> %` z3bjbj 46̟̟ǖ(RRRROOO8P\\PD@RR(RRRVVV-}/}/}/}/}/}/}$hS}gVVggS}RhRRp~D9q9q9qg:|8RR-}9qg-}9q9qVw@xRQ2Omw1x~Dxtptxtx VA\9qU`DcVVVS}S}pdVVVDggggd:DGDGRRR|||Determinants of technical efficiency of crop and livestock farms in Poland
Laure Latruffe(i), Kelvin Balcombe(ii), Sophia Davidova(ii), Katarzyna Zawalinska(iii)
(i) INRA-Unit ESR, Rennes, and University of Paris 10, France
(ii) Imperial College, Wye Campus, UK
(iii) CASE Foundation, Poland
Abstract
Poland is one of the candidate countries for European Union membership that is currently experiencing acute structural problems within agriculture. This study analyses technical efficiency and its determinants for a panel of individual farms in Poland specialised in crop and livestock production in 2000. Technical efficiency is estimated with stochastic frontier analysis (SFA) and confidence intervals are constructed. Determinants of inefficiency are also evaluated. The SFA results are compared with results using Data Envelopment Analysis (DEA).
On average, livestock farms are more technically efficient than crop farms. For both specialisations, the size-efficiency relationship is positive i.e. large farms are more efficient. The SFA findings are generally supported by the DEA results. Soil quality and the degree of integration with downstream markets are highly important determinants of efficiency. The use of factor markets (land and labour) is important for crop farms, while livestock farms can rely on family labour and own land. Also, education is a constraint to efficiency particularly for crop farms.
Corresponding address
Laure Latruffe, INRA Unit ESR, 4 alle Bobierre, CS 61103, 35011 Rennes Cedex, France (latruffe@roazhon.inra.fr)
Determinants of technical efficiency of crop and livestock farms in Poland
Introduction
Poland is one of the candidate countries for European Union membership that is currently experiencing acute structural problems within agriculture. The prevalence of small-scale farming coupled with a high average density of agricultural employment per hectare, 3.2 times the European Union level (Pouliquen, 2001), creates doubts about the future viability of Polish agriculture in an enlarged Union. The public view of the Polish Ministry of Agriculture is that, in the medium term, only 400 to 500 thousand market farms are sustainable (European Commission, 1998). This figure has to be viewed relative to the total number of farm holdings in Poland which produce for the market. According to the last agricultural survey there were 1.5 million of such farms in 2000 (GUS, 2001). For this reason, shedding light on the efficiency of Polish farms and its determinants is of substantial policy relevance.
The objective of the paper is to analyse differences in technical efficiency between Polish livestock and crop farms and to evaluate various determinants of inefficiency for both specialisations. The determinants include various farm characteristics, including size, soil conditions, use of hired or rented factors and human capital characteristics. The stochastic frontier approach is used in conjunction with the construction of confidence intervals for the efficiency scores. The results are compared with those of a previous study employing Data Envelopment Analysis in order to assess the consistency and robustness of the results.
The paper is structured as follows. The next section reviews previous studies on determinants of technical efficiency in Central and Eastern European Countries (CEECs). Section 3 presents the stochastic frontier approach, whilst section 4 explains the methodology and describes the data set. Section 5 discusses the results and section 6 compares them with results from Data Envelopment Analysis. Section 7 concludes.
Previous studies about determinants of technical efficiency in CEECs
Three groups of variables are generally investigated in studies concerning the determinants of technical efficiency at the farm level. These are characteristics of the farm and technology employed, locational and environmental variables characterising the conditions for farming, and human capital variables. Farm characteristics always include farm size although there is little agreement how to measure the economic farm size. Lund and Price (1998) underline that there is no generally accepted measure in the economics literature. Various measures have been used in various contexts: output, sales, inputs and incomes. Standard man-days and standard gross margins have also been applied. Utilised agricultural area (UAA) can be a good proxy for size in a crop farm but is deficient in characterising the size of intensive livestock farming. Curtiss (2000), for Czech crop farms, and Morrison (2000), for Slovakian crop farms, measure the size in terms of hectares and find a positive relationship with the technical efficiency. For livestock farms, Morrison uses the number of feeding days in Slovakia, while Mathijs and Vranken (2000) employ the value of total output for crop and dairy farms in Hungary and Bulgaria. These authors also found a positive relationship between size and efficiency, except for Bulgarian crop farms. Generally, the findings about the size-efficiency relationship seem to confirm an L-shaped average cost curve. Larger farms attain economies of size by spreading fixed costs over more land and output, by getting volume discounts for purchased inputs, or by achieving a better market access and higher prices for large volumes delivered (Hall and LeVeen, 1978).
The organisational type of the farm is also commonly included as a farm characteristic in studies of transition economies (Curtiss, 2000; Mathijs and Swinnen, 2001). Debate about the efficiency of different post-reform farm structures will not be reviewed here since this study focuses only on those individual farms that are of the prevailing organisational form in Poland. However, the results of previous studies do not reveal any clear efficiency superiority of family farming over corporate type of structures, such as producer co-operatives or various types of farming companies, which are prevailing in some transition economies.
Several authors also investigate the effect of specialisation on technical efficiency. Mathijs and Vranken (2000) for Hungary and Bulgaria, Curtiss (2000) for the Czech Republic and Brmmer (2001) for Slovenia, all find that highly specialised farms are more technically efficient. Mathijs and Vranken (2000) report that Bulgarian crop farms that are more capital intensive are more efficient.
Efficiency variations between farms can also be explained by the farm locational and environmental characteristics. Farm location is important since farms may operate under different climate or altitude conditions, and different soil quality. Brmmer (2001) reports that Slovenian farms are less efficient in altitudes above 600 metres. Morrison (2000) also finds that agro-climatic region has a significant impact on the technical efficiency of Slovakian farms. As noted by Bhalla and Roy (1988), land quality is also an important factor. Unfortunately, existing efficiency studies for CEECs usually do not employ this variable, although van Zyl et al. (1996) adjust farm size for differences in land quality within regions in Poland. The physical infrastructure can also differ regionally. Munroe (2001) finds that Polish farmers with higher modernisation, in terms of use of electricity and gas heating, are more technically efficient.
Finally, much emphasis has been placed on the characteristics of on-farm human capital. The effect of the principal farmers age on efficiency has been extensively studied, but the conclusions are not consistent. Mathijs and Vranken (2000) and Munroe (2001) consider age as a proxy for farming experience and find a positive relation with technical efficiency in samples of Hungarian and Polish crop farms, but a negative effect in Bulgarian crop farms and Hungarian dairy farms. Mathijs and Vranken (2000) also include education and the share of women in the household. Both variables have a positive impact on efficiency in Hungary and Bulgaria. Finally, Brmmer (2001) notes that full-time farmers in Slovenia are more technically efficient than people involved in part-time farming.
Some of the aforementioned variables will be explored in the present study of the Polish farms. However, some additional variables will be included in order to characterise soil quality and integration to the market, up-stream and down-stream.
