Determinants of efficiency in agriculture in CEE countries

Based on selected data acquired from Eurostat database the output efficiency of agriculture in CEE countries at national level is evaluated. DEA approach is applied in order of Malmquist productivity index calculation. Analysis includes one output variable (Total Agricultural Output) and three input variables (Total Labour Input, Total Utilized Agricultural Area and the Consumption of Inputs). In addition, the paper identifies the variables (average farm size, average subsidies per farm and education of managers in agriculture), which affect technical efficiency in agriculture.


Introduction
Efficiency analysis can be applied in the level of countries, as well as at the level of selected industries.Agriculture has been one of the most important sectors of national economy in the most countries of the world or at least the sector with the longest history.Its importance among European countries is highlighted by the policy framework of Common agricultural policy (CAP), which is aimed on rural and agricultural development.CAP has been working toward the modernization and sustainability of the sector since 2008.

Literature review
The paper points out the researches which conducted a research of the performance of agriculture in individual countries through the common performance indicators, the Technical Efficiency (TE) and Total Factor Productivity.Serrão (2003) examined the sources of productivity growth and the productivity differences among countries and regions over the period 1980-1998.Research included fifteen European Union countries and four East European countries.The study was based on data collected from the Food and Agriculture Organisation of the United Nations.An approach based on Data Envelopment Analysis was used to provide information on the peers of the (inefficient) country and to derive the Malmquist productivity indices.Author claimed an annual growth in Total Factor Productivity of 2.2 percent, where a major contributing factor was technical change.Negative growth in efficiency change was observed in a couple of years.France posted the most spectacular performance, with an average annual growth of 3.6 percent in Total Factor Productivity over the observed period.Bel-Lux and Denmark had a similar performance.Portugal posted a Total Factor Productivity growth decline.Turning to the performance of the five European regions defined in his research work, the Eastern European region (consisting of Romania, Bulgaria, Poland and Hungary) was the major performer, with an annual Total Factor Productivity growth of 2.6 percent.Akande (2012) measured the Technical Efficiency and TFP growth of agricultural holdings in the EU-15 region over the 11 years period by Data Envelopment Analysis.Author observed an average technical efficiency of 87% for the EU-15 region as a whole.Author divided the EU-15 region into four regional groups.Western European Region was more efficient with the highest average technical efficiency of 95% while Central European Region shared the same technical efficiency level of 85% with Southern European Region.Meanwhile, the Northern European region was the least technically efficient (84%).The annual average of 3% and 4% TFP growth rate was observed for all the regions in the EU-15.Study observed that TFP growth rate in the four regional groups were being driven by technology progress (technical change) and a decline in efficiency change particularly between the years 1999 and 2002.Subsequently up till 2005, the growth rate was driven by catch-up (efficiency change) while there seem to be technological regression.Fogarasi (2006) analysed efficiency and TFP in Hungarian sugar beet production.For 2004 and 2005 efficiency and TFP were calculated by Data Envelopment Analysis and by a Malmquist index respectively.Between 2004 and 2005 the average technical efficiency was very stable, around 0.80 for CRS efficiency and 0.87 for VRS efficiency.Between years TFP increased by 9%.The main reason for the observed TFP increase was technical progress of 8%, while technical efficiency played a limited role in improving the performance of sugar beet production.At the same time there was a clear convergence which can be identified and thus improving efficiency scores among individual holdings.Coelli et al. (2006) obtained detailed information on the TFP growth of arable farms in Belgium over a 16 years period from 1987 to 2002.The TFP measures were calculated using a Malmquist indices and DEA methodology.The results indicated an average annual rate of TFP change of 1.0% per year, with most of this being due to technical change.An inspection of the TFP change indices before and after the two CAP reforms (in 1992 and 2000) indicated that these reforms had had no discernible effect upon TFP trends.
Following authors conducted a research based on various econometric models to calculate and evaluate the significant drivers of efficiency in agricultural sectors among the studied economies.Authors considered direct investment into agriculture, subsidies, inputs quality, openness of the country and farm size as significant drivers of efficiency.Nowak et al. (2015) concerned the measurement of the technical efficiency of agriculture in the 27 European Union countries in 2010.The research was conducted based on the output-oriented DEA method assuming variable returns to scale.Moreover, in the study, the factors affecting technical efficiency were identified, and the econometric modelling of their impact was performed with the use of the Tobit model.Authors claimed that across the 27 EU Member States, the level of the technical efficiency of agriculture was diverse and the difference between the states with the highest and the lowest efficiencies was 40%.Cyprus, Denmark, Greece, France, Spain, the Netherlands, Luxembourg, Italy and Malta were identified as the countries with the thoroughly technically efficient agriculture.In turn, the least technically efficient agriculture was observed for the Czech Republic, Lithuania, Hungary, Ireland, Latvia and Slovakia.Authors considered the soil quality, the age of the head of the household and the surcharges for investments as main determinants of agriculture in observed countries.

