Technical Efficiency In The Mediterranean Countries

Preparing to load PDF file. please wait...

0 of 0
Technical Efficiency In The Mediterranean Countries

Transcript Of Technical Efficiency In The Mediterranean Countries

Abstract – This paper aims to estimate the technical efficiency of the agricultural sector in a group of Mediterranean countries involved in the process of global market liberalization. The analysis applies the stochastic production frontier model to a set of six different crop products. For each product, a panel data set covering the period 1990-2005 in fourteen Mediterranean countries is used and separate econometric results are obtained. Then cross-sectional country information is used, in the second stage, to assess the main factors contributing to technical inefficiencies and to explain efficiency differentials among Mediterranean countries. The empirical results revealed the presence of important inefficiencies in Mediterranean agricultural production with significant diversities across countries and crops. EU countries on average appear to have higher technical efficiency levels than SMC, among which some states such as Algeria, Morocco, Tunisia and Jordan display substantial misallocation of resources. The results indicate that there exists scope for increasing crop production up to 50% in some regions through expanding irrigated areas, encouraging the mechanisation of the farmers and combating the land fragmentation.
JEL Classification: C23, F12, F13, L15, Q17.
* Maître assistante à la Faculté des Sciences Economiques et de Gestion, Nabeul, Tunisie. Chercheur au LEAD, Université du Sud Toulon-Var, France, et au LEGI, Ecole Polytechnique de Tunis, Tunisie.
Région et Développement n° 25-2007

28 Nadia Belhaj Hassine
The Barcelona conference, held during November 1995, led to the negotiation and signature of partnership agreements between the European Union (EU) and several Southern Mediterranean Countries (SMC). The ambitious aim of the Barcelona declaration was to achieve a greater economic integration in the Mediterranean region through the gradual establishment of a free trade Euro-Mediterranean area by 2010.
The association arrangements, currently limited to the removal of tariff and non-tariff barriers on manufactured goods, are going to be enlarged to agricultural products. The debates over the progressive liberalization of farming products are still underway (Legrand, 2002; CIHEAM, 2002, 2005; Corrons et al., 2004; Bouët et al., 2004)
Agriculture, considered as a vital sector in the SMC, is expected to face ambitious challenges and interesting perspectives by the increasing openness of these economies
SMC enjoy a good potential in agricultural trade due to favorable climatic conditions, competitive advantage of cost of production, especially labor, and closeness to the EU markets. South Mediterranean governments implemented important agricultural development projects directed towards modernizing the agricultural sector and enhancing the efficiency and quality in the vegetal production. The government's strategies placed an increased emphasis on promoting relatively high value export goods like citrus, some fresh fruits and vegetables, wines and olive oil… at the expense of the more traditional farming productions (CIHEAM, 2002, 2005). These policies aggravated the heterogeneity that characterizes the SMC agricultural sectors, where a modern export oriented agriculture which mobilizes an important fraction of fertile lands and irrigation water, is combined with poor traditional farming mainly based on rain fed production systems and particularly vulnerable to irregular weather conditions. Rain fed agriculture represents the essential economic activity in the rural areas and suffers from the use of traditional practices, the lack of logistic and human skills, the weak productivity and the low quality of its products. This sector appears to be particularly sensitive to agricultural trade liberalization since the rural farmers may have severe difficulties to sustain competition from the more efficient EU producers (Corrons, 2000; Corrons et al., 2004)
Several empirical studies using applied general equilibrium models showed that a free trade policy can substantially boost the Mediterranean agricultural exports, with a wide expansion of the products having appreciable comparative advantages. They concluded that the opening process should be carried out with accompanying policies for restructuring the rural sector to cope with the fierce international competition (Chemengui and Dessus, 1999; Corrons, 2000; Jabarin, 2001; Muaz, 2004).