Stochastic Frontiers
Broadly, three quantitative approaches have been developed for the measurement of production efficiency: parametric (deterministic and stochastic), non-parametric based on Data Envelopment Analysis (DEA), and productivity indices based on growth accounting and index theory principles (Coelli et al., 1998). Stochastic Frontier Analysis (SFA) and DEA are the most commonly used methods. Both methods estimate the efficient frontier and calculate the firms technical efficiency relative to it. The SFA approach requires that a functional form be specified for the frontier production function. The DEA approach uses linear programming to construct a piece-wise frontier that envelops the observations of all firms. The frontier shows the best performance observed among the firms and it is considered as the efficient frontier. An advantage of the DEA method is that multiple inputs and outputs can be considered simultaneously, and inputs and outputs can be quantified using different units of measurement. Moreover, DEA allows calculating scale efficiency. However, a strong point of SFA in comparison to DEA is that it takes into account measurement errors and other noise in the data.
The stochastic frontier model was simultaneously proposed by Aigner et al. (1977) and by Meeusen and van den Broeck (1977). In contrast to a deterministic model it includes a random term representing the noise.
The model for the i-th farm is written as follows:
EMBED Equation.3 (1)
where
yi is the observed output quantity of the i-th farm;
f is the production function;
xi is a vector of the input quantities used by the farm;
b is a vector of parameters to be estimated;
vi is an error term, independent and identically distributed (iid) with N(0, (v2);
ui is a non-negative random term, accounting for inefficiency, iid, with N((i,su2), truncated to zero to ensure non-negativeness.
The technical efficiency of the i-th farm is given by EMBED Equation.3 and has a value between 0 and 1, with 1 defining a technically efficient farm. Since only the difference between both random terms wi = vi-ui can be observed, ui is predicted by its conditional expectation given the estimated value of wi: E[ui|wi] (Coelli et al., 1998). The conditional distribution of ui|wi is that of a truncated N((i*,s*2), where (i*=(wisu2-(isv2)/(su2+sv2) and s*2=su2sv2/(su2+sv2) (Jondrow et al., 1982).
The technical inefficiency effects ui are frequently estimated in a first step and the determinants of inefficiency are obtained in a second-stage regression. However, this can induce both bias and inefficiency in the estimates. Therefore, inefficiency effects are simultaneously conditioned on several specific factors and estimated using the parameterisation (Battese and Coelli, 1995):
EMBED Equation.3 (2)
where
zi is a vector of observable explanatory variables;
d0 and d are respectively a parameter and a vector of parameters to be estimated.
The construction of confidence intervals is straightforward given distributional assumptions about the random terms. TEi is a monotonic transformation of ui, therefore, the lower and upper bounds Li and Ui of the EMBED Equation.3 -percent confidence interval for ui|wi are translated directly into upper and lower bounds on TEi|wi = exp(-ui)|wi. They are given by (Horrace and Schmidt, 1996):
EMBED Equation.3 (3)
EMBED Equation.3
where the quantiles EMBED Equation.3 and EMBED Equation.3 are calculated as
Pr(Z>zL) = (a/2)[1-F(-(i*/s*)] (4)
Pr(Z>zU) = (1-a/2)[1-F(-(i*/s*)].
Z follows a standard normal distribution, therefore the quantiles are:
EMBED Equation.3 (5)
EMBED Equation.3
with F the standard normal cumulative density function.
Methodology and data employed
The production function for SFA was specified here as a Cobb-Douglas. Specifications such as the translog provide the opportunity to characterise the data in a more flexible way. However, with limited data it also tends to be overparameterised. Attempts to use a translog induced convergence failures possibly for this reason, hence the more restrictive Cobb-Douglas form was used. Using data from transitional economies, other researchers, e.g. Mathijs and Swinnen (2001), also opted for the Cobb-Douglas for a similar reason. A single output variable, total output in value, has been used. Four inputs have been included: utilised agricultural area (UAA) in hectares as a land factor, annual work units (AWU) as a labour factor, depreciation plus interests as a capital factor, and intermediate consumption as a variable factor.
As discussed in section 2, explanatory variables for inefficiency have been included in the model. The choice of variables in the final model has been subject to data availability and the requirement that the parameter g, defined as g = su2/(su2+sv2) (Singh et al., 2000), is significant. This parameter represents the share of the deviation from the frontier that can be attributed to inefficiency (the rest being attributed to noise) (Coelli et al., 1998). Most authors use hypothesis testing to evaluate the model (Curtiss, 2000; Morrison, 2000; Abdulai and Eberlin, 2001). This involves testing whether the coefficients d in (2) are significantly different from zero. Accordingly, the hypothesis H0 {dk=0 (k} is also tested (Battese and Coelli, 1995). As this study is focused on the comparison between livestock and crop specialisations, the same determinants for livestock and crop farms have to be chosen in order to compare their effects according to specialisation. Several models were tested in the light of this.
The need to specify a functional form, using SFA, is problematic because the paper attempts to analyse efficiency variations within the sector, according to size and specialisation. As pointed out by Mathijs and Swinnen (2001), small and large farms might operate with different production functions. Moreover, here livestock and crop farms are analysed separately, and it may not be valid to use the same production function for both sectors (this has been also noted by Abdulai and Eberlin, 2001, estimating efficiency of maize and beans producers in Nicaragua). Thus, as in Mathijs and Swinnen (2001), a comparison with results from DEA was carried out in order to evaluate the consistency and robustness of the results.
This study uses data from the annual survey of individual farms carried out by the Polish Institute of Agricultural and Food Economy (IERiGZ). Two sub-samples were selected according to farm specialisation, crop and livestock. The specialisation criterion was set at 65 per cent of the total output coming from crop or livestock production respectively. This threshold is not as high as in some other efficiency studies for transitional economies (e.g. Mathijs et al., 1999 applied 75 per cent for a Czech farm sample) but it is justified by the fact that the Polish farms are generally not highly specialised. According to this criterion, there were 222 crop and 250 livestock farms. However, similarly to the overall IERiGZ sample, both sub-samples are not representative for the Polish farm population since they are biased towards larger farms. For example, the 2000 agricultural yearbook of the Central Statistical Office (GUS, 2001) indicates that 23.8 per cent of Polish farms have an average land area between 1-2 ha and only 9.9 per cent have more than 15 ha. In the IERiGZ sample these shares are 1.3 and 48 per cent respectively. In both samples used in this analysis, although the smallest farms are included, the proportion of farms above 15 ha is even higher than in the overall IERiGZ sample (Table 1).
Table 1
Some basic characteristics of the sample farms are presented in Table 2. Crop farms have a larger mean UAA. Crop farms are also larger than livestock farms when size is measured by total assets. The more labour intensive character of livestock production is reflected in the fact that, when comparing size according to labour input (measured in AWU), the difference between crop and livestock farms disappears. Standard deviations show that livestock farms are more clustered around their means than crop farms according to the three size measures.
Table 2
Results
Technical efficiency estimates
Summary statistics for technical efficiency are presented in Table 3 and suggest that specialised livestock farms are, on average, more technically efficient than crop farms. Although the maximum score found within each sample is quite similar and close to the unity, the crop sample presents a minimum score that is much lower than for livestock farms and has a greater standard deviation. The results from previous empirical studies of production efficiency of crop and livestock farms in transition economies vary. Mathijs and Swinnen (2001), for East Germany, and Hughes (2000), for Hungary, report a higher average technical efficiency for livestock farms that is consistent with the estimates in this study, while Mathijs et al. (1999) and Mathijs and Vranken (2000) conclude that crop farms in the Czech Republic, Hungary and Slovakia are more efficient.
