Bio-based and Applied Economics BAE Copyright: © 2023 Krasniqi N., Blancard S., Gjokaj E., Ottaviani Aalmo G. Open access, article published by Firenze University Press under CC-BY-4.0 License. Firenze University Press | www.fupress.com/bae Bio-based and Applied Economics 12(3): 183-195, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Citation: Krasniqi N., Blancard S., Gjokaj E., Ottaviani Aalmo G. (2023). Modelling technical efficiency of hor- ticulture farming in Kosovo: An appli- cation of data envelopment analysis. Bio-based and Applied Economics 12(3): 183-195. doi: 10.36253/bae-14693 Received: May, 9 2023 Accepted: August, 1 2023 Published: October 15, 2023 Data Availability Statement: All rel- evant data are within the paper and its Supporting Information files. Competing Interests: The Author(s) declare(s) no conflict of interest. Editor: Davide Menozzi ORCID NK: 0000-0003-4570-3239 SB: 0000-0001-5996-2368 EG: 0000-0002-1649-4098 GOA: 0000-0001-8357-4971 Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Nol Krasniqi1,2,*, Stéphane Blancard1, Ekrem Gjokaj2, Giovanna Ottaviani Aalmo3 1 CESAER, INRAE, L’Institut Agro Dijon, Université Bourgogne Franche-Comté, F-21000, Dijon, France 2 Faculty of Life and Environmental Sciences, University “Ukshin Hoti”, Prizren, Kosovo 3 Department of Economy and Society, Norwegian Institute of Bio-economy Research, NIBIO, Ås, Norway *Corresponding author. E-mail: nol.krasniqi@uni-prizren.com Abstract. With a view to integration into the European Union, the efficiency and com- petitiveness of the Kosovo’ different sectors (including agriculture) must be improved. This paper assesses the technical efficiency (TE) of horticultural farms through Data Envelopment Analysis (DEA) applying output orientation. It was founded that the TE of these farms is positively affected by their size, with large-size farms presenting over- all higher technical efficiency. The research findings indicate that the degree of agri- cultural education does not have a significant impact on TE, whereas public assistance through subsidies and grants has a substantial and negative impact on TE, as con- firmed by statistical analysis. Keywords: technichal efficiency, horticultural farming, data envelopment analysis. JEL codes: Q10, Q18, C14. HIGHLIGHTS • With a view to integration into the European Union, the efficiency and competitiveness of the Kosovo’ different sectors (including agriculture) must be improved • We use a model of Technical Efficiency of Horticulture Farming in Kos- ovo with application of DEA. • FADN data used on this study are from the years 2015 to 2019, in total 5 years in a row making the total observation 779. • The study’s findings reveals that the majority of farms in the sample show a technical efficiency level below 50%. • It was found that the TE of these farms is positively affected by their size, with large-size farms presenting overall higher technical efficiency. http://creativecommons.org/licenses/by/4.0/legalcode https://doi.org/10.36253/bae-14693 https://doi.org/10.36253/bae-14693 https://orcid.org/0000-0003-4570-3239 https://orcid.org/0000-0001-5996-2368 https://orcid.org/0000-0002-1649-4098 https://orcid.org/0000-0001-8357-4971 184 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. 1. INTRODUCTION A future integration for Kosovo to the European Union (EU) raises significant opportunities but also challenges for the country’s economy. One challenge is to improve the competitiveness of several sectors, including agriculture. According to latest agriculture census, Kosovo has 1.1 million hectares of land, out of which 53% is agri- cultural land (from which 54.3% belongs to permanent grasslands, 43.6% arable land, 1.9% permanent crops and 0.3% kitchen garden), 41% is forest, and 6% belongs to other land uses (KAS, Agriculture Census, 2015). Kosovo has traditionally supported with direct payment (subsidies) and through investment grants three main agricultural sectors: cereals, horticulture, and livestock, which are divided into 21 subsectors: 11 annual and per- ennial crops (cereals and horticulture), wine, and organ- ic products, and 10 livestock sectors and milk (Kostov et al. 2020). In Kosovo, the agriculture sector employs the high- est number of people, accounting for 34% of the total employment. This sector also makes a significant contri- bution to the country’s Gross Domestic Product, which was around 8% in 2019. Additionally, agricultural prod- ucts constitute 17% of the total export value (MAFRD, 2020). Although, 60% of the population lives in rural areas in Kosovo, they do not contribute much to eco- nomic growth. According to the Kosovo Ministry of Agriculture, Forestry and Rural Development (MAFRD), “only a limited number of farms are currently able to compete and grab a greater share of the EU and foreign market”. The low competitiveness of farms can be attrib- uted to several key structural factors, including the small size of most farms, land fragmentation, outdated build- ing and equipment design, and limited access to finan- cial resources (MAFRD, 2014). Furthemore, Kosovo con- tinues to have a relatively high volume of imported agri- cultural products, which make up approximately 10% of all imports. In Europe, Kosovo ranks among the highest importers of food per capita (ERP, 2018). In this context and to attain the European stand- ards, improving the competitiveness of the agricultural sector becomes paramount. One way to help agricul- ture go towards competitiveness in domestic and foreign markets is to improve the technical efficiency (TE) of each agricultural sub-sector. Technical Efficiency refers to the ability to achieve the highest possible output level from a given set of inputs or resources. It measures how effectively inputs are utilized to produce desired out- puts within a production process or system. It is a fun- damental concept in economics and plays a crucial role in various fields, including agriculture, manufacturing, healthcare, and public services. According to Koopmans (1951, as cited in Farrell, 1957, p. 255; Charnes & Cooper, 1985, p. 72) provided a definition of what we refer to as technical efficiency, stating that an input-output vector is technically efficient if increasing any output or decreas- ing any input can only be achieved by decreasing some other output or increasing some other input. In the context of agriculture, Technical Efficiency is particularly significant as it directly impacts food production, resource utilization, and sustainability. By measuring and improving Technical Efficiency, policy- makers, farmers, and stakeholders can make informed decisions, allocate resources effectively, and drive agri- cultural development. In this study we focus on the horticultural farms from FADN data, which includes TE for vegetables cul- tivated indoor in greenhouses and vegetables cultivated outdoor. In comparing the 2019 total share of agricul- tural crops’s production to 2018, 2019’s vegetables lead with the highest percentage 33.4%, followed by fodder crops, cereals, fruits and others (MAFRD 2020). Accord- ing to the green report from MAFRD (2020) the total area cultivated with vegetable during 2019 was 18,911 ha. The crops that dominate the largest area in 2019 were potato (20%), pepper (16%), beans (15%), pumpkin (13%), onion (7%) and watermelon (6%). From the total area with vegetables, the different forms of horticulture in Kosovo, with the largest part are produced in open field. In percentage, the main area used for horticulture is in the open field with 83.5% followed by garden with 11.3% and vegetables cultivated in greenhouses with 5.2% (MAFRD, 2020). Following the introduction, section two presents a comprehensive review of the existing literature. Section three provides a detailed explanation of the research methods employed, while section four elaborates on the data utilized for estimating efficiency. Moving forward, the fifth section presents the results of the technical efficiency analysis and identifies the factors that influ- ence it. Finally, in the sixth section, the paper concludes with a summary of the analysis and discusses the policy implications derived from the findings. 