Région et Développement 29
In the coming years, therefore, one important policy issue for the SMC will be to make the whole agricultural sector more competitive, to increase efficiency in the farming sector and to enhance the rational utilization of scarce natural resources, mainly water and land.
SMC share some common features like environmental conditions, agricultural practices and cropping patterns and face similar policy and institutional challenges. They differ nevertheless in their resource endowments and their ability to meet food import requirements. These countries are expected to be affected in different ways by the free trade policy; their capacity to benefit from opportunities arising from the new trade environment being closely related to the performance of their agricultural sector. In this context, assessing the SMC's agricultural performance and their potential to compete with Mediterranean EU countries may be a useful tool for policy analysis and decision making.
The following analysis depicts farming performances through the productive efficiencies in the agricultural sectors of a panel of advanced and developing Mediterranean countries involved in the process of global market liberalization, and investigates the factors contributing to productivity improvement in these countries using the stochastic frontier models.
Since the seminal paper by Farrell (1957), extensive empirical research has been conducted on efficiency measurement in agricultural economics. Farrell's approach is based on the comparison between observed production and best-practice or frontier production. At any point of time a production frontier reveals the state of a certain technology that determines the maximum feasible output from the actual bundle of inputs. A farm producing beneath this frontier is considered as technically inefficient in its resource utilization. Technical efficiency is, therefore, calculated through the ratio of observed production to the corresponding maximum output given by the production frontier. This efficiency notion accounts for all producible outputs and all types of inputs but it only provides a scalar measure of technical efficiency which gives little information about the improper use of specific inputs. Several econometric studies have attempted to examine the sources of technical inefficiency by the regression of technical efficiency indexes on a set of explanatory variables that affect the managerial ability of farming such as age and education, inputs quality and factor endowment (Pitt and Lee, 1981; Bravo-Ureta and Pinheiro, 1993; Ferrantino and Ferrier, 1995; Hallam and Machado, 1996; Alvarez and Gonzalez, 1999; Iraizoz et al., 2003). These studies use the Stochastic Production Frontier models. The causes of technical inefficiency are investigated by a two-step procedure which first estimates the relative efficiencies using the stochastic frontier, and then analyses the effects of the exogenous farm-specific factors on efficiency.
Stochastic frontier models have been extensively applied in the past mainly at a micro level, but have gradually gained popularity in macro economic analysis in recent years. The number of studies investigating crosscountry differences in agricultural productivity levels and growth rates has

30 Nadia Belhaj Hassine
significantly expanded during the past decades, (Kawagoe and Hayami, 1985; Lau and Yotopoulos, 1989; Fulginiti and Perrin, 1998; Rao et al., 2004). This is most likely driven by the availability of some new panel data sets, such as that produced by the Food and Agriculture Organization of the United Nations (FAO) and the development of new empirical techniques to analyze this type of data, such as the data envelopment analysis (DEA) and stochastic frontier analysis (SFA) techniques (Coelli et al., 1998). The majority of these studies focus generally on the estimation of the production elasticities and the investigation of the factors contributing to productivity differentials considering one aggregate agricultural output as a dependent variable.
Aggregating multiple products into a single output may however prevent the exploration of the efficiency differences that may exist among various commodity groups. As far as we know, no studies have been devoted to depict agricultural efficiency for the Mediterranean region using disaggregated outputs.
The present study applies the stochastic production frontier model to a set of six different Mediterranean crop groups (fruits, shell-fruits, citrus fruits, vegetables, cereals and pulses). The analysis employs a two-step approach which combines cross-sectional and panel data for the estimation of technical efficiency indexes in the agricultural sectors of nine South Mediterranean Countries: Algeria, Tunisia, Morocco, Lebanon, Turkey, Jordan, Syria, Egypt and Israel; and five Mediterranean European Union Countries: France, Spain, Italy, Greece and Portugal. In the first stage, separate production elasticities and efficiency indexes are obtained for each product group and each country using a panel data set covering the period 1990-2005. Then cross-sectional country information is used, in the second stage, to assess the main factors contributing to technical inefficiencies and to explain productivity differentials among Mediterranean countries. The results obtained from the second step regression help to compute corrected technical efficiency indexes.
The paper is organized as follows: section 2 outlines the stochastic production frontier model and the specification used to estimate technical efficiency indexes followed by the procedure used to explain the inefficiency effects. Section 3 provides an overview of the data used and reports the main econometric results. Section 4 summarizes the essential findings and conclusions.
This section focuses on estimating technical efficiency indexes in the agricultural sector of some Mediterranean countries, and analyzing the main factors contributing to inefficiencies. The analysis uses a two-stage procedure. A conventional index of technical efficiency is estimated in the first stage using Stochastic Production Frontier models. The estimated indexes are then adjusted using cross sectional country information to assess the main factors contributing to technical inefficiencies.