Table 3
Graphs 1 and 2 depict average technical efficiency scores with respect to 7 size intervals for both samples. Among livestock farms, large farms (over 15 ha) are the most efficient, and substantially more than the middle-sized farms. The curve for crop farms is G-shaped, clearly showing that the smallest farms (under 2 ha) have a very low efficiency. An Analysis of Variance (ANOVA) for the efficiency estimates showed that farm size had a highly significant impact on efficiency (Table 7 in Appendix 1).
Graph 1 Graph 2
Statistical robustness
The confidence intervals calculated as in (3) are very wide and overlap. The average width is 0.51 for the livestock sample and 0.53 for the crop sample. It shows that it is difficult to identify farms that are significantly less, or more, efficient than the average.
Determinants of inefficiency
Table 4 reports the results provided by the SFA model. The first part of the table presents the estimated parameters of the production function. Expected positive signs are found for land input (UAA) for livestock farms, and for labour input (AWU) and variable inputs (intermediate consumption) for both samples. The negative sign of the land input coefficient for crop farms is surprising. However, the parameter is not significantly different from zero. Mathijs and Swinnen (2001) also found an unexpected negative sign for labour input for East German farms, and attributed it to misspecification. However, such an argument is not compelling for the capital input (depreciation plus interest) for both samples since there is a highly significant negative sign. This may be due to the characteristics of the equipment in Polish farms. A large part of the machinery owned by the farms is very old and perhaps less productive. For example, the European Commission reports that only 5.9 per cent of the total number of tractors in Poland were made in the period 1991-1996, all the others were older (European Commission, 1998).
Table 4
The second part of Table 4 presents the results for the inefficiency effects model. Several models have been estimated and models that fitted well to one specialisation appeared poorly suited for the other specialisation. Although the coefficients of the final model presented in Table 4 are not significant for some determinants, for both samples the parameter g is highly significant and the hypothesis H0 is rejected.
Most of the signs related to the inefficiency determinants are as expected and are similar for livestock and crop farms. Although the size of the farms, represented by the total output, has little impact (the coefficient is close to zero), it positively affects efficiency. This confirms the relationship depicted on Graphs 1 and 2. The degree of market integration with down-stream markets, defined as the ratio of total revenue over total output, gives an idea about how much of its output the farm sells on the market. As expected, more commercially oriented farms are more efficient. Moreover the large absolute value of the coefficient suggests that this variable is an important determinant of efficiency particularly for crop farms. The negative sign linked to the soil quality index is also expected. The greater the soil quality, the higher the efficiency. It may appear strange that soil quality is a determinant of efficiency for livestock farms, but this might be due to the fact that feed is mostly produced on-farm. It is logical, however, that the parameter for the livestock sample is much lower than the one for crop farms, meaning that soil quality matters less for the livestock sector. For both specialisations older farmers are less efficient, but the small value of the coefficients suggests that the age is not a major constraint to efficiency. An interesting result is the different sign attached to the share of hired labour for livestock and crop farms. The more hired labour they employ, the more efficient crop farms are. The opposite is true for livestock farms. This suggests that crop farms need some external labour to perform specialised tasks while livestock production that requires closer and permanent attention is more efficient with family labour as an input. The latter, as the residual claimant, might be better motivated for good husbandry.
Comparison with Data Envelopment Analysis (DEA) results
Efficiency estimation and confidence intervals
The same data set was employed to estimate technical efficiency through DEA employing bootstrapping to construct confidence intervals (Simar and Wilson, 1998 and 2000). The efficiency estimates computed by DEA under constant returns to scale are used for the comparison with the SFA estimates (Table 5). As in the case of SFA, livestock farms are, on average, more technically efficient than crop farms. Not surprisingly, the average technical efficiency estimated by DEA is lower than the one estimated by SFA, as DEA attributes the entire distance from the frontier to inefficiency. Most of the studies using both DEA and SFA report such conclusions (Fraser and Kim, 2001; Brmmer, 2001). Mathijs and Swinnen (2001) for East German farms present SFA estimates greater than DEA. However they argue that this might be a result of misspecification of the production function. In the current study, the difference between the means of DEA and SFA efficiency estimates is large, 0.17 and 0.16 for livestock and crop farms. This is not a surprising result for data from transitional economies, where substantial measurement errors can be expected (Morrison, 2000). Brmmer (2001) also found large differences when estimating technical efficiency for Slovenian farms with DEA and SFA (0.32 and 0.29 for different years).
The standard deviations given by both methods are quite similar for crop farms, but different for livestock farms, where the standard deviation of SFA estimation is very small. This is due to the high mean of SFA efficiency score.
In contrast to DEA, no farm is found to be totally efficient using SFA estimation. This difference stems from the specifics of the methods applied. In DEA the frontier is determined by the best practice observed. Both methods suggest, however, that the minimum estimated efficiency for crop farms is much lower than for livestock farms, while the maximum for both samples is very similar. This shows that in crop production some of the farms lag far behind and they might reduce the mean efficiency of the sample.
Table 5
The shapes of the curves depicting the size-efficiency relationship are consistent between both methods. Livestock farms present a U-shaped curve, although this is less clear-cut when using SFA, while the curve for crop farms is G-shaped (Graphs 3 and 4) . An ANOVA showing the significance of size for DEA estimates is presented in Appendix 1 (Table 7). Both methods confirm that the largest livestock farms are the most efficient, while the large and medium-size crop farms are almost equally efficient. It is, however, difficult to draw clear conclusions for the smallest farms.
Graph 3 Graph 4
The confidence intervals constructed with bootstrapping are narrower than the intervals obtained with (3) in the SFA case. However, they are also wide. They present a samples average width of 0.10 and 0.13 for livestock and crop farms respectively. While this result does not undermine the overall findings about average efficiency, it does indicate that researchers should be cautious about making conclusions concerning the relative efficiencies of specific farms on the basis of point estimates of efficiency alone. The results herein may reflect variability that is not representative of other samples. However, it would be the contention here that farm specific efficiency scores should be treated cautiously in any DEA analysis, where standard errors for these efficiencies have not been calculated.
Determinants of inefficiency
A Tobit regression (Green, 1993) of DEA efficiency estimates on potential determinants was also undertaken and compared with the SFA results. A Tobit regression is required because the inefficiency estimates are truncated at 0 and 1. The results are presented in Table 6.
Table 6
The results are consistent with the SFA inefficiency effects model. The same sign and relative values for livestock and crop farms are found for the size variable, share of hired labour, degree of market integration, soil quality and age. Technology is represented in the determinants by the ratios of capital over labour and land over labour. Although the negative sign related to the land intensity can be expected, the positive sign related to the capital intensity ratio is surprising. It shows that more capital-intensive farms are less efficient, which is contradictory to our expectations and is not consistent with results reported for other countries in transition (Mathijs and Vranken, 2000). However, as mentioned previously, this might be explained by the obsolete equipment used by Polish farms. An additional variable representing a proxy for integration in factor markets has been included, the share of rented land. The result is similar to that for the use of external labour. Unlike crop farms, livestock operations are more efficient if they rely on their own land. This might be due to the fact that they can achieve efficient size with much less land than crop farms and renting land might not contribute to efficiency. The positive relationship between the education dummy and inefficiency is not surprising, bearing in mind that the value of 1 means no agricultural education. Lower educated farmers are less efficient. The relative values of the coefficient suggest that education is particularly a strong determinant for crop farm efficiency.
Conclusions
This study analysed technical efficiency and its determinants for a panel of individual farms in Poland specialised in crop and livestock in 2000. Technical efficiency was estimated with the parametric method, SFA, and confidence intervals were constructed. Determinants of inefficiency were also evaluated. The SFA results were compared with results generated by DEA.