2. LITERATURE REVIEW Despite the fact that there is limited literature that demonstrates the significance of measuring techni- cal efficiency in Kosovo’s horticultural sector, there are numerous global studies that explore efficiency in this area, Iráizoz et al. (2003) measured the TE of horticul- https://doi.org/10.36253/bae-14693 185Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 tural production in a sample of Spanish farms. They discovered a significant resemblance between the two technical efficiency estimates. Other authors, Bozo- glu and Ceyhan (2007) assessed the technical efficiency of 75 vegetable farms involved in vegetable production and investigated the factors that contribute to techni- cal inefficiency in the Samsun region of Turkey. This study’s findings indicate that the technical efficiency of the sample vegetable farms ranged from 56% to 95% (82% in average) and was affected by schooling, experi- ence, credit use, participation by women, and that infor- mation score negatively affected technical inefficiency. On the other hand, factors such as age, family size, off- farm income, and farm size were positively related to inefficiency. Another study conducted by Clemente et al. (2015) focused on assessing the technical efficiency of citrus-producing properties in Sao Paulo State dur- ing the years 2009 and 2010. Their investigation involved conducting interviews with producers and employing both non-parametric data envelopment analysis (DEA) and econometric methods to determine the levels of technical efficiency and identify factors that influenced efficiency. The study’s findings demonstrated that a sig- nificant proportion of citrus-producing properties in Sao Paulo operated below optimal efficiency levels. Nota- bly, the factors of “producer schooling” and “experience as a rural producer” emerged as the primary drivers of increased efficiency. The mean technical efficiency score obtained from the study was 0.79, indicating the poten- tial for production growth while maintaining the cur- rent input proportions based on the product-oriented model. In a similar vein, a study conducted by Irz and Stevenson (2012) investigated the potential inverse rela- tionship (IR) between farm size and technical efficiency in Philippine brackishwater pond aquaculture. This paper employs a stochastic ray production function to examine the potential inverse relationship in Philippine brackishwater aquaculture, utilizing a cross-sectional sample of 127 farms. The distribution of efficiency scores spans the entire range, with an exceptionally low average value of 0.37. Farm size explains only 13% of the vari- ability in outputs that are not accounted for by physi- cal inputs, while 73% is attributed to unidentified fac- tors and 14% to random shocks. Although the findings of this study are significant for policy formulation, they present a rather negative outcome, as they indicate that variations in efficiency are influenced by unexplained factors. Consequently, further investigation and specula- tion are necessary to uncover the underlying reasons for the subpar average technical performance of farms. Previous studies about the technical efficiency in Kosovo mainly focused on livestock and the dairy sec- tor. For example, Sauer et al. (2015) analysed the effect of migration on farm TE and found migration a decreas- ing effect. More recently, Alishani (2019) investigated the effects of public support policies on technical efficiency in Kosovo, with 394 farms from FADN year 2014. To the best of our knowledge, few research deals with technical efficiency of horticultural farms in Koso- vo. Frangu et al., (2018) assessed the input efficiency of 136 greenhouse farms growing tomatoes and peppers at both the farm and regional levels. The research utilized a combination of linear regression and DEA methods to identify any external factors that impacted efficien- cy. The study concluded that technical efficiency scores varied between regions, and based on the structural and operational characteristics of the greenhouse farms growing tomatoes and peppers, it was found that there was a possibility for farms and regions with low techni- cal efficiency to enhance their input usage. While larger farm size is essential for achieving sustained higher productivity in the long term, techni- cal efficiency presents the most promising solution for enhancing productivity in the short to medium term and promoting the growth of Kosovo’s agricultural sec- tor. Vegetable production offers the best opportunity for producing viable incomes on small farms while adding significant value to the national economy. According to statistical data from MAFRD, total area for cereals is decreasing, while for vegetables, the area of cultivation is increasing. Kosovo has the quality of land to achieve this but experience shows that when the products are grown by large numbers of small farmers acting independently, without irrigation, greenhouses, cool storage, grading and packaging facilities, and sufficient processing capac- ity there will be a considerable amount of dumping on oversupplied markets at peak supply. This study not only measures efficiency, but it also examines the factors that influence efficiency, and uses this analysis to provide additional recommendations for policy. In order to achieve this objective, we employ a two-step method suggested by Simar and Wilson (2007). In the first step, we estimate the relative efficiencies using inputs and outputs and then analyse the effects of the exogenous variables on efficiency. As several authors (Iráizoz et al. 2003; Sauer et al. 2015; Wilson, 2001; Kari- mov, 2014; Latruffe, 2004; Theodoris et al., 2014; Gav- iglio 2021; Morrais, 2021; Alishani 2019), the exogenous variables are age, agricultural training of the manager/ holder, gender, irrigation system, altitude, area con- strains, total subsidies on crops, rented area (Iráizoz et al. 2003; Sauer et al. 2015; Wilson, 2001; Karimov, 2014; Latruffe, 2004; Theodoris et al., 2014; Gaviglio 2021; Morrais, 2021; Alishani 2019). https://doi.org/10.36253/bae-14693 186 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. Finally, the paper contributes to fill the research gap on efficiency in the horticulture sector in Kosovo. We focus on the farms from FADN data, which includes TE for vegetables cultivated indoor in greenhouses, and veg- etables cultivated outdoor. 3. METHODOLOGY Methodologically, we employ Data Envelop- ment Analysis (DEA) to assess the performance of a group of units. Based on the pioneering work of Farrell (1957), Charnes et al. (1978) developed the DEA model under the constant return to scale (CRS) assumption, and Banker et al. (1984) extended it under the variable returns to scale (VRS) assumption. DEA involves cre- ating a production frontier that illustrates the highest attainable output from inputs, and subsequently meas- uring the distance between each unit and the efficient frontier (Blancard and Hoarau, 2013). The best perform- ers’ group provides practical observations for construct- ing this frontier. The most efficient units are those clos- est to the frontier, and so the furthest from the frontier, the highest is the units’ inefficiency. Two approaches can be used to estimate technical efficiency (TE): parametric, which includes both sto- chastic and deterministic methods, and non-parametric, such as DEA. In agriculture and farming, each approach has its own advantages and disadvantages when it comes to measuring farm performance. Studies comparing parametric and non-parametric methods have revealed disagreements regarding these approaches, particularly in agriculture. Coelli (1995) reviewed literature on fron- tier function estimation and efficiency measurement and suggested potential applications of these methods in agricultural economics. Further to this debate Sharma (1999) compared two approaches in measuring efficiency of the swine industry in Hawaii and the study revealed the DEA method is a more robust approach for measur- ing efficiencies compared to the parametric approach, based on