Technical efficiency of meat sheep production systems in Spain

- 1 - Technical efficiency of meat sheep production systems in 1 Spain 2 3 J.P. Pérez a , J.M. Gil b, ∗ ∗∗ ∗ , I. Sierra a 4 5 a Dpto. Producción Animal y Ciencia de los Alimentos. Facultad de Veterinaria. Universidad de 6 Zaragoza, Miguel Servet, 177, 50013-Zaragoza, Spain 7 b CREDA-UPC-IRTA. Edifici ESAB – Campus del Baix Llobregat 8 Av. del Canal Olimpic, s/n, 08860-Castelldefels (Barcelona, Spain) 9 10 11 Abstract 12 The technical efficiency of sheep results in one of the most important sheep producing 13 regions in Spain has been assessed. The methodology is based on a survey from 14 representative farms (in terms of the existing alternative production systems) within the 15 region. Results indicate that the best farms, in terms of technical efficiency, are obtained 16 by extreme situations: either by extensive and well-managed farms, without pens and 17 one lambing per year (lower production but well adapted to the seasonality of prices and 18 more reduced costs), or by well managed farms with prolific ewes. Thus, maximum 19 efficiency is determined not so much by the production system as by the technical and 20 economic management to accommodate the specific circumstances of each farm. 21 Key words: Sheep production, Spain, Technical efficiency, Frontier production 22 functions 23 24 25 26 27 ∗ Corresponding author. Tel.: +34-935521210; fax: +34-935521121; E–mail address: chema.gi[email protected]u - 2 - Technical efficiency of meat sheep production systems in 1 Spain 2 3 4 1. Introduction 5 Sheep production in Aragón 1 is important not only from the social, economic and 6 biological points of view, but also for cultural and ecological aspects (Sierra, 2000). In 7 this region, 3.5 million lambs are fattened per year in more than 8,000 sheep farms. 8 Approximately 80% of the farms are in less-favoured areas and are more dependent on 9 European Union (EU) subsidies (Ashworth et al., 2000), 10 Early studies about the sheep sector in Aragón merely aimed to describe the 11 production model and its economic performance (gross margins) and to analyse the 12 impact of EU subsidies on these parameters (Sierra, 1977; and Manrique and Sáez, 13 1984; among others). However, in the context of the progressive liberalization of world 14 markets and future cuts in subsidies (especially after the incorporation of Eastern 15 European countries), an evaluation of the efficiency and profitability of sheep farms is 16 necessary in order to assess future viability. 17 Using more or less similar data sets than those considered in previous studies, the 18 aim of this paper is to apply econometric models to determine to what extent sheep 19 farms are efficient when allocating input quantities to obtain the final output. 20 Additionally, sheep farms are divided in two main groups according to their efficiency 21 (above and below mean sample values) and are characterized taking into account their 22 1 Aragón is an autonomous region in northern Spain, with a surface area of 47.616 km 2 and a semiarid climate. - 3 - lamb meat productivity as well as their cost structure, in order to identify main factors 1 affecting meat sheep farms efficiency 2 The methodology used is based on the estimation of a frontier production function 3 following Greene’s (1980) procedure. It consists of estimating a production function by 4 Ordinary Least Squares (OLS) and displacing the constant term until all the errors were 5 negative, except one, which was zero. Battese and Coelli (1988), Neff et al. (1993) and 6 Murúa and Albisu (1993), among others, provide good examples of this approach. 7 In these types of models it is considered that the main causes of inefficiencies are 8 represented by the error term but ignore the real possibility that the efficiency can be 9 influenced by factors that are out of control of the producers, such as the climate, 10 diseases, the unavailability of resources in a specific period, etc. As a result, in recent 11 years, researchers have adopted the concept of stochastic frontier production in which 12 the inefficiency is considered as only one part of the random disturbance. This approach 13 was developed by Aigner et al. (1977) and applied by Coelli (1988), Bravo-Bureta and 14 Rieger (1990); and Tzouvelekas et al. (2002), among others. 15 In this paper, a mixed approach is proposed. First, a frontier production function is 16 estimated and, in a second step, residuals from the estimated equation are analysed in 17 terms of some farm characteristics. Then, we isolate the error component that cannot be 18 controlled by the farmer. Although the paper refers to a specific region, it aims to 19 provide a useful tool, which can be applied elsewhere, to better understand the 20 performance of the lamb sector in a specific region or country. 21 The rest of the paper is structured as follows. In Section 2 the data and the 22 econometric tools are described. Main results are presented and discussed in Section 3. 23 Finally, some concluding remarks about the implications of the findings for the sheep 24 sector are outlined. 