On average, livestock farms are more technically efficient than crop farms. For both specialisations, the size-efficiency relationship is positive and, thus, large farms are more efficient. The SFA findings are generally supported by the DEA results. This study suggests that soil quality and the degree of integration with downstream markets are important determinants of efficiency. Also, the use of factor markets (land and labour) is found important for crop farms, while livestock farms could rely on family labour and own land.
The results of this study underline that policy measures that can facilitate a farm size increase might have beneficial effects on efficiency due to the positive size-efficiency relationship. In addition, for crop farms, the development of land leasing is also a substantial factor, as the share of rented land positively affects the technical efficiency. Therefore, policies stimulating the land market and assisting small farmers to move to non-agricultural employment may contribute to increased technical efficiency in the long-run.
There are two determinants that are probably the most important for technical efficiency, farmers education and market integration of the farm. The educational level of people engaged in individual farming in Poland is low. According to the 1996 agri-census only 16.2 per cent of people had a secondary education and 3.6 per cent had an education higher than secondary school (SAEPR/FAPA, 2000). This is a major constraint, impeding the adoption of new technologies. Education appears to be most important in the crop sector.
The Polish agricultural sector consists of subsistence and semi-subsistence farms that consume a large portion of their output within the household. The agri-census shows that, in 1996, 50 per cent of farms produced mainly or exclusively for their own needs (SAEPR/FAPA, 2000). This present analysis suggests that these farms might stay in the vicious circle of low technical efficiency and technological backwardness if they are not integrated into the market. From this point of view, support that can assist the commercialisation of semi-subsistence farms can be beneficial for efficiency gains and can facilitate the viability of these farms under an enlarged Union.
Acknowledgements
The authors are very grateful to Lech Goraj from IERiGZ, Warsaw, for providing access to the data. They also thank Iain Fraser for his help. The paper was written while Laure Latruffe was an academic visitor in Imperial College, Wye Campus, UK.
References
Abdulai, A. and Eberlin, R. (2001). Technical efficiency during economic reform in Nicaragua: evidence from farm household survey data, Economics Systems, 25, 113-125.
Aigner, D., Lovell, C. and Schmidt, P. (1977) Formulation and estimation of stochastic production function models, Journal of Econometrics, 6, 21-37.
Battese, G. and Coelli, T. (1995) A model for technical inefficiency effects in a stochastic frontier production function for panel data, Empirical Economics, 20, 325-332.
Bhalla, S. and Roy, P. (1988) Mis-specification in farm productivity analysis: The role of land quality, Oxford Economic Papers, 40(1), 55-73.
Brmmer, B. (2001) Estimating confidence intervals for technical efficiency: The case of private farms in Slovenia, European Review of Agricultural Economics, 28(3), 285-306.
Coelli, T., Battese, G., and Prasada Rao, D. (1998) An Introduction to Efficiency and Productivity Analysis, Kluwer Academic Publishers, Massachusetts.
Curtiss, J. (2000) Technical Efficiency and Competitiveness of the Czech agricultural Sector in late Transition The Case of Crop Production, Paper presented at the KATO Symposium, Berlin, 2-4 November 2000.
European Commission (1998) Agricultural Situation and Prospects in the Central and Eastern Countries - Poland, Directorate General for Agriculture, Brussels.
Fraser, I. and Kim, J. (2001) Estimating technical Efficiency: Does the Choice of Estimation Methodology matter?, Department of Economics and Finance, La Trobe University, Australia, August 2001.
GUS (2001) Yearbook 2000, Warsaw.
Greene, W. (1993) Econometric Analysis, Second Edition, Macmillan Publishing Company, New York.
Hall, B. and LeVeen, P. (1978) Farm size and economic efficiency: The case of California. American Journal of Agricultural Economics, 60(4), 589-600.
Horrace, W. and Schmidt, P. (1996) Confidence statements for efficiency estimates from stochastic frontier models, Journal of Productivity Analysis, 7, 252-282.
Hughes, G. (2000) Agricultural Decollectivisation in Central Europe and the Productivity of Emergent Farm Structures, PhD Thesis, Wye College, University of London, unpublished, mimeo.
Jondrow, J., Lovell, C., Masterov, I. and Schmidt, P. (1982) On the estimation of technical inefficiency in the stochastic frontier production function model, Journal of Econometrics, 19, 233-238.
Lund, P. and Price, R. (1998) The measurement of average farm size, Journal of Agricultural Economics, 49(1), 100-110.
Mathijs, E., Dries, L., Doucha, T. and Swinnen, J. (1999) Production efficiency and organization of Czech agriculture, Bulgarian Journal of Agricultural Science, 5, 312-324.
Mathijs, E. and Vranken, L, (2000) Farm Restructuring and Efficiency in Transition: Evidence from Bulgaria and Hungary, Selected Paper, American Agricultural Economics Association Annual Meeting, Tampa, Florida, July 30- August 2.
Mathijs, E. and Swinnen, J. (2001) Production organization and efficiency during transition: An empirical analysis of East German agriculture, The Review of Economics and Statistics, 83(1), 100-107.
Meeusen, W., and van den Broeck, J. (1977) Efficiency estimation from Cobb-Douglas production functions with composed error, International Economic Review, 18, 435-444.
Morrison, J. (2000) Resource Use Efficiency in an Economy in Transition: An Investigation into the Persistence of the Co-operative in Slovakian Agriculture, PhD Thesis, Wye College, University of London, unpublished, mimeo.
Munroe, D. (2001) Economic efficiency in Polish peasant farming: An international perspective, Regional Studies, 35(2), 461-471.
Pouliquen, A. (2001) Comptitivits et Revenus agricoles dans les Secteurs agro-alimentaires des PECO: Implications avant et aprs Adhsion pour les Marchs et les Politiques de lUE.
http//:www.europa.eu.int/comm/agriculture/publi/reports/ceeccomp/fullrep_fr.pdf
SAEPR/FAPA (2000) Stereotypes in the European Union concerning Polish Agriculture, Warsaw.
Simar, L. and Wilson, P. (1998) Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models, Management Science, 44(1), 49-61.
Simar, L. and Wilson, P. (2000) Statistical inference in nonparametric frontier models: The state of the art, Journal of Productivity Analysis, 13(1), 49-78.
Singh, S., Coelli, T. and Fleming, E. (2000) Measurement of Technical and Allocative Efficiency in Indian Dairy Processing Plants: An Input Distance Function Approach, Working Paper No. 3/2000, Centre for Efficiency and Productivity Analysis, The University of New England.
van Zyl, J., Miller, W. and Parker, A. (1996) Agrarian Structure in Poland: The Myth of large Farm Superiority, Policy Research Working Paper No 1596, The World Bank, Washington D.C.