the obtained results. DEA is particularly suit- able for agriculture because it allows for the assessment of relative efficiencies among multiple decision-making units (DMUs) without requiring explicit functional form assumptions or knowledge about the underlying stochas- tic production function. It considers the best-practice frontier defined by the most efficient units, providing a benchmark for comparing and evaluating the efficiencies of other units. This is beneficial in the agriculture sector, which encompasses a wide range of production systems and practices, where the assumptions of a specific func- tional form may not hold universally (Fare et al. 1994). In our study, we utilized an output-oriented model to estimate TE, which was based on both on (variable returns to scale) and (constant returns to scale). The term “Decision-Making Unit” (DMU) is used to refer to any entity that is evaluated based on its ability to transform inputs into outputs. In our study, we use this term to refer horticultural farms. Our primary goal is to evaluate efficiency based on the assumption that a DMU can produce a greater amount of output by using the same level of inputs. To achieve this, we use an output- oriented model. We chose output orientation based on the challenges that the horticulture sector in Kosovo fac- es, as described in the first part of the paper. Moreover, as following numerous studies, we decomposed technical efficiency (TE) into pure technical and scale efficiencies from CCR (Charnes et al., 1978) and BCC (Banker et al.,1984) models to identify the sources of inefficiencies. Let us consider n farms producing s output from m inputs. For the evaluated farm o, the output-oriented DEA linear programming is written as follow: max ϕ Subject to ∑n j=1 λjxij≤xio i=1,2…,m ∑n j=1 λjyrj≥ϕyro r=1,2…,s ∑n j=1 λj=1(DEA–BCC) λj≥0 (DEA-CCR ) j=1,2,…n (1) where n, m and s are number of DMUs, inputs and out- puts, respectively. DMUj consumes xij of input i and pro- duces yrj of output r; λj are the weights assigned by the linear program, ϕ is the calculated efficiency. The summary of the results obtained from the envel- opment model interpretation is as follows: if ϕ* = 1, then the DMU under evaluation is a frontier point. i.e., there are no other DMUs that are operating more efficiently than this DMU. The DMU under evaluation is ineffi- cient. i.e., this DMU can either increase its output levels or decrease its input levels (Zhu, 2014). The results of DEA TEVRS model represent pure technical efficiency (PTE). Alternatively, DEA TECRS model represents the overall technical efficiency (OTE), which consists of two components: scale efficiency and pure technical efficiency. While comparing scores from both DEA TECRS and DEA TEVRS model, if a DMU has a different efficiency score that means that the particu- lar DMU has scale inefficiency. Scale efficiency can be obtained by: SE = = OTE/PTE (2) https://doi.org/10.36253/bae-14693 187Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 After obtaining the results from the two models, we employed bootstrapping in the nonparametric model to address potential scepticism regarding the use of DEA in agriculture. Non-parametric efficiency measures are often criticized for lacking a statistical basis. However, Simar and Wilson (1998) argued that nonparametric efficiency measures do indeed have a statistical basis, and used bootstrapping to analyze the sensitivity of non- parametric efficiency scores to sampling variation. To generate the bootstrap estimates, we utilized the algo- rithm proposed by Simar and Wilson (1998) in R stu- dio, which is a statistical computing software. We used B = 2,000 bootstrap replications, and set the bandwidth at h=0.014 based on empirical evidence from Simar and Wilson (1998) that suggests small values of h provide smooth density estimates that follow the empirical den- sity function, while large values of h yield over-smooth density estimates. 4. DATA 4.1. Data source The study uses data from farms covering the entire Kosovo. Kosovo has 7 administrative regions, but a nomenclature of territorial units for statistics has not yet been introduced. It is divided into two territorial levels: municipal and settlement level; it currently has 38 munici- palities and 1,469 settlements (MAFRD 2014). In hydro- graphic terms, Kosovo is divided into river basins: The Drini i Bardhë, Ibri, Morava Binqës and Lepeneci (KAS, 2019). This sector of vegetable production in Kosovo is one of the main branches of agricultural production whilst in some regions of the Dukagjini Plain, it represents the main economic activity (MAFRD 2014). The predominant approach to horticultural cultivation involves cultivating crops in open fields for the purpose of commercial produc- tion. Among the various types of crops, vegetable produc- tion is typically the most labor-intensive. Data employed in this study are extracted from the farm accountancy data network (FADN). The devel- opment of a sustainable FADN system in Kosovo has been a focus of effort over recent years. Funded by the European Agency for Reconstruction-EU, a FADN pilot project was launched in 2004 involving 50 farms. This network expanded to 159 farms in 2005, increasing the number of farms to 300 in 2008 and 402 in 2013 and 2014. In order to make an adequate selection of the sam- ple, the FADN team applied the stratified simple random sampling. Sampling is carried out by following three fundamental criteria, which include economic size, farm type, and region. These criteria conform to the stand- ardized FADN methodology in line with the European Commission’s guidelines.1, even though it was simplified to suit the specific situation of the country. The deci- sion was made to include around 1,250 farms, which is roughly 2% of all agricultural holdings, in the FADN survey in order to ensure that the sample is as repre- sentative as possible (MAFRD, 2020). To account for the possibility of some farms declining to participate in the survey, each entity involved included approximately two additional reserve farms. In order to assess the technical efficiency (TE) of the horticultural sector of Kosovo, FADN data used are from the years 2015 to 2019, in total 5 years in a row making the total observation 779 (table 2 in appendix). The number of farms is different from year to year, the reason is that some farmers refused to participate in the upcoming years, so there was a number of reserved farms of the same typology which was used in case of refusal, besides this some farms change the category during the five years’ period. 4.2. Inputs and output selection To measure the technical efficiency, we retained four inputs and one output. The chosen inputs are widely employed in the literature for measuring technical effi- ciency. The term “total labor” refers to the amount of work completed in a year, equivalent to a full-time job. This is measured in annual work units (AWUs), which rep- resent the amount of work performed by a person who is employed full-time on a farm. In Kosovo, the mini- mum annual working hours are considered to be 1,800, which is equivalent to 225 workdays of eight hours each. The second input is land or the utilized agricultural area expressed in hectares. It consists of the land in owner- occupation, rented land and land in share-cropping. The third input is total intermediate consumption, which includes total specific costs (including inputs produced on the farm) and overheads arising from pro- duction in the accounting year. The total specific costs included specific crop costs (fertilizers and soil improv- ers, purchased manure, crop protection products) and other specifics costs (labour and machinery costs and inputs, contract work, and machinery hire, current upkeep of machinery and equipment, motor fuels and lubricants, car expenses). Farming overheads include land improvements and buildings, electricity, heating 1 Council Regulation (EC) No 1217/2009 of 30 November 2009 setting up a network for the collection of accountancy data on the incomes and business operation of agricultural holdings in the European Community https://doi.org/10.36253/bae-14693 188 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. fuels, water, farm insurance, other farming overheads expressed in the euro. Finally, we consider one more input which is the average farm capital includes cash & equivalents, receiv- ables, other current assets, inventories, plants, land improvements, farm buildings, machinery and equip- ment, and intangible assets. Output is the total value of the crop products, and of other output expressed in Euros, including that of other gainful activities (OGA) of the farms. The table 2 (appendix) presents the descriptive sta- tistics of variables for farms together indoor and outdoor in the open fields2. On average, they produced output in value of 29,319 € for the year 2015 with 139 farms in the sample for this year. In 2016 with 150 farms the total out- put value was 28,404 € while in 2017, 162 farms produced on average 29,678 €. In 2018, 143 farms produced total output on average of 26,516 € while in the year 2019, 185 farms produced total output in value of 25,550 €. 