*4.9. Models Based on Agriculture Subsidies and Agriculture Loans*

The model presented in Figure 18 describes the relation between the agricultural credit and vegetables production with a regression *p*-value at 0.01 (Figure A20). Thus, this reveals that a significant part of agriculture loans is used for vegetable production. Therefore, agriculture loans can be used as control tools for maximizing the vegetable production.

**Figure 18.** Agr\_Credit. – Vegetable\_Prod model.

An important model describes the relation between governmental agriculture subsidies as predictor and agricultural loans (Figure 19a). With a maximum R-square of 81.40, governmental agriculture subventions explain the variability in agricultural loans.

**Figure 19.** (**a**) Subsidies–Agr\_Credit quadratic model. (**b**) subsidies–Agr\_Credit model.

The Versus Order chart shows no particular form of dependence of the error terms, while the distribution of the residuals, according to the normal probability plot, is normal (Figure A21). By using a linear approach for the same scenario presented in Figure 19a, a lower R-square is obtained, respectively 72.80% (Figure 19). Still, as the model involves a linear equation, it is easier to numerically grasp the relation between the predictor and the dependent variable. The equation shows that for every extra 1% additional subsidies, the agricultural credit will decrease by an average of 0.83 percent. However, this negative correlation between agriculture subventions as predictor and agricultural loans is expected as they must work as complementary financing sources. In this direction, it is recommended to cover the gap between the value of governmental agriculture subsidies versus the value of agricultural loans.

At first glance it seems there is linear relation between total GDP and subsidies (Figure 20). Still, due to a situation in which for the medium-high total GDP value there was a very low subsidy value, the model returned an R-square of only 21.40%. Removing that data point would generate the following situation: both the R-square and adjusted R-square significantly increased, and as the equation shows, for every 1 unit GDP increase the subventions will increase with 0.004 (Figure 20a). Still, we consider the R-square to be quite low for some conclusions to be drawn. However, by analyzing Figure 20a, after the outlier removal, the total GDP–Subsidy model returned an R-square of 55.80%. However, it is hard to identify a certain pattern between total GDP and subsidies value since the percentage of subsidized farms was less than 1%. It is possible that if the subsidies values rise the effect of this

governmental financial support will be better observed in the Republic of Moldova macro-economy and, therefore, in the total GDP.

**Figure 20.** (**a**) GDP–Subsidies—outlier removed. (**b**) GDP–Subsidies model.

The conclusions mentioned at the previous described model (between total GDP and governmental agriculture subsidies) are confirmed by the model from Figure 21 which describes the relation between GVA per farm and subsidies per farm and that displays some peculiarities. Even if it is not statistically relevant, with an R-square of 31.40%, it is still possible to see why the value is low (Figure 21a,b). This is mainly due to a specific case where, for a very high value of subsidies per farm (16,100 USD) in 2014, the GVA per farm was not as high as expected, probably due to external factors. Another peculiar fact is that high values of GVA per farm are associated with low values for subsidies per farm.

**Figure 21.** (**a**) GVA\_Farm–Farm\_Subsidies quadratic. (**b**) GVA\_Farm–Farm\_Subsidies.

Actually, subsidies per farm does not change much, while GVA per farm almost doubles, so it seems that directly, subsidies per farm does not radically influence the GVA per farm. This can be due to the direction in which subsidies are invested by farm owners.

As mentioned previously, subsidies are mainly directed toward technical investments and not to yearly production; therefore, their effect may be visible on a long-term period and during the years in which the climate change effect is more intense.

An increased agricultural GVA is associated with a constant increase in the subsidies per farm (Figure 22). The data presents an outlier, where for a small increase in agricultural GVA there was a significant increase of subsidies per farm, probably due to external factors that were not considered. Still, the best model is obtained when we apply the log function to both predictor and dependent variable, having an R-square of 99.88 (Subsidies per farm Log = Log (Subsidies per farm) and Agriculture GVA Log = Log Agriculture GVA) (Figure 22a).