Appendix 1
Table 7
Table 1: Distribution of sample farms according to area in hectares
Farm
specialisationSmall farms (ha)Medium farms (ha)Large farms (ha)1-2 2.01-55.01-77.01-10 10.01-15 15.01-50 >50Livestock (%)0.49.29.28.819.244.48.8Crop (%)0.98.18.114.49.532.926.1
Table 2: Basic characteristics of sample farms: Descriptive statistics
Livestock farmsCrop farmsUAA (ha)
Mean
Standard deviation
Minimum
Maximum
21.4
20.1
1.1
161.0
48.3
84.1
1.6
754.5AWU
Mean
Standard deviation
Minimum
Maximum
2.01
0.99
0.33
8.02
1.95
2.33
0.07
27.11Total Assets (000 PLZ)
Mean
Standard deviation
Minimum
Maximum
394.2
328.2
28.6
1,947.2
560.7
681.0
34.4
6,314.3
Table 3: Descriptive results of SFA efficiency estimates
Farm specialisationMeanStandard deviationMinimumMaximumLivestock
Crop0.88
0.730.06
0.130.74
0.380.99
0.97
Table 4: Results of the stochastic frontier estimation: Production function and inefficiency effects
Livestock farmsCrop farmsParametert-ratioParametert-ratioProduction function variables
Constant
UAA
AWU
Depreciation + Interest
Intermediate consumption
0.54
0.18
0.14
-0.11
0.86
6.97 ***
6.10 ***
4.19 ***
-3.78 ***
27.8 ***
0.80
-0. 02
0.17
-0.17
0.10
5.22 ***
-0.39
4.76 ***
-3.68***
15.6 ***Determinants of inefficiency
Constant
TKY\mq
& ' 0 Y
!l!]"^"##$()**|-}--ڸڳڭڧڸukjhR0J!UhR6B*phhRB*phhR6CJmH sH hRB*CJmH phsH
hR5CJ
hRCJ
hRCJ hR5hRCJmH sH hR0JmH sH
hR0JhRmH sH hR hRH*hRH*OJQJhROJQJhR5CJOJQJ(K & ' 0 Y
!lyyCEƀE{.$da$$a$dddx$dxxa$$dha$dhddt3x3y"g!$'( $dxa$CEƀE{.d
dx`dx
((-S......6/Z0124w`dx^``Sd`S
d`
n#Sd`S
dx`dxCEƀE{.
----..........00\0^00000111111111118292L2M2N2O222222222<3=3B3C3E3F3P3R3S3U33333ҾҾjphREHUj@
hRCJUV jmhR
hRPJ hRH* jshRhROJQJ hRH*jhREHUjٜ@
hRCJUVjhRUhR hR6<3333344444444$4&4(4*4,4.40424446484<4@4B4D4F4H4J4L4X4Z4\4^4`4b4d4f4h4j4l4p4r4t4v4x4z4|4~444444466666677888888jhREHUj@
hRCJUVjhRU hR6hRH*OJQJhROJQJ hRH* hRH* jmhRhRF4667678::;o;.<r<=8=h==d$da$$Sd`Sa$$
Sd`Sa$ $da$Sd`S
d`
Sd`Sd`89999::::':(:;:<:=:>:F:G:`:a:c:d:::::::::::::::::;;;;);̼vjhRCJEHUmH sH jڝ@
hRUVmH sH j) hRCJEHUmH sH jڝ@
hRUVmH sH hRmH sH jhRCJEHUmH sH !jc@
hRCJUVmH sH jhRCJUmH sH hRCJH*mH sH hRCJmH sH ));*;=;>;?;@;E;F;Y;Z;[;\;u;v;<<<<<<<<<< <:<<<J<L<X<Z<ttg\tMtthRCJH*OJQJmH sH hRCJH*mH sH jmhRCJmH sH hRCJOJQJmH sH hRCJH*mH sH jIhRCJEHUmH sH j%@
hRUVmH sH jhRCJUmH sH hRCJmH sH "jYhR6CJEHUmH sH j@
hRUVmH sH hR6CJmH sH jhR6CJUmH sH Z<^<`<b<d<f<h<j<r<==(=*=,=.=8=:=`=b=d=f=r=t==
>JAJBLBfBhBnBpBrBŶvf^^^Y hRH*hROJQJjhRCJEHUmH sH j@
hRUVmH sH j9hRCJEHUmH sH j@
hRUVmH sH jhRCJUmH sH hRhRCJH*OJQJmH sH hRCJOJQJmH sH hRCJH*mH sH hRCJH*mH sH jmhRCJmH sH hRCJmH sH =
>JAJHKDPLPqRyR}
Sdx
Sddx$da$
dx`dx $dxa$CEƀE{.rBtBxBzB|B~BBBBBBBDDtEvEFFFFFF FKLLLLDPLPRRxU~UXXZZ)[*[G[bbXbZbjjjtt4t6tsvyyL]S^ӳhRB*CJmH phsH jhR0J!U hR5hRB*phhRmH sH hR6CJmH sH hRCJmH sH j"hR hR6 hRH*hROJQJhR hRH*;yRRRUVYZZ*[G[__qe
d
dhxdx$dxa$$
dxa$dxxddxCEƀE{._vbjjj"p qssbvsvxlY$
dxxa$
ddxdxxj
&FQdhxEƀE{.^`gdRo&d&
&F^`dxd
svyyzzmKdCEƀ,{&.dxdxxd`$dxa$$da$KL]RS^[II$!dx^!`a$J$dC$Eƀ.{&a$$da$dhxGdC$Eƀ-{&y&9ǎKt1
\\iӒ,-Wܓ
~Ôby̕vΖ##:;I[ؽبhRCJOJQJmH sH hR6CJhmH sH hRCJhmH sH hR6B*phhRB*phhRmH sH hR6CJmH sH hRCJmH sH hRB*CJmH phsH hR6B*CJmH phsH 6HQsӒi
Ô[e$ dx^ `a$dx^`dx^`$dx^`a$[ؘݙCSezmFX֜3FϞڞ'()HdmĠ/ǡ<>>@RSUhR5mH sH hRCJH*mH sH jhR0J!CJUmH sH hR5CJmH sH hR5
hR6CJ
hRCJhR6CJmH sH hRCJmH sH hR6hR9mțhϞڞ).=N`q$d$Ifa$$dxxa$dxxdhxdx^`$dx^`a$
$dx`a$qrsxeYMMMMMM$d$Ifa$$d$Ifa$kd$$IfTl4\~v#`
s04
laf4T$d$Ifa$kd$$IfTl4ִ~&v{W # nn0 4
laf4TƟ˟Пԟ$d$Ifa$$d$Ifa$ԟ՟kds$$IfTlִ~&v{W #nn0 4
laT՟ޟ$d$Ifa$$d$Ifa$$da$kdQ$$IfTlִ~&v{W #nn0 4
laTGHXcdmrYPPP d$Ifkd$$IfTlF|X+04
laT$d$Ifa$ d$If
Sdxx $d$If]a$ d$IfĠɠܠ{riiiiYYYY$d$If]a$ d$If d$Ifkd$$IfTlF|X+04
laT
/4GkbYY d$If d$Ifkdy $$IfTlF|X+04
laT$d$If]a$
GOWX^diqrx~$d$If]a$ d$Ifǡۡ{sj^^^^^$d$Ifa$dxx$da$kdA!$$IfTlF|X+04
laT"RFF:::$d$Ifa$$d$Ifa$kd "$$IfTl4r
d`[A04
laf4T"',16;<Ikd#$$IfTlr
d`[A04
laT$d$Ifa$<=ˢMkd$$$IfTl4FI
!
N
N
04
laf4T$d$Ifa$$d$Ifa$$
dxa$ˢӢݢI=$d$Ifa$kd$$$IfTlrI
p!
''''04
laT$d$Ifa$
-FGLQV\abkt}$d$Ifa$$d$If`a$ƣϣأ٣9kd%$$IfTlrI
p!