5. RESULTS AND DISCUSSIONS 5.1. Efficiency results In our study VRS, CRS and SE were evaluated for horticultural farms. The number of farms is differ- ent from year to year, the reason is that some farmers refused to participate in the upcoming years, so there was a number from the list of a reserved farm of the same typology which was used in case of refusal, and besides this some farms change the category during the five years’ period. The summary of results is presented in the table 1. The year 2017 showed the highest efficiency score with a pure technical efficiency level of 0.72 for farms hor- ticulture indoor, which means that 28% can increase the output to reach the efficiency frontier. The majority of farms in the sample show a technical efficiency level of above 50%, besides the year 2018, which is with the level of efficiency of 45%. On the contrary, farms oper- ated in the open field have a lower efficiency score, with the largest efficiency level of 0.56 in the year 2017, while similar to horticulture indoor, also at the open field, the year 2018 has the lower efficiency score below the 50%. For the farms, horticulture indoor the highest average score on scale efficiency (0.92) was in the year 2017 while the lowest score (0.77) was in 2019. While for the farms in the open field the highest average score on scale effi- 2 Descriptive statistics for each of inputs and output variables for the three categories horticultural farms indoor, outdoor in open field are presented in table 3 and 4 in appendices. ciency (0.91) in the year 2017 and the lowest in the year 2015 and the lowest score of SE (0.74). From the aver- age aggregate results for farms (Indoor and open fields together), most of the farms in the sample show a tech- nical efficiency level that is less than 50%. The highest efficiency score with a pure technical efficiency level of 0.50, which means that 50% can increase the output to reach the efficiency frontier. The lowest pure technical efficiency score (0.36) is in the year 2018. The highest average score on scale efficiency (0.94) was in the years 2017 and 2018 while the lowest score (0.87) was in 2015. From the FADN methodology, farms are defined as being commercial only when they pass the Standard Out- put of 2,000 Euros. This implies that a commercial farm is able to provide the farmer with a sufficient level of income to support the welfare of his family. Thus, based on this classification the table 5 in appendices present the technical efficiency score categorized by economic sizes of farms. The large-size farms had overall higher techni- cal efficiency under the category 6 (100,000 - < 500,000). 5.2. Biased corrected efficiency scores Figure 1 to 10 present a graphical illustration of the distribution of farms (in appendices), using box plots to Table 1. Descriptive results of efficiency estimate for horticultural open field farm. 2015 Mean SD Min max no of farms PTE 0.52 0.28 0.11 1.00 121OTE 0.36 0.24 0.10 1.00 SE 0.74 0.21 0.10 1.00 2016 PTE 0.49 0.27 0.13 1.00 130OTE 0.36 0.23 0.12 1.00 SE 0.77 0.21 0.28 1.00 2017 PTE 0.56 0.26 0.13 1.00 144OTE 0.51 0.24 0.12 1.00 SE 0.91 0.14 0.31 1.00 2018 PTE 0.46 0.28 0.13 1.00 124OTE 0.39 0.24 0.10 1.00 SE 0.87 0.19 0.21 1.00 2019 PTE 0.48 0.28 0.14 1.00 164OTE 0.42 0.24 0.12 1.00 SE 0.89 0.16 0.17 1.00 Source: Author’s composition. https://doi.org/10.36253/bae-14693 189Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 facilitate the comparison among farms efficiency score in addition to the bias corrected. For each group of farms, the box represents the 50% mid-range values of efficiency scores and biases corrected. The interquartile range (IQR) is depicted by the length of each box, and the natural limits of the distributions are defined by the whiskers (which correspond to the mean ±1.5 (IQR)). Any outliers that fall beyond the natural limits are rep- resented by round circles. Each group of farms are deter- mined based on the bias corrected scores, the allocation of farms is different than the groups obtained based on the efficiency scores. Due to this different scope of the groups, we get these differences that appeared in the charts with red colors. Groups of farms are determined based on bias-corrected scores, for example group of farms with bias corrected scores from [0-0.10[ belong to group 1, while [0.10-0.20[ belong to group 2 and the rest until group 10. From the graph its clear the homogene- ity of farms in respect to efficiency scores within each group and those to be noted are in group 7 the differ- ences within group in each year under VRS and CRS. In addition, there are substantial differences between the two measures. On average, under this determination, in the year 2019 the efficiency score is 0,46, while under the boot- strap PTE model, it is only 0.39. Further for the same year, the OTE average score of TE is 0.39, while under the bootstrap OTE the score is 0.34. For instance, none of farms found entirely efficient under the PTE model and OTE model in each year (2015-2019) do not remain so after accounting bias-corrected scores through the bootstrap procedure. In this case, farmers should con- sider increasing the output while maintaining the same inputs. The results show there is a lot of space for using efficiently the inputs, area, labour, total intermediate consumption and average farm capital. 5.3. Determinants used to explain efficiency This section explains the second stage of technical efficiency study. The objective of this stage is to identify shared common characteristics among the most efficient farms. Two step procedures are used in the same scenar- io as Irazoz et al. (2003), so OLS and analyses of vari- ance are used to determine the link between efficiency and exogenous variables. Although the one-stochastic frontier method has a clear technical advantage over the two-step procedures, the two-step procedures may be more logically appealing for policy analysis and decision making because they directly relate the exogenous varia- bles to the observed efficiency performance of the firms. Furthermore, identifying the sources of inefficiency may aid in the development of policy recommendations (Yu 1998; Theodoris 2014). In this case we want to show the effects of exogenous variables in technical efficiency of horticultural production for further policy analysis in national level. Running DEA and creating a regression model with the DEA efficiency scores as the dependent vari- able and other possible factors as explanatory factors. This is a well-known two-stage technique that has been widely criticized for producing skewed results. However, it is frequently utilized, at the very least, to figure out which determinants are relevant. Contrary to a number of authors using Tobit, in second stage data envelop- ment analysis (DEA), McDonald (2003), is not a fan of using this model. In the two articles written by McDon- ald (2009, 2010), he describes OLS as a better replace- ment and a sufficient second stage DEA model. As he mentioned for many applied researchers, familiar and easy to compute, OLS may be the best