25 - 4 - 2. Material and methods 1 Data were obtained from a survey of 49 farms from Aragón, specialized in meat 2 production. An important difference among farms was the sheep breed used. The most 3 common was Rasa Aragonesa, a rustic and extensively used breed autochthonous to the 4 region and therefore competes for pasture in less favoured areas (Sierra, 2000). Other 5 observed breeds were more prolific (in terms of more lambs per birth) or belong to other 6 genotypes that greatly increase production in more intensive systems. The questionnaire 7 included several technical and economic data from the year 2000, since the fieldwork 8 was carried out in spring 2001. Before carrying out the survey we designed a pilot 9 questionnaire that was completed by eight farms to determine whether they could easily 10 provide the requested information. Finally, we tested whether the final questionnaire 11 was coherent and consistent. 12 The main economic data requested was related to sales and main expenses (feed, 13 labour, etc.). We also asked about interests paid on capital invested as well as the 14 depreciation of installations, equipment and livestock in order to calculate the Farm 15 Income (instead of the traditional gross margin). We also took into account the family 16 workforce. 17 As mentioned in the introduction, in this paper, a Cobb-Douglas Frontier 18 Production Function is estimated using Greene’s (1980) approach assuming that the 19 error terms are non-positive (farms can not produce above the frontier), independent and 20 identically distributed. Likewise, we assumed that explanatory variables are 21 independent of the error term. The following explanatory variables have been 22 considered: 23 - Feed Costs (FC), which involves the supplementary feed for the sheep, the pasture 24 cost as well as the feed for lambs. 25 - 5 - - Depreciation of capital (DC) invested, including fixed capital (installations) and the 1 existing credits at the time of the study. 2 - Labour (L)), which includes both an approximation of the cost of the family 3 workforce as well as salaries, including possible seasonal and temporary contracts. 4 All data used are measured in monetary terms, which are generally more precise 5 than information measured in physical units. 6 7 3. Results and discussion 8 Table 1 (first column) summarises the mean values and coefficients of variation of 9 the main technical and economic variables (euros/sheep/year), without considering EU 10 subsidies. As can be observed, the labour productivity (number of sheep per work unit) 11 as well as the other two technical variables considered (the number of lambs born per 12 ewe and the number of lambs sold per ewe) showed a lower variability than most of the 13 different cost items considered in Table 1. Moreover, as total sales variability is also 14 relatively low (22%) in relation to the different cost items, the performance of meat 15 sheep farms in Aragón (measured by Farm Income) exhibited a large variability 16 (100%). 17 (Insert Table 1) 18 Further analyses suggested that the main out-farm expenses were feed and labour 19 (taking into account both the family workforce and salaries). Again, these costs showed 20 the largest variability, specially the feed costs in pastures, which is a good indicator of 21 the existing diversity in the feed systems used in Aragón as well as the difficulty of 22 rationing them. 23 A novelty in this study is that it specifically considers interests from credits and 24 depreciation of installations and equipment (in euros per ewe per year). In these cases, 25 - 6 - the variability found is representative of the existing diversity in sheep farms 1 (installations are fully depreciated or are new, no as opposed to major debt, etc). The 2 average Farm Income was negative, indicating that the sector (at least in Aragón) is 3 very dependent on EU subsidies. 4 The new intensive systems have helped to increase production (Stefanou and 5 Saxena, 1983) but also have increased costs, making the problem even worse. Many 6 farmers have not considered the efficiency of several production factors, specially when 7 they try to improve results using the same local breed (Rasa Aragonesa). In fact, 8 although some authors (Valls, 1983) have shown that combining the ability of the 9 mentioned local breed to be fertile during the seasonal anoestrus period (provided that 10 an adequate reproductive management is implemented) and an improved litter size can 11 generate similar productivity (in terms of number of lamb/ewe/year) than more prolific 12 breed, the associated costs are very high making the system inefficient from an 13 economic perspective. 14 Technical efficiency was calculated from the Cobb-Douglas Average Production 15 Function mentioned in the previous section. The estimated model (t-ratios in 16 parentheses) is given by: 17 (2.51) (3.17) (10.83) (-0.24) DC 0.02 L 0.29 FC 0.76 0.26P +++= ˆ 18 0.86R 2 = 100.65F AV = 3.61PB = − 19 where all variables are expressed in logarithms, P ˆ is the estimated farm output, F AV is 20 the statistic to test the joint significance of all explanatory variables, and B-P is the 21 Breusch-Pagan test for heteroscedasticity 22 As can be observed, the production function was correctly specified. The value of 23 the B-P statistic was under the critical value at the 5 % level of significance (7.81), and 24 - 7 - the value of the Adjusted R 2 was relatively high (0.86). All the variables had the 1 expected signs and their coefficients were statistically significant at the 5% level. Given 2 that the variables are measured in logarithms, the coefficients were interpreted as 3 elasticities. For example, increasing the feed cost by 1% would increase output value by 4 0.76 %. 