''''04
laT$d$Ifa$$Ld$If`La$٣"@TXY^ejpv|}$d$If`a$$d$If`a$$d$Ifa$$d$If`a$$d$Ifa$otal output
Share of hired labour
Degree of market integrationa
Soil quality indexb
Age
0.28
-0.001
0.16
-0.07
-0.11
0.001
1.99 **
-2.48 **
0.92
-0.63
-1.80 *
0.87
0.94
-0.001
-0.29
-0.40
-0.23
0.001
5.38 ***
-6.36 ***
-1.49
-2.79 ***
-2.14 **
0.44Significance of the model
Parameter g
0.23
3.69 ***
0.56
5.79 ***H0 {dk = 0 (k}rejected ***rejected ****, ***, ***: significance at 10%, 5%, 1%
a The degree of market integration is given by the ratio of total revenue over total output. The greater it is, the more commercially integrated the farm is.
b The soil quality index is based on a soil survey. Smaller values represent lower soil quality. The range in the whole sample is between 0.15 and 1.61.
Table 5: Descriptive results of DEA efficiency estimates
Farm specialisationMeanStandard deviationMinimumMaximum% farms with efficiency score of 1Livestock
Crop0.71
0.570.15
0.180.38
0.181
17
3
Table 6: Results of the Tobit analysis: Inefficiency determinants
Livestock farmsCrop farmsParametert-ratiocParametert-ratiocConstant
Total output
Ratio of capital/labour
Ratio of land/labour
Share of hired labour
Share of rented land
Degree of market integration
Soil quality index
Education dummyd
Age2.02
-0.002
0.05
-0.007
0.07
0.03
-0.54
-0.24
0.05
0.000914.67 ***
-8.58 ***
6.80 ***
-1.84 *
0.36
0.42
-4.71 ***
-3.54 ***
1.34
0.553.01
-0.001
0.03
-0.002
-0.18
-0.22
-0.85
-0.43
0.23
0.0001 9.09 ***
-3.33 ***
3.43 ***
-0.77
-0.59
-1.15
-4.52 ***
-2.33 **
2.10 **
0.03
Pseudo-R2
0.71
145.2 ***
0.12
65.4 ***c For the pseudo-R2, the significance is evaluated with a Chi2 test.
d The education dummy is rather a non-education variable as it takes the value 0 if the farmers have some agricultural education, and 1 if they have not.
Table 7: Impact of size on efficiency: ANOVA analysis, F-test statistics
SpecialisationDEA estimated
technical efficiencySFA estimated
technical efficiencyLivestock
Crop 2.39 **
4.52 ***15.70 ***
7.11 ****, ***, ***: significance at 10%, 5%, 1%
Graph 1: Technical efficiency for livestock according to size (SFA)
Graph 2: Technical efficiency for crop according to size (SFA)
Graph 3: Technical efficiency for livestock according to size (DEA)
Graph 4: Technical efficiency for crop according to size (DEA)
Notes
PAGE 1
PAGE 30
A more detailed comparison between SFA and DEA can be found in Coelli et al.(1998).
The difference in the shape of the curves between livestock and crop farms might reflect the different influence of the components of the total technical efficiency. Another DEA conducted under variable returns to scale shows that for both specialisations the relationship between pure technical efficiency and farm size is a U-shaped curve, while the relationship between scale efficiency and farm size is a $d$If`a$$d$If`a$$d$If`a$$d$Ifa$UI9--$d$Ifa$$=d$If`=a$$d$Ifa$kd&$$IfTlrI
p!
''''04
laTHJNPR\^|~
EFMN1cʿʿ謢ʿʿʝ hR5hRCJmH sH hRmH sH hRCJH*mH sH hR
hRCJH*
hRCJ j"hRCJmH sH hRCJH*mH sH hRCJmH sH hRCJOJQJmH sH <$&02D$d$Ifa$$d$If`a$DFd~UE55$d$If`a$$=d$If`=a$kd'$$IfTlrI
p!
''''04
laTxqhZJ>>$d$Ifa$
dxx
d<d<<dxkd($$IfTl4FI
!
N
N
04
laf4T"EF3kdU)$$IfTl4ֈ
hd_p#O 04
laf4T$d$Ifa$FPUZ_dinsuwy{$d$Ifa$$d$Ifa${|}B4+dxx
d<kds*$$IfTlֈ
hd_p#O 04
laTdkd+$$Ifl4Fr"K
04
laf4$d$Ifa$$d$Ifa$ 3H^sYMMMMMMM$d$Ifa$kdZ,$$IflrrFZ"K
04
la$d$If`a$$d$If`a$$!d$If`!a$$d$Ifa$$d$If`a$$d$Ifa$!'1;AGLSX_fmt{$!d$If`!a$$d$Ifa$$d$If`a$$d$If`a$$d$If`a$$d$If`a$$!d$If`!a$$d$If`a$$d$If`a$YMMA1$d$If`a$$d$Ifa$$d$Ifa$kdy-$$IflrrFZ"K
04
la$d$If`a$$d$Ifa$M1YP@7dxx
d<<dx<kd.$$IflrrFZ"K
04
la1@Ncqf]]P
@d$If`@ d$Ifkd/$$IfTlFMR04
laT d$If$d$Ifa$ TM?
d<dxkd0$$IfTlFMR04
laT
d$If`
d$If`
@d$If`@,lu1r3t3 $dxa$h]h&`#$$
dxa$
gmuv00112p3r3x3z3 hR5hRCJmH sH hRCJOJQJU
hR6CJ
hRCJjhR0J!CJUh;60JmHnHuhR0JmHnHu
hR0JjhR0JUhRjhRU%G-curve. Thus, the total technical efficiency for livestock farms depicted on Graphs 1 and 3 represents mostly pure technical efficiency, while the total technical efficiency for crop farms depicted on Graphs 2 and 4 represents mostly scale efficiency.
The classification into small, medium-sized and large reflects the farm distribution in Poland. If it is compared with other countries, most of the farms should be categorised as small.