option. Through- out the paper when referring to DEA, he dealt with the single output, output-oriented case. After comparing, in a stage 2 analysis, OLS, 2LT and 1LT marginal effects were similar. Output oriented frontiers are constructed under both the assumptions of variable returns to scale (VRS) and constant Return to Scale (CRS). The effect of the determinants is investigated with Ordinary Least Squares (OLS) regressions on each of the three TE scores for the period of 2015 to 2019 with total of 779 observa- tions specialized in horticultural farming. This method- ology is used by Latruffe (2017) to measure effect of sub- sidies on technical efficiency, contrary to us he used only variable return to scale (VRS) as our purpose is not only to measure the effect of subsidies on technical efficiency, in addition to that also we tend to measure other deter- minants which effect on technical efficiency on horticul- tural farming. Regarding the determinants that affect the efficien- cy scores, the most common variable used are farm size, the age of holder, qualifications, experience and special- ization of the farmer and combination of inputs (Iráizoz et al., 2003). In this study they found limitations to get this information in their sample data, while in our case, we could have accesses to raw data from FADN and get this information. We classified farm level data based on specialization of farms in horticultural (open field and indoor), match them with farm code and efficiency results of each farm. These similar determinants men- tioned above were used also from Sauer et al., (2015) who investigated the effect of migration on farm tech- nical efficiency in Kosovo. Another important study is to analyse the managerial drivers and practices due to https://doi.org/10.36253/bae-14693 190 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. business planning in farm and relation to the technical efficiency. Results from the research by Wilson on influ- ence of management characteristic on technical effi- ciency of wheat farmers in eastern England shows that, those farmers who seek information, have more years of managerial experience, and have a large farm are also associated with higher levels of technical efficiency (Wilson, 2001). Age and education are commonly cited as factors that may impact technical efficiency (Karimov, 2014). He stressed the important role formal education (uni- versity degree and educational background in agricul- ture) and informal education such as participating in workshops and seminars of farmers are associated with efficiency-improving results (Karimov, 2014). Other authors stressed that farmers that are more educated are considered more likely to be efficient farms associ- ated with higher scores of TE (Latruffe, 2004). And there is a strong significance between agricultural trainings and efficiency (Theodoris et al., 2014). Farmers who are younger may have a greater tendency to adopt innova- tive technologies aimed at reducing input usage. In con- trast, older farmers may have greater efficiency due to their extensive experience in addressing efficiency-relat- ed issues (Hadley, 2006). Exceptionally to these authors, Gaviglio (2021) found in his research that in fact, the level of education does not significantly improve the lev- el of efficiency (Gaviglio, 2021). In terms of the socio managerial aspect, we involved the variables age of owner/manager of farm, level of edu- cation in agriculture with only practical agricultural experience, basic agricultural training, full agricultural training, with the aim of seeing if the level and type of experience in agriculture affects the inefficiency. Other variables were: specialization of farm that produce veg- etables indoor and in open field, form of irrigation, irri- gation system used on the farm, not applicable (when no irrigation on the farm), surface, sprinkler or drip. In similar research on effects of irrigation in technical efficiency Morrais, (2021) results indicated that farms with irrigation had higher average technical efficiency compared to non-irrigators, which implies that irriga- tion technology has a significant effect on the efficiency gain for those groups. We also included variables on alti- tude of farms and the location, areas facing natural and other specific constraints. Also, we divided regions in two main plains of Kosovo, Dukagjini Plain and Koso- vo Plain, to see which farms are more efficient based on their location, although the plain of Dukagjini is well known for cultivating vegetables due to weather condi- tions, farm experience and tradition etc. However, the other part (mainly, the east part of the Kosovo plains) in recent years has benefited from increased investment in this sector based on data from Agency for Agricultural Development of Kosovo. Following other determinants, we included size to measure this we used the total output expressed in phys- ical units (kg) of vegetables produced by farms. Alike Iraizoz et al. (2003), they explain that they expected to obtain a positive coefficient for size, because horticultur- al production could present scale economies, in our case we follow this conclusion. We included the same deter- minants involving the combination of inputs. Additionally, we considered the total output coming from other gainful activities (OGA) directly related to the farm such as processing of farm products. We meas- ured this by the share of total OGA in total Output (%). We want to see if there is higher technical efficiency on farms that diversify their activities, and if large farms operate more efficient with higher share of OGA. Furthermore, we included as other determinants the share of subsidies to total output. Various stud- ies have investigated the effect of subsidies on farms’ technical efficiency, and in general the effect reported is negative. According to Minviel and Latruffe (2017) direct payments are common negatively associated with farm technical efficiency. In another study of the impact of support policies on technical efficiency of farms in Kosovo, subsidies had negative effect on technical effi- ciency (Alishani, 2019). Drawing from these related studies, Latruffe (2017) conducted a study on the impact of subsidies on technical efficiency with respect to envi- ronmental outputs. The study highlights that the policy implications are important, as a farm utilizing subsi- dies to increase environmental good outputs or reduce environmental bad outputs may have a lower traditional technical efficiency as compared to a farm receiving the same level of subsidies but using them solely for pro- ducing marketed outputs. Therefore, the effect of sub- sidies on traditional technical efficiency could be nega- tive for the former farm and positive for the latter farm. However, this case doesn’t necessarily fully apply to for Kosovo’s scenario because cross-compliance subsidies are still not introduced in national level support, but it still remains a recommendation from EU commission to Kosovo for initiating this form of subsidies. Moreover, other determinates we used are the share of paid labour to total AWU (annual working unit), and the share of rented land to total UAA (Utilized agricul- ture area). For these two variables results, a study by Alishani (2019) found out the paid labour to total AWU affects negatively the technical efficiency score, while the determinant of rented land to total UAA was insig- nificant. https://doi.org/10.36253/bae-14693 191Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Finally, we incorporated the factor of machinery and equipment into the analysis, which encompasses various items such as tractors, motor cultivators, lorries, vans, cars, and other farming equipment that are valued in euros. In the study by Sauer et al., (2015) results show that physical capital (machinery and farm equipment) decreases technical inefficiency, but this stands mainly because of outdated machinery and equipment. Our model with all determinates of inefficiency is presented on the table 8 (appendices). The adjustment shows corrected R squared coefficients of 0.38 for VRS, 0.44 for CRS and 0.26 