5 From the economic point of view, the results indicate that the sheep sector in 6 Aragón is more intensive in labour that in capital. Moreover, the sum of the coefficients 7 of the explanatory variables was unity, implying that the sheep sector is characterised by 8 constant returns of scale (if all inputs increased on the same proportion the output will 9 increase in the same percentage). In any case, this hypothesis was tested, and the value 10 of the corresponding statistic was 1.07, which was well under the critical value at the 11 5% level of significance (3.84), indicating that the constant returns to scale hypothesis 12 could not be rejected. After estimating the average production function we added to the 13 constant (-0.26) the maximum positive residual (in our case 0.43) in order to displace 14 the estimated function to obtain the frontier production function (P*). In other words, P* 15 represents the maximum level of output that could be attained for a given combination 16 of inputs. 17 From the frontier production function we have calculated the average level of 18 inefficiency in the sheep sector in Aragón using the measure proposed by Timmer 19 (1971). This method relates, for each farm, the real output produced (P i ) and the 20 potential output that could be obtained in the frontier (P* i ), using the actual combination 21 of inputs. Taking into account the functional form adopted for the frontier production 22 function, the Timmer’s measure of Technical Efficiency is given by: 23 Technical Efficiency i = P i / P* i i=1...49. 24 - 8 - Results indicate that the average technical efficiency of sheep farms in Aragón is 1 0.66. In other words, these farms could improve production by 34 %. As a final step in 2 our analysis we have tried to assess main characteristics of most efficient farms. To 3 achieve this objective, farms were divided into two main groups: above or below the 4 average efficiency (0.66). For each group we have calculated the average values of the 5 main economic variables in terms of euro/sheep/year (columns 2 to 5 in Table 1). The 6 more inefficient farms (38.7% of the sample) were those with higher production costs 7 and relatively little rationalization of labour in relation to the number of head produced 8 and/or the actual needs at specific moments during the production process. The most 9 inefficient farms showed the lower labour productivity (335 head/ work unit) and with 10 lesser reproductive intensification (1.08 lambs sold per ewe per year). These farms are 11 paying higher interests, partly due to the important investments made in installations 12 and/or the herd that are either too expensive or not adjusted to the farm size. In some 13 cases, important investments are due to the desire to increase the number of head in 14 order to adopt better handling practices. 15 Finally, the most inefficient farms show the lowest Farm Income values. In fact, 16 farms with a technical efficiency between 0.39 and 0.50 exhibit an average sales value 17 of 59.2 € per ewe per year, compared with a value of 74.83 € in the most efficient farms. 18 However, it seems that facilities are not used to capacity, as labour and feed costs are 19 relatively high. As a result, the Farm Income of inefficient farms is lowest (-50.33 €) 20 making them highly dependent on EU subsidies. 21 With respect to the eight most efficient farms, one third had costs and expenses that 22 were adjusted to their possibilities and capacity, especially one farm which practiced 23 one mating per year, did not use hormonal treatments, did not keep animals in pens and 24 the degree of capitalization was average. Accordingly, the traditional extensive system 25 - 9 - can provide excellent results under adequate management, correct planning and 1 coordinating all the existing resources throughout the productive year (reproduction, 2 feed, labour, installations, etc.) (Sierra, 1994). In this case the technical efficiency was 3 0.97, which is comparable with the farm with the maximum efficiency (1.00), a well 4 managed farm that uses prolific breeds. 5 On the other hand, prolific breeds did not seem to guarantee a high level of 6 efficiency. Moreover, farms with a prolific genetic base should be managed correctly. 7 Feed needs to be rationed, investments should be optimised, in terms of a long term 8 planning of depreciation and interest, and farm size has to be defined avoiding excess 9 capacity. None of these farms reach the optimum production level and economic results 10 are only average (efficiency index around 0.57-0.60). With adequate herd handling and 11 management skills, efficiency levels could be increased considerably. 12 13 4. Conclusions 14 Using statistical tools to analyse the efficiency of sheep farms, this paper has shown 15 that the optimal Farm Income can be obtained by rationalising main inputs in the 16 production process, adapting them to real needs and availability. The best results, in 17 terms of efficiency, have been obtained by extreme farms, either extensive, well18 managed farms, without pens and one lambing per year (less final production but low 19 costs and a correct orientation towards the seasonality lamb prices), or well managed 20 prolific farms. Thus, maximum efficiency is determined not so much by the production 21 system as by the technical and economic management to accommodate the specific 22 circumstances of each farm. 23 24