t3v3x3z3$
dxa$2
00BP. A!M "#$%pDd
$ TB
SA?2j.KpāDU`!j.Kpā8 XJ|xcdd`` @c112BYL%bL0Yn&|B@?6 5rC0&dT20]I?1@001d++&1k[~!}y;.P.P56zҚY 'qQ |
#;}`
& #J+a\ ~6{?!0Gd>IG7`e4*Ra*A*45`*Lfi`#
W&00\cZ[@||~rp7Kl)AMW`pzedbR
,.Ie(_ΞVDg:Dd
TB
SA?2ΥDxE"S8U`!xΥDxE"S8F XJFxcdd``Vfd``baV d,FYzP1C&,7\N! KA?HZ
@+5rC0&dT20]1@001d++&1k[~!}9;C.:]j mU2qB, 0& OB\XsGLc%0,ŧL@(\4RH>}S=pb
c
g.p\
ev0o8+KRsʽӢAhݏ`cHDd
hB
SA?2~22/ratGU`!~22/ratG @|Txcdd``.bd``baV d,FYzP1C&,7\ A?d-πjx|K2B*R 01d++&1k[~!}lv<.B]j mhy̛Ε0L@2W
dD6s 3ٝsj3C1o('Y0p<#DP 27)?蟂@;WY@Fhlm%4pS
{4 -F&&\r=4Iv.F} F33X?xYH7Dd
TB
SA?2_!`uE6v}6U`!u_!`uE6v @2XJCxcdd``bd``baV d,FYzP1C&,7\ A?d-ǀjx|K2B*R15@0@penR~KVo >P.P56c4fiը$d^Wy >Nsa`ev>Eh) 2BXQk[͇\
.4
0y{I)$5"kCv.F Ff~ΗDd
$ hB
SA?228I"'TU`!28I"'Tf@|xcdd``6fb``baV d,FYzP1C&,7\>! KA?HZ |πjx|K2B*R15@0@penR~6ƥo >P.P56c\
ҪQIȼ\$sЄa.ܐbr<$w6!b.#FZ]\{{pцdgrیϽ+@FTUW9v/>[xC[jn33ZQ
Ɂdhh&An<`憘kC*/a=巢V|1tdnʆ^
.h>UwPLLJ%'BAh={Dd
,TB
SA?2ZKaBU"x˨6U`!.KaBU"x˨ XJxcdd``> @c112BYL%bL0Yn&B@?6 5bnĒʂT+~3);a&br?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry7 FЬ#2Data
U1WordDocument646ObjectPool92Ь#2_1084021192 F22Ole
CompObjfObjInfo !$%(-2567:=>?ABCDEFHIJKLN
FMicrosoft Equation 3.0DS EquationEquation.39q`
ln(yi
)=f(xi
,)+vi
"uiEquation Native _1084006567F22Ole
CompObj
f
FMicrosoft Equation 3.0DS EquationEquation.39qMش\$
TEi
=exp("ui
)
FMicrosoft Equation 3.0DS EquationEquation.39qObjInfo
Equation Native i_1084023174F22Ole
CompObjfObjInfoEquation Native g_1084086371F22K`X
i
=0
+zi
FMicrosoft Equation 3.0DS EquationEquation.39q]$X&
1"()Ole
CompObjfObjInfoEquation Native @_1084087002F22Ole
CompObjfObjInfo
FMicrosoft Equation 3.0DS EquationEquation.39q]r(
Li
=exp"i
*"zL
*()Equation Native _1084087033'F22Ole
CompObj f
FMicrosoft Equation 3.0DS EquationEquation.39q]r-dz
Ui
=exp"i
*"zU
*()
FMicrosoft Equation 3.0DS EqObjInfo!"Equation Native #_1084087055$F22Ole
&CompObj#%'fObjInfo&)Equation Native *6_1084087077",)F22uationEquation.39q]__
zL
FMicrosoft Equation 3.0DS EquationEquation.39q]4
zUOle
+CompObj(*,fObjInfo+.Equation Native /6_10840871671.F22Ole
0CompObj-/1fObjInfo03
FMicrosoft Equation 3.0DS EquationEquation.39q]^
zL
="1
1"/2()1""i
*/*
()[]{}Equation Native 4_10840872883F22Ole
8CompObj249f
FMicrosoft Equation 3.0DS EquationEquation.39q]LP
zU
="1
1"1"/2()1""i
*/*
()[]{}ObjInfo5;Equation Native <1TableSummaryInformation(8@?m`bQUu=؇`cn``ÄI)$5[Dv.F Ɔb#3X?GDd
@hB
SA?2Z34#/$VI6U`!.34#/$VI@ |xcdd``> @c112BYL%bL0Yn&&&! KA?HZ17T
obIFHeA*C0d@Hfnj_jBP~nbúߗ_dTT0]Z 1cA?m`bQUu=؇`cn``ÄI)$5:[
;C#|
caGmJDd
\
hB
SA ?28Er@.F}U`!8Er@.F>`@"!|VxkATCEC۵=Ҋk-Vi!lK! IhZ
zxRsGPy;nIe^ \Y(ԍ(!""Զmz)Y/ug+R2joeQg͗AX®L;0)IC:Zh?wz!z=î#<Ё&6/ɯ8_zު{<]#.nsM\z:\"r7o1طB:Oz}|%`2y}Y$߅`2)F!n[rt#]|U}3͙kf>sc<7<`̴qDrriZ|{8Qoԃ,eU[&K㻤-go#~Zn70%Mw_;Ԥh3C?Wp~F㯪ְ~z/TS~\yJ;LU[pofxޅ%gw#A)jFZU`!a2/>jFZ^``;$0\xkALmhlZ=l=1UV-Vh!lT6m~4D(zCɛ(4A:v6 üof滟!q
l$QBDD8]$V'Z iG8c;6? puz( _84}v),e_%t`)ߞW~\|}OY8~Nחm22[
&gOyx$p-0$Z8YFޛw]]WC[<u.. ߬g3Ek>{*eg[ mj+\zɵ
Oz`mCFs3u5ܺE୩xǸp%
pkzRc,fTSisu2PV<ȷs8p _4P.cGkkS_Nv֩z,rjJo:XFq.3U'lT5r$$If!vh55
5s5#v#v
#vs#v:Vl40+55
5s5//4f4T$$If!vh555555n5n5#v#v#v#vn#v:Vl40+555n5//////////////4f4T$$If!vh555555n5n5#v#v#v#vn#v:Vl05555n5/////////////////4T$$If!vh555555n5n5#v#v#v#vn#v:Vl05555n5////////////4T$$If!vh5+55#v+#v#v:Vl05+55//4T$$If!vh5+55#v+#v#v:Vl05+55//4T$$If!vh5+55#v+#v#v:Vl05+55//4T$$If!vh5+55#v+#v#v:Vl05+55//4T$$If!vh5A5555#vA#v#v#v#v:Vl405A5555//4f4T$$If!vh5A5555#vA#v#v#v#v:Vl05A5555//
//
4T$$If!vh5
5N
5N
#v
#vN
:Vl405
5N
/4f4T$$If!vh5
5'5'5'5'#v
#v':Vl05
5'//////4T$$If!vh5
5'5'5'5'#v
#v':Vl05
5'////4T$$If!vh5
5'5'5'5'#v
#v':Vl05
5'////4T$$If!vh5
5'5'5'5'#v
#v':Vl05
5'///////4T$$If!vh5
5N
5N
#v
#vN
:Vl405
5N
/4f4T$$If!vh5O55555 #vO#v#v#v#v#v :Vl405O55555 ///4f4T$$$If!vh5O55555 #vO#v#v#v#v#v :Vl05O55555 //
//
4T$$If!vh5K
55#vK
#v#v:Vl405K
55/4af4$$If!vh5K
5555#vK
#v#v:Vl05K
55///////4a+$$If!vh5K
5555#vK
#v#v:Vl05K
55////////4a$$If!vh5K
5555#vK
#v#v:Vl05K
55/////4a$$If!vh5R55#vR#v#v:Vl05R55////4T$$If!vh5R55#vR#v#v:Vl05R55//4TOh+'0
@LX
