Doi: 10. 1016/j agwat
Download 181.32 Kb. Pdf ko'rish
|
1 2
Bog'liqfrija2009
Water use and technical efficiencies in horticultural greenhouses in Tunisia Aymen Frija a , * , Ali Chebil b , Stijn Speelman a , Jeroen Buysse a , Guido Van Huylenbroeck a a Department of Agricultural Economics, Gent University, Coupure Links 653, 9000 Gent, Belgium b Institut National de Recherches en Ge´nie Rural, Eaux et Forets (INRGREF), B.P. 10, 2080 Ariana, Tunisia 1. Introduction The North African countries located in the Southern Mediter- ranean region are among those that experience the most severe water shortages and face the greatest challenges in terms of future water availability. Most of these countries have a semi-arid climate with limited and variable rainfall. Moreover, much of the rainfall is lost through evaporation ( Hamdy and Lacirignola, 1999 ). During the last three decades, it has been common policy in these North African countries to develop irrigation infrastructure and to control renewable water resources to increase stability in terms of water supply ( Ben Mechlia, 2004 ). Agriculture, being an important component of food security policies, is the main consumer of this supplied water. On average, agriculture accounts for around 80% of total water consumption in Tunisia, Morocco, and Algeria. However, these supply policies have led to substantial use of irrigation water at heavily subsidized cost ( Thabet et al., 2005 ). In the early 1980s, there was a shift in agricultural water policies towards demand management. This alternative has been widely applied at a global level and involves rationalization of the current demand, instead of increasing the current supply. In other words, this means greater efficiency in the allocation and use of water in agriculture. Rainfall in Tunisia varies greatly, ranging from an average of less than 100 mm year 1 in the South, to more than 1000 mm year 1 in the extreme north of the country. In the north, however, the topography is mountainous, leaving relatively little cultivable land in the high rainfall areas. As a result, most agricultural activity is undertaken in areas with limited and highly variable rainfall, making irrigation necessary to stabilize or increase production by reducing climatic uncertainty. Currently, almost 385,000 ha (7% of useful agricultural land) are irrigated in Tunisia. The irrigation sector consumes 80% of the available water resources and provides 35% of the agricultural production value, 95% of horticultural crops, 30% of dairy production, almost 22% of agricultural sector exports, and 26% of total agricultural employment ( Ministry of Agriculture and Water Resources, 2004 ). Moreover, the demand for irrigation water continues to rise, due largely to the development of new irrigation projects and the intensification of irrigation within existing areas. Therefore, during recent decades concerns regard- ing the efficient use of water resources in the country have increased. These concerns have been addressed in three ways: (i) modernizing the management of collective irrigation systems by enhancing the role played by water users’ associations and by promoting user participation in all aspects of management, (ii) Agricultural Water Management 96 (2009) 1509–1516 A R T I C L E I N F O Article history: Received 24 January 2008 Accepted 28 May 2009 Available online 5 July 2009 Keywords: Irrigated production Data envelopment analysis Tobit model A B S T R A C T We measure the technical efficiency of unheated greenhouse farms in Tunisia, and propose a measure for irrigation water use efficiency (IWUE) using an alternative form of the data envelopment analysis (DEA) model. Technical efficiency measures the degree to which (all) farm inputs are used efficiently. IWUE is a measure of the efficiency of irrigation water use when other inputs and output are kept constant. As a second stage, a tobit model is used to identify the degree to which technical efficiency and IWUE correlate with a set of explanatory variables. A comparison of the efficiency scores obtained from constant returns to scale (CRS) and variable returns to scale (VRS) specifications shows that most farmers in our sample are producing at an efficient scale. Under the CRS assumption, the average technical efficiency of the sample was 67.3%. A similar pattern of scores was shown for IWUE; although in this case the average IWUE was even lower (42%). This implies that when all other inputs remain constant, the current output could be produced using, on average, 58% less irrigation water. We conclude that farmers’ technical training in greenhouse management, investments in water saving technologies and the existence of a fertigation technique on farm have a significant and positive effect on their level of IWUE. However, IWUE is significantly and negatively affected by the proportion of total farm land allocated to greenhouses. ß 2009 Published by Elsevier B.V. * Corresponding author. Tel.: +32 92646192; fax: +32 92646246. E-mail addresses: frijaaymen@yahoo.fr , Aymen.Frija@ugent.be (A. Frija). Contents