for SE. Similar results are found by different author, Iraizoz et al. (2003) obtained coef- ficients of 0.31 and 0.68, and in addition, they found similarities to different authors as cited in (Parikh ,1995) who obtain a coefficient of 0.214, (Sharma, 1999,) with a coefficient of 0.23, and (Wadud and White, 2000) with a coefficient of 0.66. Concerning the socio managerial aspect in our results, the determinant age of the holder does not have significance with TE scores. Under the VRS and CRS, full agricultural training significantly does not affect the TE scores, while under the scale efficiency, there is a strong positive significance of full agricultural train- ing to TE scores. These results have relation to different reports that shows either formal or informal education in the field of agriculture remains insufficient com- pare to EU and neighbouring countries. According to the report from (National Research Programme of the Republic of Kosovo from 2010), research and techno- logical development (RTD) in agriculture is still a mar- ginal undertaking in Kosovo, despite the fact that agri- culture is an important economic sector. Compared to other countries in EU and the region, Kosovo has the lowest budget allocated for research per GDP, amount- ing to 0.1%. Only 0.19% of budget was allocated (0.05% of GDP), while in 2016, around 0.33% of budget (0.1% of GDP) (Kaçaniku, 2018). Considering the differences of horticultural farms if they operate more efficiently in open field or indoors in greenhouses, results show not any significance. In terms of irrigation system used on the farm, drip system shows significance on 10% under VRS and CRS. This system of irrigation is the most recommended to use in crops, as drip irrigation reduces deep percolation, evaporation and controls soil water status more precisely within the crop root zone (Singandhupe, 2003). Furthermore, Lattrufe and Desjeux (2014) indicate that farm size in Kosovo increases integration into the output market and that irrigated crop output is more marketable than livestock output. With respect to demographic contents in term of altitude, there is no particular significance. There is strong significance in scale efficiency to altitude below 300m and above 600m. For specific vegetables there are different requirements to produce yields, for example potato according to Haverkort (1990) it is shown that is adapted to a wide range of environments and hints are given on further exploitation of its potential in the vari- ous ecosystems. For the regional determinant, there is a negative significance under the scale efficiency in Duk- agjini Plain, although it is well known for cultivating veg- etables, this confirms our supposition that investment is being increased in the recent years in the east part of the Kosovo plain and the area covered by vegetables. As regard to farm size measured as total production in kg, the study shows a positive relationship with tech- nical efficiency under VRS, CRS and SE, with a strongly significance of 1%, in this case, the most efficient farms produce more in physical units. These results were also found by Iraizoz et al. (2003) in horticulture production in Spain. With respect to cultivation costs per hectare of land as a determinant relating to a combination of inputs, results show statistically no significant correlation with technical efficiency, contrary to Iraizoz et al. (2003) who found negative correlation indicating that higher culti- vation costs do not guarantee better results, in terms of efficiency. Following other determinants, the partial productiv- ity indices (output per unit of land and output per unit of labour), and the outcome are as expected, because the farms with higher productivity is an indicator for obtaining higher levels of technical efficiency. There is strong statistically significant under VRS, CRS and SE. With respect to subsidies, as we expected there, is a negative and statistically strong significant correla- tion between this determinant and technical efficiency under VRS and CRS. Public expenditure on Kosovo’s agriculture and rural development is based on two pil- lars; Grant aid to encourage investments in the means of production (the Rural Development Measures) and pay- ments for quantities of horticulture and livestock pro- duced (Direct Payments). In term of commercialization, direct payments have positive effects for horticultural and fruits farms (Kostov et al. 2020). Regarding size, the authors suggest that the impact on commercialization will be more significant if a larger number of semi-subsistence farms receive pay- ments based on their size. In Kosovo, eligibility require- ments for direct payments related to fruits and vegeta- bles (open field) have lower size thresholds compared to most other payments, making them more attainable for semi-subsistence farmers (Kostov et al. 2020). In every year, expenditure on direct payments has exceeded the https://doi.org/10.36253/bae-14693 192 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. amount contracted for investment grants, and overall accounts for 56% of the total public expenditure in the agricultural sector. At the outset, it should be recognized direct income support has a vital role to play in the man- agement of the transition from a production-oriented to a market-oriented food production sector. Last but not least, the lack of producer organization in the fruit and vegetable sector, lack of specialist advice and train- ing, and lack of support for innovation, are not being addressed. Continuing the following determinants, uti- lized agricultural areas rented by the holder does not have any significance on TE scores, while paid labour to total annual working unit is statistically significant under VRS only at 10%, contrary to Alishani (2019), this deter- minant affected negatively the technical efficiency score. With respect to other gainful activities in farms concerning the diversification of economic activities, and contrary to what we expected, the results shows this determinant presents negative and statistically signifi- cant correlation with TE scores, indicating that higher time spending on processing horticultural products does not guarantee a better TE score. While as we expected based on results, large farms operate more efficient under VRS and SE with higher share of OGA. Lastly, machinery and equipment’s decrease techni- cal inefficiency on farms, there is a negative and statisti- cally strong significant correlation with TE scores under CRS and SE. This finding are is similar as Sauer et al. (2015) on migration and farm technical efficiency evi- dence from Kosovo. This is consistent with our study’s observations regarding the continued use of old technol- ogy and machinery by farmers in Kosovo. With regards to five years of research data, from the results we can show that the year 2017 is strongly posi- tive correlation with TE scores under CRS, VRS and SE, and it has the highest average TE score compare to oth- er year. This mean that farms in horticulture operated more efficiently in the year 2017. 5.3. Implications (limitations of our study) A limitation of using the FADN (Farm Accountancy Data Network) for measuring technical efficiency is the potential for selection bias. The FADN database col- lects data from a sample of farms that voluntarily par- ticipate in the program. This self-selection process can introduce bias if participating farms differ systematically from non-participating farms in terms of their charac- teristics or behavior. Therefore, the findings based on the FADN data may not be representative of the entire agri- cultural sector, potentially limiting the generalizability of the results. To address these limitations, we employed appropriate statistical techniques, consider conducted robustness checks. The future research idea is to com- pare nonparametric methods with parametric methods for measuring technical efficiency scores in the agricul- ture sector in Kosovo. This study will offer valuable sci- entific insights for researchers and provide assistance to policymakers in addressing the issue of inefficiency. 