Région et Développement 31

2.1. Technical efficiency measurement

The Stochastic Production Frontier models (SPF) developed by Aigner, Lovell and Schmidt (1977) and Meeusen and Van den Broeck (1977) has propelled a vast amount of applied works on the econometric estimation of technical efficiency (Coelli and Battese, 1996; Pascual, 2001; Iraizoz et al., 2003; Joshua, 2005). This approach assumes that the gap between current and best practice may not be completely under the producer's control, and allows for the introduction of statistical noises resulting from random events such as weather conditions and equipment failures (Green, 1993).

The SPF is based on a parametric specification of the technology with

inefficiency effects. The disturbance term in a regression of output-input

relationship is considered as composed of two elements: a symmetrical error term ( ε ) that accounts for random effects and assumed to be independently and








2 ε








disturbance ( v ) which represents systematic effects that are not explained by

the production function and therefore considered as technical inefficiency. By

decomposing the error term, the frontier production function can be expressed


(1) yi = α + x iβ + εi − vi

where yi is the output of country i, xi the vector of inputs and α and β are
parameters to be estimated.

The stochastic frontier of equation (1) was extended to accommodate panel data by Pitt and Lee (1981) and Schmidt and Sickles (1984). The panel data model can be written as:

(2) yit = α + x itβ + λ t + εit − vi

where λt is the time effect. The inefficiency term can be fused with the constant, by setting α i = α − vi , to obtain a standard panel data model:

(3) yit = αi + x itβ + λ t + εit

Where αi can be estimated by the fixed or random effect estimator, according to the correlation between the individual effects and the explanatory variables (Alvarez and Gonzalez, 1999; Wang and Schmidt, 2002; Green, 2003).

The relative indexes of technical efficiency are then computed from the comparison of the estimated αi for each country to its maximum estimated value. For a logarithmic specification, these indexes are measured as follows:

(4) TEi = exp(αi − max α j )

32 Nadia Belhaj Hassine

where TE i is the technical efficiency level of country i. TE i takes the value 1 for the country with the largest individual effect, which is considered as producing on the production frontier and is said to be technically efficient, the remaining countries operating at some level of inefficiency obtain indexes lower than 1.
To make the economic model suitable for econometric analysis it is convenient to approximate the frontier production in (3) by a flexible mathematical function such as the translog form. The model to be estimated takes then the following form:

∑ ∑ ∑ (5) Ln(yit) = αi + β jLn(x jit) + 12

β jk Ln(x jit )Ln(x kit ) + υit



where β jk = βkj for j ≠ k , the subscripts i and t refer to the ith country and tth
period respectively, j and k represent inputs.
2.2. Determinants of technical efficiency and adjusting procedure
Measuring technical efficiency by individual effects can be misleading since this approach fails to disentangle heterogeneity unrelated to efficiency from the inefficiency itself. This ambiguity is likely to be particularly problematic in analysis based on aggregate country-level data, as the broad variation in the countries economic characteristics leads to a substantial unmeasured heterogeneity in the data (Greene, 2003).
Alvarez and Gonzalez (1999) performed a two-stage procedure for adjusting technical efficiency indexes from the heterogeneity captured by the fixed effects in a model applied to Spanish dairy farms. Their approach uses cross-sectional information on a farm's characteristics to estimate corrected technical efficiency indexes using panel data.
Following these authors, we assume that the heterogeneity captured by the fixed effect can be adjusted by complementary information about the specificities of countries. A large individual effect implies that there are unobservable factors that make one country more productive than another. Our intent is to disentangle the part of individual effect due to management from the part attributed to complementary factors. The methodology consists in regressing the estimated fixed effects on a set of explanatory variables reflecting countries characteristics:

(6) αˆ i = δ0 + ∑δ jz ji + ωi

where z ji are countries specific variables and ωi a random variable
The fitted value α~i is corrected by the largest positive residual to yield: (7) α*i = α~ i + maxωˆ j

Région et Développement 33
The adjusted index of technical efficiency uses α*i to yield:
(8) TE*i = exp(αˆ i − maxα*j )
3.1. Data
The empirical application in this study considers panel data at the national level for agricultural productions in nine SMC involved in the partnership agreements with the EU such as: Algeria, Egypt, Israel, Jordan, Lebanon, Morocco, Syria, Tunisia and Turkey and five EU Mediterranean countries presenting a strong potential in agricultural production as: France, Greece, Italy, Portugal and Spain for the period 1990-2005. The data we use come from the FAO (FAOSTAT), World Bank, AOAD and Eurostat databases as well as from the different reports of the FEMISE and the ESCWA. Our data set includes observations on the main crops grown in these countries, the inputs that are used and countries’ characteristics. The model comprises some non observable explanatory variables; we approximate these variables by available proxies. The variables used in the empirical analysis are summarized as follows:
– Output and input: we considered six aggregate product categories: fruits
(apricots, dates, figs, peaches and nectarines, pears, apples, plums, grapes), shell-fruits (almonds, peanuts, hazelnuts, pistachios), citrus fruits (lemons, oranges, tangerines, other citrus fruits), vegetables (artichokes, carrots, cucumbers and pickles, strawberries, watermelons and melons, pepper, potatoes, tomatoes), cereals (rice, wheat, maize, barley) and pulses (beans, peas, chick-peas, lentils, vetches). Inputs are classified into five groups: cropland, irrigation water, fertilizers, labor and capital. The data for the input use by crop for each country are constructed according to the information collected from recently published reports by FAO, FEMISE, ESCWA and the Ministries of Agricultural in the considered countries. Labor is calculated in terms of days worked for each of the selected crops, cropland is calculated in hectares of utilized agricultural area, capital is measured in terms of hours of machines used, fertilizers are evaluated by kilograms used and water is calculated by cubic meters allocated for each crop. Table A1 in the appendix presents summary statistics on the sample.
– Country characteristics: we use variables on the agricultural productive
capacity of each country such as: agricultural land area, part of irrigated area, water resources and agricultural machinery; environmental variables as: average precipitations and part of agricultural area incurring severe and very severe degradation; and land fragmentation evaluated by the part of exploitations having an area under five ha. Country statistics are summarized in table A2 in the appendix.
3.2. Empirical results
Equation (5) was computed by OLS using the WITHIN estimator for a fixed-effects model. We applied White's estimator of the variance-covariance

34 Nadia Belhaj Hassine

matrix to correct for heteroskedasticity. We achieved different groups of estimations: we started with appraising input elasticities from the translog production function using data for each group of products on the panel of the considered countries, and then we stacked the different groups in one model. Table 1 summarizes the inputs elasticities evaluated at the geometric mean of the sample. The elasticities are globally positive and significant at the 5% level. Water, cropland and labour have globally the largest values, indicating that the increase in Mediterranean agricultural productions depends mainly on these inputs. Fertilizers have a limited effect in the production of selected crops. Except for cereals and pulses groups, the fertilizers elasticity is weak and rarely significant, this may be explained by the fact that the selected crops are not very intensive in fertilizers; moreover, several farmers in SMC use fertilizers as a complementary factor to organic manure which is much less expensive. Water and labour appear as the most important factors in citrus fruits and vegetables productions, these crops being highly water and labour intensive. Fruits and shell-fruits are on the other hand capital intensive. Cropland seems to be the main factor in the production of cereals and pulses since these commodities require large agricultural plots.