lists available at ScienceDirect Agricultural Water Management j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / a g w a t 0378-3774/$ – see front matter ß 2009 Published by Elsevier B.V. doi: 10.1016/j.agwat.2009.05.006 reformulating the water pricing system by introducing a cost recovery objective and (iii) developing incentives to enhance and promote the adoption of water saving technologies at farm level ( Al Atiri, 2004 ). In this context, we analyse the current performance of irrigation water use efficiency (IWUE) and its determinants at the farm level in Tunisia. We focus on small-scale irrigated greenhouse produc- tion schemes in the Tunisian ‘‘Sahel’’ region. This choice is motivated by the socio-economic importance of greenhouse production in the region, the constraints on water resources and the increasing price of irrigation water. Our methodology is data envelopment analysis (DEA), which enables us to examine the efficiency of an individual input (i.e. water), while keeping other inputs constant. We use a tobit model to assess the effects of socio- economic and structural variables on the levels of technical and irrigation water use efficiencies obtained. 2. Methodology: efficiency assessment using data envelopment analysis The measurement of technical efficiency is based upon deviations of observed output or input vectors from the best production or efficient production frontier. If a production units’ actual production point lies on the frontier it is perfectly efficient. If it deviates from the frontier then it is technically inefficient, with the ratio of the actual to potential production defining the level of efficiency of the individual firm. Our measure of technical efficiency provides an indication of how the use of all inputs can be minimized in the production process of a given farm, while continuing to produce the same level of output. Additionally, we consider the possible reduction of a subset of inputs while keeping other inputs and the output constant. This generates a ‘‘sub-vector efficiency’’ measure. If this is applied to the possible reduction in water use the efficiency measure produced can be called ‘‘water use efficiency’’ or in the case of irrigated production, ‘‘irrigation water use efficiency’’ (IWUE). Parametric and non-parametric methods are the two main approaches used to measure technical efficiency. The results from both methods are highly correlated in most cases ( Wadud and White, 2000; Thiam et al., 2001; Alene and Zeller, 2005 ), indicating that both methods are valuable and the choice can be based on a researcher’s preference. A major advantage of non-parametric DEA for this study is that the calculation of sub-vector efficiency for irrigation water use is relatively straightforward ( Speelman et al., 2008 ). 2.1. DEA models Farrell (1957) introduces the relative efficiency concept, according to which, the efficiency of a decision making unit (DMU) can be evaluated by comparing it to the other DMUs in a given group. This concept was extended by Charnes et al. (1978) who developed the first DEA model, called CCR (Charnes, Cooper and Rhodes), to incorporate many inputs and outputs simulta- neously. In this way, DEA provides a straightforward approach for calculating the efficiency gap between the actions of each producer and best practices, inferred from observations of the inputs used and the outputs generated by efficient firms ( Wadud and White, 2000; Malano et al., 2004; Haji, 2006 ). Explicitly, DEA uses piecewise linear programming to calculate the efficient or best practice frontier of a sample of DMUs. The DMUs on this technical efficiency frontier will have an efficiency score equal to 1. Less efficient DMUs are measured in relation to the efficient ones. Moreover, different units of measurement for the various inputs and outputs can be combined within the DEA models. The first DEA CCR model assumed constant returns to scale (CRS). For a DMU producing an output Y, using an input X, it is feasible to produce aY using aX amount of input (a is a scalar). However, in practice this may not always be observed, as increasing the input does not usually result in a proportionate increase in output ( Speelman et al., 2008 ). For instance, when the amount of irrigation water is increased, there is not always a proportional increase in crop volume. For this reason, a variable returns to scale (VRS) option might be more suitable for technical efficiency measures in the case of irrigated farms ( Rodrı´guez Dı´az et al., 2004b ). The first DEA model used to assess technical efficiency under the VRS assumption was developed by Banker et al. (1984) and was called the BCC (Banker, Charnes and Cooper) model. To identify whether CRS or VRS applies to production using the DEA technique, we calculate and compare the efficiency values under both assumptions. The use of the VRS specification permits the calculation of technical efficiency (TE) without the scale efficiency (SE) effects ( Coelli, 1996 ). According to Coelli et al. (2002) , scale efficiency can be obtained by the ratio