**Figure 22.** (**a**) GVA\_Agriculture–Farm\_Subsidies Log. (**b**) GVA\_Agriculture–Farm\_Subsidies.

The Figure 23 displays how subsidies influence subsidies per farm. As it can be noticed, subsidies per farm mostly increase with the increase of subsidies. The quadratic are better representations of the current situation due to the high value subsidies per farm point, which is also associated with the highest value for subsidies. The S value, at 1502, can be considered small for a scale ranging from 2000 to 16,000. Both the agricultural GVA–subsidies per farm model and subsidies–subsidies per farm model where possible in conditions of a relatively constant dynamics of agriculture farm number. However, their significance is expected to decrease in a long-term macro-economic analysis, when agricultural farm number can register higher variations.

**Figure 23.** Subsidies–Farm\_Subsidies model.

All models described above will contribute to a better understanding of the Moldavian agriculture sector, offering tools that can be used in order to ensure an efficient intervention that targets to maximize the performances of this sector in different scenarios and, therefore, contribute to the economic sustainability of agriculture sector.

As with the majority of studies, the design of the current study is subject to limitations. One limitation of the present paper's analytical framework is the lack of a multifactorial connection between the economic subsystem and social and environmental subsystems. This would potentially establish a better agriculture sustainability approach. However, the economic sustainability of agricultural activities is considered the basic constraint for the survival of farm systems over time [45]. Therefore, if agriculture sustainability is targeted, economic sustainability must be the starting point.

The second limitation concerns the economic indicators list that could be extended in order to elaborate on a more complex analytical framework. According to Latruffe et al. [46], a wider range of indicators has been proposed to capture various economic properties of farming systems that are associated with sustainability, such as profitability, liquidity and stability or autonomy/dependence [47]. However, subsidy-dependence can be considered an aspect of autonomy: if farms are highly dependent on public support, any policy reform that reduces subsidies could put farm sustainability at risk [46]. Other studies [48] that multidimensionally analyzed agricultural economic sustainability used specific indicators that analyse the imports and exports of agricultural products, as well as the gross regional product values compiled with the crop productivity indicators.

The analysis of agricultural economic sustainability can be oriented towards different types of fish farms according to their production potential. Therefore, studies which assessed the sustainable economic development for family farms [49] developed the analytical framework based on indicators such as return on equity, return on assets, operating expense, debt to assets, debt to total output, depreciation expense or gross margin.

The economic dimension of agricultural sustainability is characterized in other studies by using various economic and socio-economic indicators [50–54]. However, the evaluation of the economic sustainability of farms involves a wide range of sustainability themes that are difficult to combine in a unique approach [50,51]. Therefore, by using a multifactorial approach that considers labor, land profitability and productivity, as well as vitality marginalization, diversification of farmer labor, competitiveness, incidence of agricultural value added and fixed investment in agriculture, Dillon et al. 2014 [53] concludes that efficient agricultural structures, appropriate technologies as well as the diversification of income sources for farms are critical elements for the achievement of the agricultural sector's economic sustainability. Other studies [50,54] concluded that the diversification of farm activities significantly contributes to the economic sustainability of agriculture, while using an analytical framework structured on the following indicators: value of production, value added, farm ability to generate income, autonomy, diversification of the production, business diversification and multifunctionality.

The third limitation of the present study concerns the lack of governance and institutional capacities indicators. This is because governance is an essential driver of both agricultural productivity and sustainability in developing countries [55]. However, those indicators are related to agricultural policies and technological interventions that would rather concern the political economy approach, although correlating the two sustainability subsystems (economic and political subsystems) could raise the analytical framework level of complexity.

However, analytical framework studies that are predominantly based on economic and production indicators revealed that the effect of agricultural GDP growth on poverty reduction is at least twice as high as the effect of GDP growth coming from other sectors.