6. CONCLUSION AND POLICY IMPLICATION The study’s findings reveals that the majority of farms in the sample show a technical efficiency level below 50%. The insufficient level of sore of TE implies that the remaining potential output could not be real- ized due to technical inefficiency. This means that 50% can increase the output to reach the efficiency frontier. The highest efficiency scores are in the region of Prizren and Prishtina, the biggest regions in Kosovo. Concern- ing the exogenous factors affecting the efficiency scores with respect to subsidies, there is a negative and statisti- cally strong significant correlation between this determi- nant and technical efficiency under VRS and CRS. Time spending on processing horticultural products does not guarantee a better TE score, although large farms oper- ate more efficient under VRS and SE with higher share of OGA. The results suggest that farmers should con- sider increasing the output while maintaining the same inputs. The findings indicate that there is considerable room for improvement for using efficiently the inputs, area, labour, total intermediate consumption and aver- age farm capital. In this respect, policy makers MAFRD3 should consider these low results of technical efficiency of farms to focus on a better program for extension ser- vices in order to promote better use of inputs. Vegetables are produced often in rather small areas and is very labour intensive; this fact fits the current sit- uation with plenty of underemployed family labour and unemployed. However, it seems that sooner than later, the abundance in the workforce will be gone, mainly because young people do not see agriculture as a busi- ness but just as an unwanted heritage. The mechaniza- tion is again low due to the small parcel sizes, but also due to missing financial means. Tractors, ploughs, trail- ers are old and just bigger farmers can afford machines like good sprayers or e.g., carrot-harvesting machines. The situation is improving when dealing with bigger farmers with modern orchards of 5 ha and more. Small farms struggle to access the market and to be commer- cialized, in this case considering the results from our 3 Ministry of Agriculture, Forestry and Rural Development. https://doi.org/10.36253/bae-14693 193Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 study it was found that the TE of these farms is positive- ly affected by their size with large-size farms presenting overall higher technical efficiency. For instance, small farms in horticultural sector should consider gathering in cooperatives. Kosovo’s agricultural policy is focused on semi- commercial and commercial farmers; the difference is that commercial farmers bring all their products to the market whereas the semi-commercial ones keep a sub- stantial part of their harvest for on-farm consumption. There seems to be a shortage in modern stor- age facilities for all kinds of vegetables; stor- age, cold storage (4°C) and cooling rooms (-15°C), and warehouses under a controlled atmosphere. Nonetheless, there is sufficient support in various forms such as investment grants for the processing industry, as well as subsidies for primary production. Further- more, there is a larger group of donors like the Euro- pean Commission (as funds cannot be used for IPARD because the ADA is not accredited yet), USAID, GIZ, SDC, and others. However, the performance of the veg- etable processing sector is not yielding satisfactory results, eventually as there was too much support and in an uncoordinated form. Investments should be focused on the direction of strengthening the primary produc- tion by indirect support through processing companies, and improvement of hygiene conditions and certifica- tions with food safety standards in order to have easy access to the EU market. REFERENCES Alishani, A. (2019). The Impact of Support Policies on Technical Efficiency of Farms in Kosovo. Review of Agricultural and Applied Economics, XXII (2): 46-54. https://doi.org/10.15414/raae.2019.22.02.46-54 Ankush, A., Singh, V., Kumar, V., Singh, D.P. (2018). Impact of drip irrigation and fertigation scheduling on tomato crop - An overview. Journal of Applied and Natural Science, 10(1), 165-170. https://doi. org/10.31018/jans.v10i1.1597 Blancard, S., Hoarau J.F. (2013). A new sustainable human development indicator for small island devel- oping states: A reappraisal from data envelopment analysis. Economic Modelling, 30, 623-635. https://doi. org/10.1016/j.econmod.2012.10.016 Bozoğlu, M., Ceyhan, V. (2007). Measuring the techni- cal efficiency and exploring the inefficiency determi- nants of vegetable farms in Samsun province, Tur- key. Agricultural Systems, 94(3), 649-656. https://doi. org/10.1016/j.agsy.2007.01.007 Charnes, A., Cooper, W.W., Rhodes, E. (1978). Measur- ing the efficiency of decisions making units. Euro- pean Journal of Operational Research, 2(6), 429-444. https://doi.org/10.1016/0377-2217(78)90138-8 Clemente, F., Lírio, V., & Gomes, M. (2015). Technical efficiency in Brazilian citrus production. Bio-Based and Applied Economics, 4(2), 165–178. https://doi. org/10.13128/BAE-13597 Coelli, T.J., Battese, G. (1995). A Model For Technical Inefficiency Effects in a Stochastic Frontier Produc- tion Function for Panel Data. Empirical Economics, 20, 325-332. https://doi.org/10.1007/BF01205442 Coelli, T. J. (1995). Recent Developments in Frontier Modelling and Efficiency Measurement. Australian Journal of Agricultura1 Economics, 39(3), 219-245. https://doi.org/10.1111/j.1467-8489.1995.tb00552.x Commission Implementing Regulation (EU) 2015/220 (2015) laying down rules for the application of Council Regulation (EC) No 1217/2009 setting up a network for the collection of accountancy data on the incomes and business operation of agricultural hold- ings in the European Union. European Commission, DG AGRI, B-1049 Brussels. https://eur-lex.europa. eu/legal-content/EN/ALL/?uri=celex:32015R0220 Commission, E. (2020). Commission Staff Working Document - Kosovo report 2020. Brussels. https:// ec.europa.eu/neighbourhood-enlargement/system/ files/2020-10/kosovo_report_2020.pdf Cooper, W.W., Seiford, L.M., Zhu, J. (2011). Data Envelop- ment Analysis: History, Models, and Interpretations. Dans W.W. Cooper, Seiford L.M., & . Zhu J., Hand- book on Data Envelopment Analysis, Second Edition, New York, Springer Science+Business Media, LLC., 41-70. https://doi.org/10.1007/978-1-4419-6151-8_1 Daraio, C., & Simar, L. (2007). Advanced Robust and Nonparametric Methods in Efficiency Analysis: Methodology and Applications. Springer Science & Business Media. EC.(2020). European Commission. Communication on EU Enlargement Policy, Kosovo2020 Report. Brussels, 6.10.2020 SWD (2020) 356 final. https://www.ecoi. net/en/file/local/2040144/kosovo_report_2020.pdf Erjavec, E., Tina, V., Rednak, M., Ciaian, P.,&. Lazdinis, M. (2020). Agricultural policies and European Union accession processes in the Western Balkans: aspira- tions versus reality. Eurasian Geography and Econom- ics, 62(1), 46-75. https://doi.org/10.1080/15387216.20 20.1756886 ERP. (2018). Kosovo’s Economic Reform Program 2018- 2020. Brussels. European Commission. https:// ec.europa.eu/neighbourhood-enlargement/system/ files/2019-04/kosovo_erp_2018-2020.pdf https://doi.org/10.36253/bae-14693 194 Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Nol Krasniqi et al. Fare, R., Grabowski, R., & Grosskopf, S. (1985). Technical Effi- ciency of Philippine Agriculture. Applied Economics, 17(2), 205-214. https://doi.org/10.1080/00036848500000018 Fare, R., Grosskopf, S. and Lovell, C.A.K. (1994) Produc- tion Frontiers. Cambridge University Press, Cam- bridge. https://doi.org/10.1017/CBO9780511551710 Färe, R., Grosskopf, S., & Margaritis, D. (2008). Efficien- cy and Productivity: Malmquist and More. Dans H. Fried, C. A. Knox Lovell, & S. Schmidt, The Meas- urement of Productive Efficiency and Productiv- ity Growth, New York: Oxford University Press, Inc., 522-638. Farrell, M.J. (1957). The measurement of productive effi- ciency. Journal of the Royal Statistical Society, 120 (3), 253-282. https://doi.org/10.2307/2343100 