\_it indicate the country-specific fixed e ffects and the error term, respectively.

The above Equation (1) can be rewritten in the form of a panel data error correction model if it is assumed that the variables are non-stationary and co-integrated. Thus, the equation incorporating both long-term and short-term coe fficients, together with the error correction term Equation (2), has the following form:

$$
\Delta Y\_{it} = \phi\_i \big( Y\_{it-1} - \lambda\_i' X\_{it} \big) + \sum\_{j=1}^{p-1} \partial\_{ij}^\* \Delta Y\_{it-j} + \sum\_{j=0}^{q-1} \gamma\_{ij}'^\* \Delta X\_{it-j} + \mu\_i + \varepsilon\_{it} \tag{2}
$$

,

where:

$$\phi\_i = -(1 - \sum\_{j=1}^p \partial\_{ij})\_\prime$$

$$\lambda\_i = \sum\_{j=0}^q \gamma\_{ij} / \left(1 - \sum\_k \partial\_{ik}\right)\_\prime$$

$$\partial\_{ij}^\* = -\sum\_{m=j+1}^p \partial\_{im\prime}$$

$$\gamma\_{ij}^\* = -\sum\_{m=-j+1}^q \gamma\_{im\prime}$$

and Δ represents the di fference operator.

In order to confirm the long-term relationship between the variables, the coe fficient associated with the error correction term, namely φ*i* must be negative and statistically significant, and its values must be between [−1; 0]. Moreover, it helps us evaluate whether the model is specified correctly and to determine the speed of adjustment of the system to long-term equilibrium following an exogenous shock. First, it should be noted that one advantage of the ARDL technique on panel data is the accuracy (consistency) of the estimated coe fficients when the dependent variable is I(1) and the independent variables have di fferent integration orders of (I(0) and I(1)). Secondly, another advantage is given by the flexibility of the estimated coe fficients, in the sense that it allows us to evaluate the influence of the independent variables on the dependency in both the long term and in the short term.

Considering all of the results related to the evaluation of the characteristics of the analyzed variables, an ARDL model (1.1) was estimated, including one lag for the dependent variable and, respectively, one lag for the independent ones. The decision to include a lag in the model is closely linked to the value of the Akaike Information Criterion (AIC) and the total number of panel-level observations. Considering that N = 24 and T = 21 (N ∗ T = 504), the inclusion of several lags in the model significantly reduces the number of observations; the ARDL model is sensitive in this respect. It should be mentioned that the main analysis was started for a group of 24 countries in Europe (Austria, Belgium, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, France, Germany, Hungary, Italy, Latvia, Lithuania, Luxembourg, North Macedonia, Netherlands, Norway, Poland, Romania, Slovak Republic, Slovenia, Spain, Sweden, and United Kingdom) for the period 1996–2017. The choice of the analysis period was strictly determined by the availability of data and, respectively, by the variable of expenditure on research and development, for which the values stop in 2017. The ARDL model (1.1) is estimated using three specific estimators, namely the Dynamic Fixed E ffects (DFE), Pooled Mean Group (PMG), and Mean Group (MG). Then, the Hausman test helps us determine if the PMG estimator or the MG is best suited to model the evaluated data. It should be noted that the DFE estimator considers that both the short-term and long-term coe fficients, together with the error correction term, are identical for all panel members (for the analyzed countries) only the constant is di fferent, depending on the country. On the other hand, the MG estimator assumes the exact opposite (the short-term and long-term coe fficients, together with the error correction term, are di fferent for all panel members), and the PMG estimator is the intermediate version between the two, considering a common long-term trend for all countries, with a respective short-term heterogeneity between coe fficients. The final step in the analysis was the validation of the final models by evaluating their robustness. For this purpose, three variables (related to both the agricultural sector and the size of sustainable development) were introduced in the analysis, taking into account the continuous discussions at the international level related to the problem of natural resources and their diminution. For the present analysis, the following variables were considered as control variables:


These were used to check if the relationships found in the main regressions remain stable in the presence of environmental and long-term sustainability factors, i.e., FOREST, ENG, and RENEW.