Table 1: Estimates of the input elasticities






Cropland Water Fertilizer Labour Capital R²

0.21** 0.37** 0.037* 0.28** 0.19** 0.96

0.19** 0.12** 0.036 0.24** 0.29** 0.96

0.29** 0.43** 0.05 0.35* 0.12* 0.95

0.28** 0.16* 0.09 0.23** 0.26* 0.89

0.26** 0.75** 0.06** 0.38** 0.01 0.98

* and ** indicate significance at the 5% and at the 1% levels respectively.

0.49** 0.08* 0.23* 0.03* 0.09** 0.97

0.48** 0.07* 0.34** 0.08* 0.1 0.93

Relative indexes of technical efficiency are computed from the comparison of the estimated individual effects as in equation (4). The results are reported in table 2, yielding mean technical efficiency equal to 0.6 with a standard deviation of 0.23. Significant differences among crops and countries can be noticed. On average, over the period under consideration, EU countries exhibited higher technical efficiency indexes than SMC. It emerges from the results that Turkey, followed by France, Italy and Spain appear as the most efficient countries, their average efficiency index varying between 74% and 85%. Greece, Lebanon, Israel, Egypt and Syria seem to be operating at a middle efficiency level ranging between 50% and 70%. Algeria, Morocco, Tunisia, Jordan and Portugal suffer from enduring substantial inefficiencies in their farming operations, since their relative efficiency index is inferior to 50%.

The results indicate that SMC are more efficient in fruits, citrus and vegetables productions than in shell fruits, cereals and pulses. Among these countries, Turkey followed by Lebanon and Morocco appear as the most efficient countries. EU countries on the other hand seem to be highly efficient in vegetables and cereals cropping. France and Spain show the greatest efficiency indexes in these productions.

Région et Développement 35

Table 2: Technical efficiency indexes

Algeria Spain France Greece Italy Portugal Israel Jordan Lebanon Morocco Syria Tunisia Turkey Egypt

0.34 0.61 0.85 0.74 0.78 0.28 0.75 0.45 1 0.34 0.65 0.38 0.87 0.49

0.61 0.71 0.96 0.72 0.69 0.34 0.72 0.45 1 0.37 0.5 0.35 0.89 0.77

0.17 1 0.13 0.98 0.95 0.29 0.57 0.43 0.65 0.88 0.53 0.33 0.81 0.43

0.68 0.69 0.33 0.39 0.68 0.15 0.79 0.28 0.58 0.28 0.75 0.21 1 0.32

0.54 1 0.97 0.76 0.88 0.55 0.43 0.6 0.39 0.74 0.51 0.57 0.99 0.66

0.63 0.91 1 0.82 0.73 0.54 0.57 0.33 0.59 0.32 0.39 0.68 0.88 0.61

0.28 0.23 1 0.45 0.48 0.25 0.43 0.26 0.61 0.32 0.39 0.38 0.53 0.48

The determinants of efficiency differentials among the selected countries are investigated through the regression of the efficiency indexes on a set of explanatory variables representing some countries characteristics. The variables used are soil quality measured by the part of agricultural land incurring severe and very severe degradation, agricultural land size which reflects the country's production capacity, the part of irrigated area which measures the country's production capacity of irrigated crops, the climate evaluated by average rainfall, agricultural machinery to approximate the degree of agricultural sector mechanisation, land fragmentation measured by the part of exploitations having an area under five hectares and country's total water resources to reflect the availability of irrigation resources.