Domanska et al. (2014) measured the agricultural total factor productivity change in 27
European Union countries.The analysis included the years 2007 2011 and the research was conducted based on Malmquist productivity index completed with decomposition on technological changes and technical efficiency changes.In addition, the determinants of total productivity were identified in the study, and econometric modelling of their effect on TFP was conducted using Tobit model.The study demonstrated a small increase in agricultural TFP for the whole sample of 27 EU countries over the examined period.The reason of this increase was mainly the changes in technical efficiency.An effect of technological changes was in turn relatively low and of negative character.The analysis of the factors determining TFP changes demonstrated that factors like: percentage of farm managers with complete agricultural education, average farm area and economy openness measured as a ratio of total export to total import were the stimulants.In turn, the variable like the share of farm managers aged above 55 http://dx.doi.org/10.15414/isd2016.s12.11years appeared to be dissimulated.It is worth to emphasize that soil quality, additional payment to investments and number of students' in agricultural and related fields of study was insignificant from the total productivity changes point of view.Economic performance of European farms is strongly dependent on subsidies and public supports in form of direct payments.In fact there are some countries like Finland, Slovenia and Slovakia where the sum of total subsidies is greater than agricultural value added.According to Matthews (2014) those are the countries in which the agriculture does not add anything to GDP and farming is a pure consumer good financed by the taxpayer in return for the non-economic values.Bojnec and Latruffe (2013) investigated the links between size, subsidies and performance for Slovenian farms over the period from 2004 to 2006.Authors' analysis revealed that both preand postaccession farms' performance measured in terms of technical efficiency was positively related to farm size in Slovenia.Authors also found that small farms are less technically efficient but more allocative efficient and profitable.The persistence of small farms in Slovenia may be associated with the provision of generous subsidies, which was negatively related to farms' technical efficiency but positively related to their profitability.Latruffe et al. (2008) analysed the impact of public subsidies on farms efficiency in EU new member states before and after accession to the European Union.Four countries were considered: Hungary, the Czech Republic and Slovenia, who acceded to the EU in 2004, and Romania, whose accession was in 2007.The study found that subsidies had a negative impact on farm technical efficiency in Hungary over the period 2001-2005, in the Czech dairy corporate sector over the period 2000-2004, in Slovenia over the period 1994-2003, and in the Romanian crop sector in 2005.

Data and Methods
Model works with one outputoutput of agricultural industry for each country in millions EUR.This output is produced as a result of set of inputs.Land in form of utilized agricultural area (UAA) measured in 1000 ha, employment in sector of agriculture in 1000 persons and consumption of inputs are considered while calculating TE and TFP.Consumption of inputs involves data on the amounts of seeds and reproductive material, energy and lubricants, fertilizers and soil improvers, crop protection products and pesticides, veterinary expenditures, animal feed, maintenance of machinery, maintenance of buildings, agricultural services, other products and services.Consumption of input is measured in millions EUR.We used aggregate data gathered from the EUROSTAT database.For the calculation of Malmquist index the Stata 12.0 statistical program was used.
Table 1 describes the descriptive statistics of the data used.The vast variation of variables was found.

Data Envelopment Analysis
The employment of DEA models enable to measure efficiency involving multiple inputs and outputs.It is based on seminal work of Farrell (1957) and it is a non-parametric approach toward efficiency measurement using linear programming, accounting for multiple outputs and inputs.These models can be constructed either as output oriented (maximization) or input oriented (minimization).Input-oriented models refer to the amount by which all inputs could be proportionally reduced without a reduction in output, while output-oriented models answer the question by how much can output quantities is proportionally expanded without altering the input quantities (Coelli, 1995).Uses of input or output oriented model provide similar values under constant return to scale but are unequal when variable return to scale is assumed.
In the case of efficiency we employ output-oriented model with CSR in form: where φ is efficiency rate for each decision-making unit (DMU, CEE state in this case), λ refer to linear combination of inputs and outputs, Y is vector of outputs and X vector of inputs.The condition λ≥0 indicates CSR.