dpx(Technical efficiency of Polish farmsll180Normal.dotsofiad2Microsoft Office Word@F#@!p4@2@2F՜.+,0 hDocumentSummaryInformation8GPCompObjMqp
Imperial College at WyeMz'%Technical efficiency of Polish farmsTitle
FMicrosoft Office Word Document
MSWordDocWord.Document.89q
"8@8Normal_HmH sH tH p`p
Heading 1-$
&Fdx@&^`5CJKHmH sH u^@^
Heading 2$$x@&a$B*CJaJhmH phsH ub`b Heading 3$$
&F@&a$CJOJQJ^JaJmH sH uT@T Heading 4$$@&a$5CJ\aJmH sH uH`H Heading 5
&F<@&CJaJ\@\ Heading 6$$7dh@&`7a$CJaJmH sH uL@L Heading 7$$dhx@&a$CJaJ`@` Heading 8 $$dhx@&`a$CJaJmH sH u` @` Heading 9 $$Idhx@&`Ia$CJaJmH sH uDA@DDefault Paragraph FontVi@VTable Normal :V44
la(k@(No List6@6
Footnote Text@&@@Footnote ReferenceH*L>@LTitle$a$CJOJQJ^JaJmH sH uLJ@"LSubtitle$a$5CJ\aJmH sH uJoA2J
Styl1$
&Fxxa$5CJ\aJL"@LCaption$`a$CJaJmH sH uLP@RLBody Text 2$a$CJaJmH sH uPB@bP Body Text$`a$CJaJmH sH u^C@r^Body Text Indent$`a$CJaJmH sH udOdNormalny (Web)
dd#CJOJPJQJ^JaJmHsHu4 @4Footer
9r .)@.Page Number4@4Header
9r 0U@0 Hyperlink>*B*pR@pBody Text Indent 2$7dhxx`7a$CJaJmH sH upS@pBody Text Indent 3$7dhxx`7a$CJaJmH sH u>Q@>Body Text 3$da$4+@4Endnote Text >*@>Endnote ReferenceH*|%^ԈV6K&'0Y!ly
"g %S&&&&&&6'c''8(1*+++,`,-.#.}... /%/=/u//269>>@A
A)ADDFFFGGCLKLMLUUUZ[]]` `Hcecud}djjl1nJpYrtt
uuvvvIwwx3yyz:{{ ||}}p~5ZAʄuŇ|ֈۈ
%,3<FPTUcgkosx}2:BCHMQWX]bflmqvĊŊ܊%+089:tŋʋϋԋًދPQalmnxڌ !*4=>CJOUZ[djs|ύ#)*2;@FNSTY`flrxyƎҎӎ؎َF}ʐϐ %*/49>CEGIKLMˑԑՑޑ.C`sĒΒؒ#(/6=DKPWaku{Ǔȓ͓Γדؓ3AVWbhpy<ǖɖ̖ϖҖޖߖEݙ0000000000000
Q 00l0l0l
Q 00"0"0"0"0"0"
Q 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Q 00u/0u/0u/0u/j0u/0>0>
Q 00A0A0A0A0A0A0A0A0A0A0A0AQ 0A0A0A0A0A0A0A0A0A0A0A0A0A
Q 00j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j0j0j 0j 0j 0j 0j0j0j0j0j 0j0j0j0j0j 0j0j0j0j0j 0j 0j0j0j0j0j 0j0j0j0j0j 0j0j0j0j0j 0j 0j0j0j0j0j 0j0j0j0j0j 0j0j0j0j0j 0j 0j0j0j 0j 0j 0j 0j 0j 0j0j 0j0j 0j0j 0j0j 0j0j 0j 0j0j0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j0j0j0j0j0j 0j0j0j0j0j0j 0j0j0j0j0j0j 0j0j0j0j0j0j 0j0j0j0j0j0j 0j 0j0j0j0j0j0j0j 0j0j0j0j0j0j0j 0j0j0j0j0j0j0j 0j0j0j0j0j0j0j 0j0j0j0j0j0j0j 0j 0j0j 0j0j 0j0j 0j0j 0j0j 0j 0j 0j 0j 0j 0j0j0j0j0j 0j 0j 0j 0j 0j 0j 0j0j 0j0j 0j0j 0j0j 0j0j 0j0j 0j 0j0j0j 0j 0j 0j 0j 0j 0j 0j 0j 0j 0j0j0j0j0j0j0j0j0j0j 0j0j0j0j0j0j0j0j0j0j 0j0j0j0j0j0j0j0j0j0j 0j0j0j0j0j0j0j0j0j0j 0j0j0j0j0j0j0j0j0j0j 0j 0j0j 0j0j 0j0j 0j0j 0j0j 0j 0j0j0j0j 0j0j 0j0j 0j 0j0j 0j0j 0j0j 0j 0j0j0j0j0j0j0j@0@0@0@0Z0Z0Z0@0@0Z0
0@ 0@ 0@ 0
000TUlmĊŊ89lmKLԑՑדؓVWǖɖʖ̖͖ϖЖҖӖޖߖEݙX0X0X0DX0X0|X0X0X0X0 X0X0
8X0X0
`X0X0X0X0$X0X0X0X0,X0X0dX0X0@X0X0xX0X0X0X0xX0X0 !X0X0"#X0X0$%X0X0&'X0X0()X0X0*+(X0X0,-`X0X0./CX0X0000000000000000 0 0 00 '''*-38);Z<rB[z3RWXZ[\^cey(4=yR_svKqԟ՟G"<ˢ٣DF{1t3z3SUVY]_`abdfghijklmnopqrstuvwxyzx3T&&&n(((+++5-I-K--....!.7.K.M.S.g.i. ///%/9/;/:::::::::: #*!!./"$fu5D>jN߆"$ˁse}"$&PξZlX5˄m"$6_BOIׂ=m88s"$9I";29{Y"$:KQ1`*"$gBe!ccU"$
QDx8խY^1T2$N)= Ha"$LliU:SE@O"$!l*32[@.0( nR
B
S ?h .( <'( <.( ,( |-( |,( -( ?@ABCDEFGHIJKLMONPQRSTUWVXYZ\[]^_`abdcefgJJ%%66++kk ++||5UU3@@99KKkk"q"qtIuIuuuuugvgvxxyy {8{8{D{||~|~|||<~D~P~Z~Z~~99ĀЀ#,,]eq{{ssÇÇv{{==
!"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMONPQRSTUWVXYZ\[]^_`abdcefg9'*urn:schemas-microsoft-com:office:smarttagsStateBg*urn:schemas-microsoft-com:office:smarttagscountry-region8f*urn:schemas-microsoft-com:office:smarttagsCity=e*urn:schemas-microsoft-com:office:smarttags PlaceType=d*urn:schemas-microsoft-com:office:smarttags PlaceName9h*urn:schemas-microsoft-com:office:smarttagsplacehgfedhgdehfghfghghghghghghfgghghgggghghgghghghghghghghhfdehghghgh'ghfhhfh'h'dehedhhfgghf'dehedhfhgghf'hgZ[!
&
af-7:AT^Y^KR!!%%%%%%e&i&&&&&''''''*(6(X(\( ))))%)')q)s)w)|)))))))*!*T*V*++++++,,,,+---m-r----------.6......///u00004444555588-:3:v<|<==>>:V?VVX]X33333333333333333333333333333333333333333333333333333333ǖǖɖɖʖʖ̖͖ϖЖҖӖ7YgfAI^Z}F"h^`o(.^`.pLp^p`L.@@^@`.^`.L^`L.^`.^`.PLP^P`L.h^`.h^`.hpLp^p`L.h@@^@`.h^`.hL^`L.h^`.h^`.hPLP^P`L.^`.^`..^`...p ^p`)@^@`OJQJo(()x^`()H^`()^`()^`()7^Z}FgfA;6Rֈ
%,3<FPTUcgkosx}BWlmĊŊ89:tʋԋދPQalmnx=Z)SxҎ؎}ʐϐ%/9CGKLMˑԑՑĒWǓ͓דؓ3VWhyǖɖ̖ϖҖ@