The estimation results are reported in table 3. The estimated coefficients globally have the expected signs and are significantly different from zero. Average precipitations, irrigated areas, agricultural machinery and land size have a positive impact on the efficiency of resources use while water availability and land fragmentation enhance inefficient behaviour. Soil quality does not seem to have a significant impact on technical efficiency. The positive correlation between agricultural machinery and efficiency reveals that mechanized farmers are more efficient and modernizing the agricultural sector can contribute to promoting the productivity of the sector. The positive impact of precipitations on efficiency can be explained by the fact that an important part of cereals and pulses, some fruits and shell fruits are produced in rain fed areas. These commodities are particularly sensitive to weather conditions and to the lack of rainfall characterizing the Mediterranean climate. An increase in rainfall can then contribute to a substantial rise in productivity and efficiency. Wider irrigated areas affect efficiency favourably, since irrigation is considered as a risk-reducing input that tends to increase mean yield and reduce its variability when rainfall is inadequate. Agricultural land size has a positive but limited impact on efficiency, this result may be explained by the fact that countries with higher agricultural areas by exploring scale economies tend to be more efficient than those suffering from narrow farming areas.

36 Nadia Belhaj Hassine

The availability of water resources seems on the other hand to encourage the waste of resources as it has a negative impact on efficiency. Land fragmentation is also negatively correlated with efficiency. Land fragmentation may lead to sub-optimal usage of factor inputs due to inadequate monitoring, the inability to use certain types of machines, and wasted space among borders1. A high percentage of land fragmentation may also reflect the existence of an important number of small farms with limited financial resources, low skills and inefficient traditional production methods.
Table 3: Estimation of the second stage regression

Soil quality Average precipitations Water resources Agricultural machinery Land fragmentation Part of irrigated area Agricultural land size R²

0.24 0.62** -0.49** 0.13** -0.14* 0.24** 0.04* 0.45

1.56 4.4 -7.98 3.19 -2.48 3.3 1.92

The second stage regression allowed the evaluation of efficiency indexes adjusted by the countries’ heterogeneity according to equation (8) of the model. The results are presented in table 4. The adjusting procedure leads to relatively similar results to those obtained in table 2. These results reveal that the observed heterogeneity between Mediterranean countries is mainly explained by the differential in resource management between these countries.

Table 4: Adjusted Technical Efficiency Indexes

Algeria Spain France Greece Italy Portugal Israel Jordan Lebanon Morocco Syria Tunisia Turkey Egypt

0.31 0.59 0.79 0.65 0.71 0.25 0.71 0.51 0.97 0.32 0.61 0.35 0.85 0.46

0.55 0.64 0.99 0.76 0.98 0.31 0.75 0.46 0.95 0.34 0.45 0.32 0.96 0.69

0.19 0.86 0.12 0.85 0.8 0.21 0.63 0.26 0.5 0.67 0.4 0.26 0.84 0.33

0.49 0.43 0.25 0.27 0.68 0.15 0.66 0.29 0.6 0.21 0.67 0.38 0.95 0.23

0.42 0.79 0.9 0.52 0.92 0.43 0.54 0.69 0.44 0.58 0.47 0.45 0.98 0.52

0.65 0.68 0.98 0.71 0.75 0.59 0.72 0.42 0.45 0.38 0.49 0.76 0.93 0.52

0.32 0.25 1 0.53 0.56 0.3 0.75 0.44 0.48 0.34 0.51 0.36 0.63 0.46


The analysis carried out in this paper aimed to provide estimates of technical efficiency in the agricultural sector in a group of Mediterranean countries involved in the process of global market liberalization. The study

1 A recent study conducted by (Raghbendra, Nagarajan and Prasanna, 2005) in southern India, showed that land fragmentation has a significant negative impact on production efficiency.
EfficiencyCountriesEfficiency IndexesMediterranean CountriesSector