Malmquist productivity index
Malmquist index is used to compare the development of agricultural performance over the time periods.It is the measure of productivity change and it decomposes this productivity change into technical change and technical efficiency change (Coelli, 1995).Malmquist productivity index is a geometric mean of two production functions based on the distance functions, as follows: which can be further adjusted to: where the outputs and inputs are yt,xt in the basic period, yt+1,xt+1 are output and output in the next period.Notation  0  and  0 +1 represents distance of the DMU in the basic and next period.The resulting product of Malmquist index (M0) is change in productivity.It includes change of technical efficiency and technological change.Whenever the M0 >1 it signalizes the enhanced productivity. http://dx.doi.org/10.15414/isd2016.s12.11

Results and Discussion
Following results was based on the calculation of Malmquist indices across CEE countries.The survey was conducted in the period from 2007 to 2012.First, we calculated the output efficiency scores under CRS (Constant Returns to Scale).The score higher than 1 indicates, the DMU is inefficient relative to the other DMUs.As you can see in table 2, the average efficiency score in CEE countries (2007-2012) was 1.0545 which shows, that if countries tend to be effective, they have to increase they output in average by 5.45 % (by the constant primary input level) to be efficient.Furthermore, the output efficiency scores varied across CEE countries and in the investigated period, as well, but they were relatively close to the level of average CEE efficiency score.The most efficient countries were Hungary, Romania and Slovenia with efficiency score 1.0000 and the least efficient countries were Latvia (1.2117) and Slovakia (1.1005).The best performer country during the period was Slovakia, which efficiency score in 2007 was 1.1812, while the efficiency score in 2012 was 1.0000, which indicates that Slovakia became effective relative to other countries.

Table 3: Efficiency Scores
Note: effdenotes output efficiency score under CRS Source: own processing based on Eurostat data Now, we turn to the decomposition of TFP growth in agricultural sector across CEE countries (Table 3).Malmquist productivity index (change in TFP, tfp_ch) can be decomposed into efficiency change (eff_ch) and technical efficiency change (tech_ch).Average TFP change in CEE region was 0.9914 which indicates a decline in TFP by 0.86 % in average over the observed period.This decline was associated by the decline in efficiency change (0.9946) and also in the decline of the technical efficiency change (0.9968).The best performer countries in terms of TFP change were Latvia (1.0141), Hungary (1.0097) and Bulgaria (1.0078).The least performer states were Slovakia (0.9578) and Czech Republic (0.9667).The catch up effect was observed only in case of Latvia, where the average efficiency change exceeded 1.The increase in the technical efficiency change was slightly up to 1 in case of Bulgaria and Hungary.Furthermore, we analysed, the common determinant of efficiency among efficient DMUs and among inefficient DMUs.We took average utilised agricultural area (a_uaa), average subsidies per farm (a_subs_pf) and the Training of the managers into consideration.Based on Eurostat database, the Training of the managers was divided as follows: i; managers with practical experiences only (pract_e), ii; managers with basic agricultural training (basic_t), iii; managers with full agricultural training (full_agri_t).We divided the countries into three clusters according to their efficiency scores under CRS (CRS_eff) as follows: 1. cluster: CRS_eff=1; 2. cluster: 1<CRS_eff<1.05;3. cluster: 1.05<CRS_eff.Efficient countries owned in average smaller utilised agricultural area than the inefficient ones and the average value of subsidies per farm was much lower than in the case of efficient countries than in inefficient countries.As the efficiency grow, grow the share of managers with only practical agricultural experiences.The share of managers with full agricultural training was about 15 percentage points higher in inefficient countries than in efficient ones.The efficient countries had common the following: i; small average utilised agricultural area, ii; low value of average subsidies per farm and, iii; high share of managers with only practical experiences.

Country
In terms of output efficiency under CRS we observed comparable results like authors mentioned in the literature review.The agricultural sectors among investigated countries didn't differ rapidly among themselves.We found a slightly decrease in TFP change, efficiency change and technological efficiency change.Average TFP change in CEE region was accounted at 0.9914, so there is a discrepancy between our findings and the findings of the authors mentioned in literature review (Akande, 2012;Fogarasi, 2006;Domanska, 2014)

Conclusion
The aim of this paper was to evaluate the performance of agriculture in ten CEE countries.We employed output oriented DEA model to calculate efficiency scores of CEE countries.The average efficiency score in CEE countries (2007-2012) was 1.0545.Analysis revealed that the output efficiency scores varied across CEE countries and in investigated period, as well.The most efficient countries were Hungary, Romania and Slovenia and the least efficient countries were Latvia and Slovakia.Furthermore, the Malmquist productivity index was used to measure the change in TFP over the observed period.We found a slightly decrease in the level of TFP, which was associated in regress in efficiency change and in technical efficiency change.The most noticeable increase in TFP change was observed for Latvia, Hungary and Bulgaria.Furthermore, we analysed the common determinants of efficiency among efficient countries and among inefficient countries.We took average utilised agricultural area, average subsidies per farm and the training of the managers into consideration and found that the efficient countries had common the following (in average): i; small average utilised agricultural area, ii; low value of average subsidies per farm and, iii; high share of managers with only practical experiences.

Table 3 : Decomposition of TFP
. Our findings in case of technological efficiency change are in line with the findings of Akande (2012), who also observed a decrease in technological efficiency change.