For the Republic of Moldova's agriculture, analytical framework studies were conducted in order to increase the competitiveness of the horticultural sector [56]. This revealed that with the increase of the share of the horticultural production sales income regarding the total sales income, being more than 20%, the enterprises will be more competitive on the market, registering higher values of total productivity factors. Furthermore, a study [56] confirmed that horticulture may represent a solution for developing the Moldavian agriculture sector. This is due to the fact that the agriculture farms that had a share of the horticultural production sales income in total sales income of more than 20% performed better in terms of financial productivity.

However, no similar studies concerning the economic sustainability of the Republic of Moldova's agriculture sector were found in the literature.

The present study confirms that agriculture subsidies and agriculture loans can be also used directly in order to raise the share of agriculture GDP in total GDP. However, Lopez et al. [57] in their study related to agriculture subsidies in China has issued a hypothesis according to which the larger the share of agriculture is in the economy, the costlier it will be to subsidize; also, in the long term, subsidy governmental programs may cause farm dependence [46] and if stopped or reduced, may cause significant economic disturbances. In the present study, an indirect correlation is observed between governmental agriculture subsidies value and agriculture loans. In spite of this correlation, the value of subsidies is considerably low compared to agriculture loans and only a small percentage of total agriculture farms managed to access them during the analyzed years. Thus, agriculture governmental subsidies cannot be considered financial sources for the substitution of agriculture loans. However, the relation between agriculture loans and subsidies presented in present study can be attributed to the fact that, in general, subsidy financing programs are a result of governmental policy intention of supporting the agricultural economic sector during a certain period of time. Therefore, the high level of governmental subsidies is proportional with other measures, both being part of governmental support programs which targets to improve the economic sustainability of agriculture sector. Therefore, the dynamics and size of subsidies can emphasize the intensity of governmental supporting program of which they belong rather than its real financial value. This hypothesis can also explain the direct correlation between the value of farm subsidies and agriculture GVA. Thus, if the subsidies' size and dynamics reveal the intensity of then governmental supporting program, this must be observed on the agricultural GVA dynamics. However, the relationship between the agriculture sector's performance and governmental supporting programs intensity can be inverse [58] if inconsistent financing policy is applied over long periods—the increase of financial support must be applied before a relative downsizing of the agriculture sector. Recent studies [59] have revealed the positive effect of subsidies in modern maize agriculture, only if this financial support is invested in crop-growing technology, specifically in seeds and fertilizers. Therefore, Vozarova and Kotulic [40] concluded that removal of agricultural subsidies would contribute to increasing the income disparities between rural and urban areas, which would lead to an exit of domestic farmers from the industry. For this reason, according to Won and Kennedy [60] most countries use some form of subsidy in order to protect their agriculture, since studies [61] have been proven it to be the most effective mechanism for accelerating the growth of the agricultural sector. However, although several studies proved the effects of subsidies and other financial tools on productivity [62–64] or the efficiency of agriculture [65–69], the subject still remains open for discussion, as it depends on numerous series of climate, social or political variables.

In present study, the significantly high correlation between total GVA and agricultural GVA reveals the significant influence of the agriculture sector in assuring and improving the Republic of Moldova's economic performance. This confirms the findings of Timofti et al. [3], who emphasized that the financing and developing of the Moldavian agriculture sector is a cornerstone in achieving modernization; also, Dinu et al. [19] encourage the development of the agriculture sector since it has great importance for the economic growth of the Republic of Moldova. A country that is not self-sufficient in food production can be more vulnerable to commercial pressure and the global food crisis [70].

The direct correlations between agricultural loans and vegetable production can be due to the peculiarities of vegetables production systems. Therefore, most of the vegetable production is obtained in greenhouse, intensive crop production systems, since the seasonality of these crops can significantly influence their profitability [6]. Thus, the influence of the agricultural loan value on vegetable production can be justified because the production intensity of greenhouse-based systems is significantly dependent on the level of financial investment in high-performance equipment, which assures a maximization of growth performance and a good seasonality compared to the competitors from other countries.