Frangu, B., Popp, J., Thomsen, M., & Musliu, A. (2018). Evaluating Greenhouse Tomato and Pepper Input Efficiency Use in Kosovo. Sustainability, 10(8), 1-14. https://doi.org/10.3390/su10082768 Gaviglio, A., Filippini, R., Madau, F.A. et al. (2021). Tech- nical efficiency and productivity of farms: a periur- ban case study analysis. Agric Econ. 9(11), 1-18. htt- ps://doi.org/10.1186/s40100-021-00181-9 Gjokaj, E., Leeds, S., Halimi, K. (2018). The impact of financial support program in creation on farm jobs. Bulgarian Journal of Agricultural Science, 24(6), 933– 941. Gjokaj, E., Kopeva, D., Krasniqi, N. & Nagy, H. (2021). Factors Affecting the Performance of Agri Small and Medium Enterprises with Evidence from Kosovo. European Countryside, 13(2) 297-313. https://doi. org/10.2478/euco-2021-0019 Haverkort, A. (1990). Ecology of potato cropping systems in relation to latitude and altitude. Agricultural Sys- tems, 32(3), 251-272. https://doi.org/10.1016/0308- 521X(90)90004-A Iráizoz, B., Rapún, M., & Zabaleta, I. (2003). Assessing the technical efficiency of horticultural production in Navarra, Spain. Agricultural Systems, 78(3),387-403. https://doi.org/10.1016/S0308-521X(03)00039-8 Irz, X., & Stevenson, J. (2012). Efficiency and Farm Size in Philippine Aquaculture. Analysis in a Ray Produc- tion Frontier Framework. Bio-Based and Applied Eco- nomics, 1(2), 175–198. https://doi.org/10.13128/BAE- 10556 Kaçaniku, F. (2018). The Situation of Research in Kosovo. Published by “Kosovo Education and Employment Network - KEEN” Project. https://www.research- gate.net/publication/334400746_The_Situation_of_ Research_in_Kosovo Karimov, A.A. (2014). Factors affecting efficiency of cotton producers in rural Khorezm, Uzbekistan: Re-examining the role of knowledge indicators in technical efficiency improvement. Agricultural and Food Economics, 2(7), 1-16. https://doi.org/10.1186/ s40100-014-0007-0 KAS(2015). Agriculture Census in The Republic of Koso- vo 2014 Finale Results. Prishtine. Kostov, P., Davidova, S., Bailey, A., Gjokaj, E., Halimi, K. (2020). Can Direct Payments Facilitate Agricul- tural Commercialisation: Evidence from a Transition Country. Journal of Agricultural Economics, 72(1), 72-96. https://doi.org/10.1111/1477-9552.12390 Latruffe, L., Dakpo, K. H.,Desjeux, Y., & Giffona J. H, (2017). Effect of subsidies on technical efficiency excluding or including environmental outputs: An illustration with a sample of farms in the European Union,” Working Papers 266259, Institut National de la recherche Agronomique (INRA), Departe- ment Sciences Sociales, Agriculture et Alimenta- tion, Espace et Environnement (SAE2). https://hal. archives-ouvertes.fr/hal-01583790 Latruffe, L., & Desjeux, Y. (2014). Perpetuation of subsist- ence farming in Kosovo: the role of farm integration in input markets. Post-Communist Economies, 26(1), 137- 148. https://doi.org/10.1080/14631377.2014.874671 Latruffe. L., Balcombe K., Davidova. S.,&Zawalinska K. (2004). Determinants of technical efficiency of crop and livestock farms in Poland. Applied Economics, 36(12), 1255-1263. https://doi. org/10.1080/0003684042000176793 Lingard, J. (2003). A Comparative Advantage Anal- ysis of Kosovan Agriculture. Post-Commu- nist Economies, 15(3), 417-434. https://doi. org/10.1080/1463137032000139089 MAFRD. (2014). Agriculture and Rural Development Pro- gramme 2014-2020. Prishtine. MAFRD. (2017;2018;2019;2020;2021). Green Report, Kos- ovo. Yearly Reports. Prishtinë. McDonald, J. (2009).Using least squares and tobit in sec- ond stage DEA efficiency analyses. European Journal of Operational Research, 197(2), 792-798. https://doi. org/10.1016/j.ejor.2008.07.039 McDonald, J. (2010). Efficiency in the Domesday economy, 1086: evidence from Wiltshire estates. Applied Economics, 42(25), 3231-3240. https://doi. org/10.1080/00036840802112430 Minviel, J. J., & Latruffe, L. (2017). Effect of public subsi- dies on farm technical efficiency: a meta-analysis of empirical results. Applied Economics, Taylor & Fran- cis Journals,49(2), 213-226. https://doi.org/10.1080/00 036846.2016.1194963 Morais, G. A. S., Silva, F. F., Freitas, C. O. de, & Braga, M. J. (2021). Irrigation, Technical Efficiency, and Farm https://doi.org/10.36253/bae-14693 195Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Bio-based and Applied Economics 12(3): 165-182, 2023 | e-ISSN 2280-6172 | DOI: 10.36253/bae-14693 Size: The Case of Brazil. Sustainability, 13(3), 1132. https://doi.org/10.3390/su13031132 Osmani, F., Gorton, M., White, J. (2013). Agricultural households, poverty and the rural labour market in Kosovo. Post-Communist Economies, 25(2), 241–252. https://doi.org/10.1080/14631377.2013.787756 Parikh, A., Ali, F., Shah, M.S. (1995). Measurement of economic efficiency in Pakistani agriculture. Ameri- can Journal of Agricultural Economics, 77(3), 675– 685. https://doi.org/10.2307/1243234 PPSE. (2020). Promoting Private Sector Employment, Assessment of support schemes in Kosovo [report]. RI/CC 1680 (2018). Committee for the Farm Accountan- cy Data Network (FADN). Farm Return Data Defini- tions from accounting year 2018. European Commis- sion, DG AGRI, B-1049 Brussels. RI/CC 1750 (2019). Committee for the Farm Account- ancy Data Network (FADN). Definitions of Variables used in FADN standard results. Brussels. European Commission, DG AGRI, B-1049 Brussels. Sauer, J., Gorton, M., & Davidova, S. (2015). Migration and farm technical efficiency: evidence from Kosovo. Agricultural Economics, 46(5), 629-641. https://doi. org/10.1111/agec.12159 Sharma, K. R., Leung, P., & Zaleski, H. (1999). Technical, allocative and economic efficiencies in swine produc- tion in Hawaii: a comparison of parametric and non- parametric approaches. Agricultural Economics, 20(1), 23-35. https://doi.org/10.22004/ag.econ.174734 Simar L., Wilson P.W. (1998). Sensitivity Analysis of Effi- ciency Scores: How to Bootstrap in Nonparametric Frontier Models. Management Science, 44(1), 49-61. https://www.jstor.org/stable/2634426 Simar L., Wilson P.W. (2007) Estimation and inference in two-stage, semi-parametric models of production processes. Journal of Econometrics,136(1), 31-64. htt- ps://doi.org/10.1016/j.jeconom.2005.07.009 Theodoridis, A., Ragkos, A., Roustemis, D., Arsenos, G., Abas, Z., & Sinapis, E. (2014). Technical indica- tors of economic performance in dairy sheep farm- ing. Animal.The international journal of animal bioscience, 8(1), 133–140. https://doi.org/10.1017/ S1751731113001845 Wadud, A., White, B. (2000). Farm household efficiency in Bangladesh: a comparison of stochastic frontier and DEA methods. Applied Economics, 32(13), 1665- 1673. https://doi.org/10.1080/000368400421011 Wilson, P., Hadley, D., Asby, C. (2001). The influ- ence of management characteristics on the techni- cal efficiency of wheat farmers in eastern England. Agricultural Economics, 24(3),329-338. https://doi. org/10.1111/j.1574-0862.2001.tb00034.x Yu, C. (1998). The effects of exogenous variables in effi- ciency measurement - A monte carlo study. European Journal of Operational Research, 105 (3), 569-580. htt- ps://doi.org/10.1016/S0377-2217(97)00076-3 Zhu, J. (2014). Quantitative Models for Performance Evaluation and Benchmarking - Data Envelopment Analysis with Spreadsheets. Springer. https://doi.org/10.36253/bae-14693 Towards a holistic approach to sustainable risk management in agriculture in the EU: a literature review Linda Arata1, Simone Cerroni2, Fabio Gaetano Santeramo3, Samuele Trestini4, Simone Severini5,* Modelling technical efficiency of horticulture farming in Kosovo: An application of data envelopment analysis Nol Krasniqi1,2,*, Stéphane Blancard1, Ekrem Gjokaj2, Giovanna Ottaviani Aalmo3 Incorporating expert knowledge in the estimate of farmers’ opportunity cost of supplying environmental services in rural Cameroon Claudiane Yanick Moukam1,*, Calvin Atewamba2 Farmers’ acceptance of a micro-irrigation system: A focus group study Maria Sabbagh1,*, Luciano Gutierrez2 Reducing food-related economic loss to improve food security and cattle trade in the Sahel: the case of agropastoral systems in Senegal Abdrahmane Wane1,*, Aliou Diouf Mballo2, Cabrelle Lauriane Dzoukou Homsi3, Pathé Diakhaté4, Rahimatou Memboup5