TE CRS /TE VRS . Obtaining similar values for CRS and VRS efficiencies for a given farm demonstrates high scale efficiency. For this reason we consider both assumptions in this study. The study of efficiency using DEA can be orientated toward inputs or outputs. The difference lies in whether the objective is to continue using the same amount of inputs while producing more output (output-orientated DEA), or to produce the same amount of output with fewer inputs (input-orientated DEA). We choose input orientation because, in the context of increasing water scarcity, it is more relevant to consider potential decreases in water use than increases in output ( Rodrı´guez Dı´az et al., 2004a ). Following the Banker et al. (1984) BCC–DEA model, and considering a dataset of K farms (k = 1, . . ., K), each of them using N inputs x nk (n = 1, . . ., N), for producing M outputs y mk (m = 1, . . ., M), each farm becomes the reference unit. We then solve the following linear program (1) K times (once for each farm). The model specification (1) is the dual form of an equivalent primal model specification that maximizes the outputs for given inputs. In applied analysis the dual specification is preferred because it involves fewer constraints. For a general exposition of primal and dual DEA models see, e.g. Coelli et al. (1998) : Min u ; l u (1) s:t: X K k¼1 l k y m;k y m;o (2) X K k¼1 l k x n;k u : x n;o (3) X K k¼1 l k ¼ 1 (4) l k 0 (5) where u represents technical efficiency and hence the percentage of radial reduction to which each of the inputs is subjected; l k is a vector of k elements representing the influence of each farm in determining the technical efficiency of the farm under study (farm 0 ), x n0 and y m0 are, respectively, the input and the output vectors of the farm 0 . The equation P K k¼1 l k ¼ 1 is a convexity constraint, which specifies the VRS framework. Without this convexity constraint, the DEA model will be a CCR model describing a CRS situation. The concept of ‘‘sub-vector efficiency’’ is introduced to account for a specific IWUE score for each farm ( Speelman et al., 2008; Lilienfeld and Asmild, 2007; Oude Lansink and Silva, 2004; Oude Lansink et al., 2002; Fa¨re et al., 1994 ). This IWUE score u t for a given A. Frija et al. / Agricultural Water Management 96 (2009) 1509–1516 1510 farm can be calculated using the following program (2): Min u t ; l u t (6) s:t: X K k¼1 l k y m;k y m;o (7) X K k¼1 l k x nt;k x n;o (8) X K k¼1 l k x t;k u t : x t;o (9) X K k¼1 l k ¼ 1 (10) l k 0 (11) u t can have a value between 0 and 1, where a value of 1 indicates that the farm 0 under study is a best performer located on the efficient frontier, and there is no potential to reduce irrigation water use without reducing the output level. A score less than 1 indicates the existence of irrigation water use inefficiency at the farm level, which means that water saving can be achieved. For example, a u t value of 0.6 shows that the observed farm should be able to produce the same level of outputs using only 60% of its current level of irrigation water when compared to its benchmark, which is constructed from the best performers with similar characteristics. This also means that 40% of the irrigation water used at the farm level could be saved. A graphical representation illustrates the difference between the technical and the sub-vector efficiency in DEA approach ( Fig. 1 ). We consider three farms using two inputs (water and another input) and producing a single output. Farms B 1 and B 2 are the best performers and linear combinations of their input use define the production frontier which envelopes all the observed data. The observation A is, however, inefficient. The radial contraction of the input vector x i [water, input 2 ] produces a projected point on the frontier (A 0 ). The obtained point A 0 is a linear combination of the observed data points B 1 and B 2 , with the constraints (7, 8 and 9) in program 1, ensuring that the projected point cannot lie outside the feasible set. The technical efficiency of farm A, relative to B 1 and B 2 , is given by the ratio u t = OA 0 /OA. The IWUE of farm A could be demonstrated using a scenario in which water quantity is reduced while keeping input 2 and the output constant. In this case A is projected to A 0 and the IWUE is given by the ratio u t = O 0 A 0 /O 0 A. Here, the benchmark is reached by multiplying the total volume of water initially used on farm A by its estimated IWUE score u t If, for example, A uses 10 units of water to produce its current level of output, then a calculated IWUE score of 0.6 would indicate that the benchmark of A uses only 10 0.6 = 6 units of water to produce the same output as A. This also implies that A can save 4 units of irrigation water. 