The present study also emphasizes the importance of Moldavian wheat production for the economic growth. Therefore, a direct correlation was identified between wheat production value and total GVA. Wheat and maize production quantity is also strongly and directly correlated with total plant production, revealing the importance of the production performance of these two crops for the Moldavian agriculture sector. Thus, the policy of supporting the improvement of production technologies for maize and wheat in order to maximize their growth rate can raise the economic sustainability of the agriculture sector.

Although advanced data analysis frameworks are performed, it is hard to establish a general model that will perform perfectly in all circumstances. This was also confirmed by Nowak et al. [71], who argue that to evaluate the performance and effectiveness of agriculture is quite complicated, not least due to the instability of the climatic conditions but also due to the wide variety of households in view of their economic strength and production profile.

#### **5. Conclusions**

The agriculture sector has a major impact on the Moldavian economy, a fact revealed by the significant model between agricultural GVA and total GVA. The Republic of Moldova's agriculture policies must focus on maintaining a high and constant financial standard for governmental agriculture supporting program, in order to improve the economic sustainability of this production sector. However, a negative significant correlation was identified between agriculture subsidies supporting programs and agriculture loans. Therefore, it is recommended to assure the complementarity of the two funding sources (agriculture supporting programs and agriculture loans), in order to obtain a better economic performance of the agriculture sector.

The supporting programs prove to be highly efficient for increasing the production quantity of crops cultivated in greenhouse, intensive production systems, such as vegetables. Thus, since vegetable production has both a social and economic impact (improving the living standards of rural communities and increasing the agri-food exports) it is recommended to continue and even increase the governmental agriculture supporting programs for maximizing the production of these crops.

The Moldavian agriculture sector recovered its production potential after the drop registered in the year 2012 caused by drought and after the restrictions imposed by the Russian Federation on Republic of Moldova agri-food imports and exports, between the years 2013–2014. The direct significant correlation model between the wheat and maize production and total agriculture production reveals the importance of these crops to the Moldavian economy. It is recommended to focus the governmental financial support on improving the technologies that lead to a superior productivity of these crops in order to improve the economic sustainability of the Moldavian agriculture sector.

Future avenues of research should target the improvement of the present analytical framework by expanding the current dataset with other indicators, describing the existing relations between the economic, social and environmental systems so it could be used as an efficient tool for identifying better agricultural sustainable strategies. As such, since the complexity of the analytical framework is positively influenced by the number of analyzed parameters and the dataset size, other relevant economic indicators like liquidity, stability, autonomy/dependence, return on equity, return on assets, operating expense, debt to assets, debt to total output, depreciation expense, gross margin, farm ability to generate income, diversification of the production and, also, business multi-functionality, are recommended to be added in the analysis.

Moreover, the Republic of Moldova's governance and institutional capacities can provide relevant parameters in enhancing the agriculture sector's analytical framework, so the political system could be also linked to the economic, social and environmental systems. Lastly, as more data would be available, several other modelling techniques could be used. For example, future studies could consider multiple linear regressions enhanced by using lasso, ridge or elastic net regularization methods; support vector regressions; or an extended array of ensemble learning prediction methods like Ada Boost, Gradient Boosting, XGBoost, Bagging, GBM and CatBoost.

**Author Contributions:** D.S.C. contributed with the study conception, data acquisition, drafting of manuscript, respectively data analysis and interpretation and literature review. M.M.T.R. contributed with data and models interpretation, respectively critical revision. C.G.Z. contributed with data analysis procedures and validations. A.T.R. contributed with drafting of the manuscript, critical revision and data interpretation. G.A.Z. contributed with drafting of the manuscript and data acquisition. D.N. contributed with critical review and drafting of the manuscript. S, .-M.P. contributed with the study conception, data and models interpretation, critical revision, literature review and drafting of the manuscript. All authors have read and agreed to the published version of the manuscript

**Conflicts of Interest:** The authors declare no conflict of interest.