2.2. Tobit model As a second stage in the study, we select a set of variables as potential determinants of technical efficiency and IWUE. We use tobit regression, which is an alternative to ordinary least squares regression (OLS) for situations in which the dependent variable is bounded from below or above (or both) either by being censored, or by corner solutions ( Wooldridge, 2002 ). In the former case (censored) observations outside the limiting interval are recorded as the border values. That is, if the range is given by the interval [a;b], observed y < a is recorded as y = a, and likewise observed y > b is recorded as y = b. Accordingly, the tobit model is defined as follows: u t ¼ X R r¼1 b r z r þ u r (12) where u t ¼ u t i f 0 u t 1; 0 i f u t 0; 1 i f u t 1 8 < : (13) where u t is the IWUE calculated from program 2 and used as a dependent variable, and Z is a (R 1) vector of independent variables related to attributes of the farmers/farms in the sample. 2.3. Case study and data collection We collected our data from small-scale greenhouse farmers in the region of Teboulba, located in the eastern central area of Tunisia and within the governorate of Monastir (cf. Fig. 2 ). This region belongs to Nebhana irrigation district where water scarcity is an important issue. The total agricultural land area in Teboulba is approximately 1914 ha, of which 600 ha are irrigated. The governorate of Monastir includes almost 39% (572 ha) of the total land area used for unheated greenhouses in Tunisia. Teboulba region provides almost one third of the total production of this governorate ( Regional administration of Agricultural Develop- ment, Monastir, 2004 ). The agricultural sector in Teboulba is dominated by rainfed olive plantations for olive oil production. Greenhouses, which can easily be integrated with the olive plantations, were largely developed after the establishment of the water transfer program that started in the early 1980s. Water is transferred from the northern and central parts of the country to the southern and coastal areas, where fresh water resources are scarce. In view of its coastal situation, bordered in the east by the Mediterranean Sea, fishing is an important economic activity in the region. Irrigation water prices in Teboulba are among the highest in Tunisia. The price is about 0.15 TND m 3 (1 s = 1.8 TND), while supplies in some other regions of Tunisia are priced at a minimum rate of 0.04 TND m 3 . Volumetric water pricing is applied in Teboulba. A water meter is installed at each farm and individual Fig. 1. Differences between (overall) technical efficiency and sub-vector input- oriented efficiency (based on Oude Lansink et al., 2002 ). A. Frija et al. / Agricultural Water Management 96 (2009) 1509–1516 1511 water consumption is measured and charged by water users’ associations that are responsible for water supply at the local level. Greenhouses are a means of overcoming climatic adversity, using the temperature of the soil as a free energy source. Greenhouses differ according to climatic conditions and the complexity of the technology used to ensure an optimal heat balance for the greenhouse thermal mass. In Teboulba, green- houses are made from a plastic covering installed on metal ropes. Farmers do not use heating systems inside the greenhouses. Their height is generally between 7 and 9.6 m, their width is between 2.8 and 4.7 m, and their length can be as much as 60 m. This provides a covered surface area of 420–540 m 2 . Each farm has a tank for water storage, which is normally supplied with water from a well or a public source. The tank is either directly connected to greenhouses using plastic water pipes or drip irrigation systems, or connected through a fertigation unit, which allows irrigation water to be mixed with liquid fertilizers. Within our survey, all farmers who own a fertigation unit connect it using drip irrigation. According to our findings two main types of ‘‘irrigation systems’’ can be distinguished in the region. These comprise either a drip irrigation system linked to a fertigation unit, or a system using only plastic pipes and without any specific technology for conserving water. We collected data in October 2005 from 47 farmers who own 16.2% (97.6 ha) of the total irrigated land area and hold 13.8% (276 greenhouses) of the greenhouses located in the region of Teboulba (2060). Farmers mainly produce tomatoes, melons and peppers. Our questionnaire contained two main sequences: (i) a farmer identification sequence (socio-economic and demographic char- acteristics) and (ii) a farm identification sequence (cultivated crops, quantities and costs of inputs; quantities and values of outputs, etc.). To limit the number of inputs/outputs used in the DEA, total output (production quantities) was converted into monetary values. The inputs considered are: land (hectares), irrigation water (m 3 ) and labor (number of working days on the farm during 1 year). Basic statistics for the sample farmers are shown in Table 1 . Several variables are expected to affect technical efficiency. Table 2 lists the explanatory variables (i.e. those relating to farmer characteristics, production technology, and farm struc- ture) used in the tobit model to explain the both technical efficiency and IWUE. 3. Results 3.1. Technical and irrigation water use efficiencies The General Algebraic Modeling System software (GAMS) is used to solve the programs (1) and (2) and provides efficiency scores for each individual farm. The frequency distribution of the obtained efficiency scores is reported in Table 3 . Fig. 2. A map showing the localization of the study area (Teboulba) Table 1 Basic statistics for the data used in the data envelopment analysis model. Total output (Tunisian Dinars) Inputs Land (ha) Water (m 3 ) Labor (days/year) Average 7,863.8 0.29 4,008.6 171 Standard deviation 5,577.1 0.18 2,920.8 141 Minimum 1,309.0 0.05 350.0 15 Maximum 24,923.0 0.85 10,000.0 660 A. Frija et al. / Agricultural Water Management 96 (2009) 1509–1516 1512 Table 3 shows that many farms are operating at an efficient level with regard to their overall production process. Under the VRS assumption, more than 89% of farms have a technical efficiency higher than 50%. The scale efficiency (ratio of technical efficiency scores calculated under CRS and VRS assumptions) of the studied sample is approximately 0.9 indicating that the majority of farmers in Teboulba operate at an efficient scale. On the other hand, the average IWUE score in the sample was only 41.8 and 52.6% for CRS and VRS, respectively. Again, the scale efficiency (ratio of IWUE scores calculated under CRS and VRS assumptions) is quite high, showing that IWUE is not affected by the scale of the operation. The results indicate that, on average, the observed outputs could have been obtained by using less irrigation water, while keeping other inputs constant. This evidence suggests that farmers could save substantial amounts of water by improving their IWUE. Table 4 illustrates the technical efficiency level and the IWUE for three groups of farmers identified as follows—(i) the most efficient overall (i.e. technical efficiency between 75 and 100%), (ii) the next most efficient (technical efficiency between 50 and 75%), and (iii) the least efficient (technical efficiency between 0 and 50%). The table shows that the average IWUE is always lower than the average technical efficiency for production as a whole. In fact, while the average of the technical efficiency is around 92% for the most efficient group, the average IWUE for this group is only about 80%. For the least efficient group, the average technical efficiency obtained is 45%; however, IWUE for this group is only 11.4%. Fig. 3 depicts the cumulative efficiency distributions, confirm- ing that under CRS and VRS specifications, the proportion of farmers with poor IWUE scores is always higher than the proportion of those having poor scores for technical efficiency. To determine the relationship between IWUE and technical efficiency, we use a pairwise correlation to investigate equality between both efficiency vectors. The test was statistically significant ( Table 5 ). Under both CRS and VRS assumptions, technical efficiency and IWUE are highly and positively correlated. 3.2. Determinants of the efficiency scores In this section we estimate a tobit model in which technical efficiency and IWUE scores are explained by a set of socio- economic explanatory variables. Given that scale efficiencies are Table 2 Definition of variables used in tobit regressions. Variable Definition Mean value of the sample (for continuous variables) Proportion of farmers (for zero/one variables) Farmer characteristics Age of the manager In years 43 Training program 1 for farmers who have following a training program in greenhouses techniques; 19.1 0 otherwise 80.9 Land ownership 1 for farmer owner of his land 80.9 0 otherwise 19.1 Experience Date of starting greenhouses activities (year) 24.8 Technological characteristics Soil analysis 1 if the farmer did a soil analysis during last 3 years 40.5 0 otherwise 59.5 Investment in water saving 1 if farmer did an investment in water saving technology during last 3 years 38.3 0 otherwise 61.7 Fertigation 1 for farms containing fertigation technology 85.1 0 otherwise 14.9 Farm characteristics Farm size Farm size in hectares 2.1 Proportion of greenhouses Proportion of total farm allocated to greenhouses (%) 10.8 Well 1 for farms containing a well 53.2 0 otherwise 46.8 Table 3 Frequency distribution of technical and irrigation water use efficiencies for the studied farms sample. Efficiency level (%) Technical efficiency IWUE CRS VRS CRS VRS Number of farms % of farms Number of farms % of farms Number of farms % of farms Number of farms % of farms 0 < E < = 25 0 0.0 0 0.0 24 51.1 14 29.7 25 < E < = 50 13 27.6 5 10.6 10 21.2 12 25.5 50 < E < = 75 19 40.4 19 40.4 4 8.5 4 8.5 75 < E < = 100 15 31.9 23 48.9 9 19.1 17 36.2 Average 67.3 75.6 41.8 52.6 Scale efficiency 0.88 0.95 E denotes efficiency level, which pertains to either technical efficiency or irrigation water use efficiency in this table. CRS denotes constant returns to scale, while VRS denotes variable returns to scale. A. Frija et al. / Agricultural Water Management 96 (2009) 1509–1516 1513 high for both types of efficiency, we regress only the CRS specification scores. Download 181.32 Kb. Do'stlaringiz bilan baham: |
1 2
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling