Professional Documents
Culture Documents
Determinants of Economic Efficiency: A Case Study of Hazelnut (Corylus Avellana) Farms in Samsun Province, Turkey
Determinants of Economic Efficiency: A Case Study of Hazelnut (Corylus Avellana) Farms in Samsun Province, Turkey
To cite this article: Osman Kilic , Vedat Ceyhan & Isil Alkan (2009) Determinants of
economic efficiency: A case study of hazelnut (Corylus�avellana) farms in Samsun Province,
Turkey, New Zealand Journal of Crop and Horticultural Science, 37:3, 263-270, DOI:
10.1080/01140670909510272
Helfand 2003). Similarly, efficiency analysis has been by the K x TV input matrix (X) and M x TV output
applied relatively little in Turkey. There have been matrix (Y). Using piece wise technology, an input-
some studies on efficiency in Turkish agriculture oriented measure of technical efficiency (TE) can be
(Gunden et al. 1998; Zaim & Cakmak 1998; Akturk calculated for the i-th farm as the solution to linear
2000; Demirci 2001; Cinemre et al. 2006; Cinemre programming (LP):
& Ceyhan 2006; Ozcelik et al. 2006), but there has
Minimise,^ 8
thus far been no study carried out by applying data
envelopment analysis (DEA) to hazelnut farms. Subject to -_y, + YA>0 (1)
Therefore, the objectives of this paper were to calculate
farm level efficiency measures and to evaluate the 8x,. - Xk > 0
determinants of cost efficiency in hazelnut farms in X>0
the Black Sea Region, Turkey.
where 8 is the technical efficiency score having a
value 0<8<l. If the value equals 1, the farm is on the
frontier; the vector X is an TV x 1 vector of weights
MATERIALS AND METHODS that defines the linear combination of the peers of
the i-th farm.
DEA model for hazelnut farms The input-based minimum cost for the i-th
Data envelopment analysis (DEA) was used to farm can be obtained by solving the following LP
calculate efficiency measures. DEAis one of several problem:
techniques that can be used to calculate the best Minimise^,.,wTj x*
practice production frontier (Coelli et al. 1998;
Kumbhakar & Lovell 2000). Thanassoulis (2001) Subject to -_y, + YÀ>0 (2)
pointed out that further detailed analysis and possibly
inspection of the best and the worst performers are x* -Xk>0
then necessary to understand the production process
and derive useful information that may help both where w, is a vector of input prices for the i-th hazelnut
the worst and the best performers to make further farm; superscript T is the transpose function; x* is
improvements in efficiency. The DEA approach the cost-minimising vector of input quantities for the
provides an analytical tool for determining effective i-th hazelnut farm calculated by the LP, given the
and ineffective performances. input prices wj and output level y:, and A is a TV x 1
Efficiency is defined in this research in a relative constant vector. Equation 1 and Equation 2 represent
sense as the distance between observed input-output the cost minimisation under constant returns-to-
combinations and the best practice frontier. The scale (CRS) technology. CRS means that output
Farrell input-oriented measure of efficiencies was increases in proportion to changes in all inputs. The
used as a measure of efficiency since farms tend economic efficiency (EEt CRS) of the i-th hazelnut
to have a greater control over their inputs than farm is calculated as:
over their outputs. Farrell (1957) proposed that the
efficiency of a firm consists of two components: x,. (3)
(1) technical efficiency, which reflects the ability That is, EEiCRS is the ratio of the minimum cost
of a firm to obtain maximal output from a given to the observed cost, given input prices and CRS
set of inputs; and (2) allocative efficiency, which technology (Coelli et al. 1998).
reflects the ability of a firm to use the inputs in Coelli et al. (1998) pointed out that the CRS model
optimal proportions, given their respective prices is only appropriate when the farm is operating at an
and the production technology. These two measures optimal scale. Factors such as imperfect competition
are then combined to provide a measure of cost and financial constraints may prevent a firm from
efficiency (also called economic efficiency). The operating at optimal scale. Since hazelnut farms in
Farrell measure equals 1 for farms on the efficiency the research area conducted their activities under
frontier, and then decreases with inefficiency. imperfect competition, and because the size of many
We constructed a DEA model assuming that each hazelnut farms made them ineligible for institutional
hazelnut farm produces a quantity of hazelnut (y,) loans, we transformed Equation 1 to the variable
using multiple inputs (x,) and that each farm (i) returns-to-scale (VRS) technology model by adding
is allowed to set its own set of weights for both the convexity constraint NYK = 1, where ATI is an
inputs and output. The data for all farms are denoted TV x 1 vector of ones and X is an TV x 1 vector of
Kilic et al.—Determinants of economic efficiency 265
constant to Equation 1. In this example, the TE Data
scores under VRS was calculated using Equation 1, This study, after identification of survey objectives,
with the convexity constraint added to decompose used a well-designed state-of-the-art instrument
the technical efficiency scores into two components: to capture information that is of great interest and
"pure technical efficiency" (PTE), which reflects relevance to the questions under study. During the
the ability of a firm to obtain maximal outputs at an sampling process, following identification of the
optimal scale; and "scale efficiency" (SE), which study population, the sample frame was defined and
reflects the distance of an observed firm from the sample size was determined by the simple random
most productive scale size. Scale efficient farms sampling method (Yamane 1967). We used a 90%
are of appropriate size and thus do not need to be confidence level and 10% precision level when
reorganised to improve output or earnings. Scale determining optimum sample size. Random numbers
efficiency was calculated as the ratio of the technical generated from a random number table were used to
efficiency score of the farm under CRS technology select farms from the population. Using a structured
to the technical efficiency score of the farm VRS survey of 151 hazelnut farms selected by random
technology. Farms were classified as scale efficient sampling, we collected the input-output data used in
if the SE = 1 or if the TEVRS = TECRS. Farm-level scale this study during the production year of 2005-06.
inefficiency was determined by comparing technical
efficiency score under non-increasing returns to The cost efficiency of hazelnut farms was
scale (MRS) with the technical efficiency score modelled in a six-input, single output framework.
under CRS. If SE < 1 and TENIRS = TECRS, farms were Quantity of hazelnut produced in 2006 was used to
classified as scale inefficient owing to increasing measure output (kg/year). Six inputs were used for
returns to scale (1RS). If SE < 1 and TENIRS > TECRS, cost efficiency analysis: working capital (us$/year;
farms were classified scale inefficient owing to 1 us$ =1.6 Turkish liras), labour (h), harvest costs
decreasing return to scale (DRS). The allocative (us$/year), nitrogen (N) use (kg/year), phosphorus
efficiency was calculated residually by: (P) use (kg/year), and orchard size (ha). Other
data collected were demographic characteristics
AE^EE^ITE, (4) of farmers, farmer characteristics, and hazelnut
production characteristics.
In this study, the two-stage approach was preferred
The sample hazelnut farms produced 2.7 t/year,
in assessing the influence of various factors upon
on average. The minimum hazelnut production was
inefficiency because of several advantages (e.g., prior
0.5 t and the maximum was 12 t. To reach their
assumptions are not required regarding direction
present level of production, hazelnut farms used
of influence, accommodation of more than one
c. US$8500 of working capital, 3.21 h labour, and
variable with continuous or categorical variables).
350 kg fertiliser, mostly in N. In addition, sample
A Tobit regression of inefficiencies on potential
farms paid c. US$1800 for harvesting hazelnuts. In
determinants was used because the inefficiency
efficiency analysis, we assumed that all farms faced
scores are truncated at 0 and 1. The Tobit model is
the same relative price for labour, N and P, land, and
specified as follows:
capital. Costs of labour, N and P, land, and capital
were us$1.54/h, us$0.25/kg, us$0.35/kg, US$1369/
yt = (5) ha, and us$0.10 per us$, respectively.
0 if y* < 0
We included in Table 4 the descriptive statistics
of variables such as hazelnut production, working
y* = ßx, JV(O, a2) capital, hazelnut yield, labour use, N and P use
where yt is the measure of economic efficiency associated with returns to scale.
for farm i, y* is the unobservable variable, xt are The variables included in the tobit analysis can
explanatory variables that influence the economic be divided into three groups: personal characteristics
efficiencies of the farms, and ß and u are parameters of farmers (the age and education level of farm
of the model and the random error term, respectively operators, and family size); farm characteristics
(Greene 2000). (farm size, the number of plots, hazelnut orchards/
For statistical analysis, SPSS statistical package total farmland, credit use), and hazelnut production
was used. Efficiency scores were measured using characteristics (the age of hazelnut orchards, location
DEAP 2.1. The LIMDEP program was used to of orchards, varieties, sucker control, the share of
estimate the effects of explanatory variables on family labour, and marketplace). The variables of
economic efficiency scores. location of orchards, variety of hazelnut, and sucker
266 New Zealand Journal of Crop and Horticultural Science, 2009, Vol. 37
Table 1 Descriptive statistics of input and output measures and factors influencing efficiency for a sample of hazelnut
farms.
harvesting and labour costs by 44% to perform that the hazelnut farms could reduce their input use
as well as other similar, best practice farms in the by 16% without output reduction. Fifty-eight percent
sample. of the sample farms had a higher technical efficiency
Efficiency measures for sample hazelnut farms are coefficient than the mean technical efficiency.
presented in Table 3. In the research area, the primary The analysis of the measures of technical
source of economic inefficiency was allocative. efficiency showed that pure technical inefficiency
Almost 93% of the sample hazelnut farms were was the primary cause of technical inefficiency. The
allocatively inefficient. These farms used the wrong mean pure technical efficiency was 0.67. The mean
input mix given input prices, so that their costs scale efficiency was 0.79, with a standard deviation
were 34% higher than the cost-minimising level. of 0.169 (Table 3). Individual analysis of the farms
The estimated technical efficiency measures for the indicated that 10% of the sample hazelnut farms
sample of hazelnut farms varied from 0.52 to 1.0, had constant returns to scale (CRS), whereas the
with a sample average of 0.85. This result indicated percentages of hazelnut farms exhibiting increasing
Table 2 Mean value of some comparative characteristics for the average Turkish
hazelnut farm and sample hazelnut farm. (Average Turkish hazelnut farm was
based on the results of Turkstat (2006), Fiskobirlik (2006), and Anon. (1992).)
Table 4 Summary of returns to scale results. Different letters reflect the significance at the 5% level for difference
between returns to scale. (1RS, increasing return to scale; CRS, constant returns-to-scale; DRS, decreasing return to
scale.)
returns to scale (1RS) and decreasing returns to scale more working capital than inefficient ones (P <
(DRS) were 84% and 6%, respectively (Table 4). 0.05). Therefore, scale-efficient farms obtained more
The 15 scale-efficient hazelnut farms had larger hazelnut yield compared with inefficient farms (P <
hazelnut production compared with the hazelnut 0.05) (Table 4).
farms having 1RS. In addition, scale-efficient farms Table 5 represents results of the tobit model on
used less labour per year, less N and P per ha, and the relationship between economic efficiency and
Estimated
Variable coefficient SE P [1 Z 1 > z]
Personal characteristics
Age of farm operators (years) 0.43 0.17 0.0130
Education level of farm operators (in years) 0.56 0.16 0.0055
Family size (person) 0.14 0.12 0.2523
Farm characteristics
Farm size (ha) 0.26 0.16 0.9843
No. of plots (unit) -0.34 0.15 0.0130
Hazelnut orchards/total farmland 0.51 0.98 0.6065
Credit use (us$/year) 0.49 0.12 0.0100
Hazelnut production characteristics
Age of hazelnut orchards (years) 0.32 0.21 0.8809
Location of orchards 0.32 0.50 0.5198
Varieties 0.61 0.31 0.8448
Sucker control 0.78 0.45 0.0805
Share of family labour 0.82 0.63 0.1935
Marketing place 0.55 0.22 0.8017
Log likelihood 37.08 0.0000
Table 6 Differences between economically efficient and inefficient hazelnut farms. (Figures in parentheses are
SEs.)
its determinants. Most signs related to efficiency level of operators and credit use were other
determinants were as expected. All variables determinants of economic efficiency. Strategies for
evaluated under farm characteristics were positive, a better farm-level education and farm extension
with the exception of number of plots. The credit programmes focused on sucker control may help to
use coefficient was positive, which indicated that the increase efficiency. Farmer training and extension
farms that used more credit were more efficient than activities are relatively low-cost methods of achieving
the lower ones (P < 0.01). The coefficient of plot increases in productive efficiency. However, such
number indicated that hazelnut farms having smaller increases strongly depend on the effectiveness
plots were more efficient than others (P < 0.05). The of the presentations by research and extension
variables of share of family labour, farm size, and organisations. Turkey's Ministry of Agriculture
share of hazelnut orchards in farmland were also a and Rural Affairs (MARA) has tried to prepare
positive sign. However, they were not statistically and implement extension and training programmes
significant (P > 0.10). throughout the country, including the research area.
The variables of the education level and age Branches of some input sales companies have also
of the operator indicated that more educated and contributed by transferring information to hazelnut
older operators were more efficient (P < 0.05). The farmers. However, the efficiency of such efforts is
variable of family size had negative effects ; however, still unsatisfactory owing to bureaucracy, limited
it was statistically insignificant (P > 0.10). investment, insufficient numbers of skilled extension
All the coefficients of the hazelnut production persons, and information gaps on technical and
characteristics group were not statistically significant economic aspects of hazelnut farming. Training
(P > 0.10), with the exception of sucker control. focusing on sucker control and cultural practices
All insignificant variables positively influenced may help increase efficiency in this research area.
the economic efficiency. The estimated coefficient In addition, encouragement of young rural people
for sucker control indicated that increasing sucker to stay in agriculture might have beneficial effects
control led to more economically efficient farms (P on economic efficiency. Providing farmers with
<0.10). greater access to credit would require government
Based on the results of the comparative efficiency support through legal and regulatory frameworks.
analysis, the economically efficient hazelnut farms Although the Turkish government has given an
carried out their activities on a smaller number of interest rate subsidy to all farmers, Kilic et al.
plots with a relatively high level of working capital. (2005) reported that hazelnut farmers did not use
However, their P use was lower than on inefficient this credit because of high transaction costs. Thus, a
farms. In addition, farmers on economically efficient government-supported pilot programme that reduces
farms were better educated and more experienced. the transaction costs of providing credit to farmers
Efficient farms also had higher hazelnut production would have the potential to increase efficiency. Such
and hazelnut yield/ha (Table 6). a credit programme could help farmers expand their
facilities from small-scale to modern facilities and
benefit from economies of scale, which may also
increase economic efficiency.
CONCLUSION
DEA was used to calculate efficiency measures and to
determine the factors affecting economic efficiency
in Samsun hazelnut farms. The mean technical, ACKNOWLEDGMENTS
allocative, and economic efficiencies for hazelnut We thank anonymous referees for their constructive
farms were 0.85,0.66, and 0.56, respectively. Of 151 comments and suggestions. We acknowledge Ondokuz
farms included in the analysis, 47 were technically Mayis University, Turkey for financial assistance.
efficient, 5 were allocatively efficient, and 6 were
economically efficient. Analysis of technical
efficiency showed that most of the hazelnut farms
were both operationally and scale inefficient. Most REFERENCES
hazelnut farms could become more efficient by Akturk D 2000. An investigation on measurement of
adjusting input use and operation scale. efficiency of cotton in farms of soke district.
The most important determinant of economic Unpublished PhD thesis, University of Ankara,
efficiency was sucker control. The age and education Ankara, Turkey (in Turkish).
270 New Zealand Journal of Crop and Horticultural Science, 2009, Vol. 37
Anonymous 1992. Report of hazelnut planting area. Faculty Kilic O, Cinemre HA, Ceyhan V, Bozoglu M 2005.
of Economics, Department of Econometrics, Alternative optimum production plans instead
Ankara, Turkey (in Turkish). 75 p. of hazelnut in lowlands of Carsamba and Terme
Banker RD, Charnes AA, Cooper WW 1984. Some models Districts of Samsun Province. Prime Ministry State
for estimating technical and scale inefficiency in Planning Organization, Research Project Number
data envelopment analysis. Management Science TAP: 012, Samsun, Turkey (in Turkish).
30: 1078-1092. Kumbhakar SC, Lovell CAK 2000. Stochastic frontier
Battese GE, Malik SJ, Gill MA 1996. An investigation analysis: Cambridge, Cambridge University
of technical inefficiencies of production of wheat Press.
farmers in four districts of Pakistan. Journal of
Agricultural Economics 47: 37-49. Latruffe L, Balcombe K, Davidova S, Zawalinski K 2002.
Determinants of technical efficiency of crop and
Cinemre HA, Ceyhan V, Bozoglu M, Demiryurek K, livestock farms in Poland. Working Paper 02-05,
Kilic O 2006. The cost efficiency of trout farms Institut National de la Recherche Agranomique,
in the Black Sea Region, Turkey. Aquaculture France.
251: 324-332.
Cinemre HA, Ceyhan V 2006. Application of utility- Laura Gow MS, Langemeier M 1999. An efficiency
efficient programming to determine economic analysis of cattle back grounding in Kansas: paper
efficiency of Turkish farmers in the Central presented at Western Agricultural Economics
Anatolian Region of Turkey. New Zealand Journal Association Annual Meeting, Fargo, ND.
of Crop and Horticultural Science 34: 381-391. Ozcelik A, Ceyhan V, Bozoglu M, Cinemre HA 2006.
Coelli T, Rao DsP, Battese GE 1998. An Introduction to Technical efficiency of vegetable farms in Samsun.
efficiency and productivity analysis. Massachusetts, 7. National Agricultural Economics Congress,
United States, Kluwer Academic Publishers. Antalya, Turkey (in Turkish).
Demirci S 2001. Performance analysis of sugar factories Pinherio A 1992. An econometric analysis of farm
and total factor productivity: an application of level efficiency of small farms in Dominican
malmquist index. Project no. 2001_17, Publication Republic. Unpublished MSc thesis, University of
no. 66. Agriculture Economics Research Institute Connecticut, Storrs.
of Turkey, Ankara, Turkey (in Turkish). 45 p.
Sidhu SS 1974. Relative efficiency in wheat production
FAO 2007. Data base. http://www.fao.org [accessed 26 in the Indian Punjab. American Economic Review
November 2007]. 64: 740-751.
Farrell MJ 1957. The measurement of productive
efficiency. Journal of Royal Statistical Society Squires D, Tabor S 1991. Technical efficiency and future
Association 120: 253-281. production gains in Indonesian Agriculture. The
Developing Economies 29: 258-270.
Fiskobirlik 2006. Data base. http://www.fiskobirlik.org.tr
[accessed 25 December 2006]. Thanassoulis E 2001. Introduction to the theory and
Greene WH 2000. Econometric analysis. 4th ed. New application of data envelopment analysis, a
York, Prentice Hall International, Inc. foundation text with integrated software. Norwell,
United States, Kluwer Academic Publishers.
Gunden C, Miran B, Sari MA 1998. The development of
productivity and efficiency in turkish agriculture: Turkstat 2006. http://www.tuik.gov.tr [accessed 25
an application of data envelopment analysis. December 2006].
National Conference of Agricultural Economics
Association, October 7-9, Ankara, Turkey (in Tzouvelekas V, Christos CP, Christos F 2001. Technical
Turkish). efficiency of alternative farming systems: the case
of greek organic and conventional olive-growing
Helfand SM 2003. Farm size and the determinants of farms. Food Policy 26: 549-569.
productive efficiency in the Brazilian Center-
West. Selected Paper at the 25th International Yamane T 1967. Elementary sampling theory. NJ,
Conference of Agricultural Economist, August United States, Prentice Hall, Inc.
16-22, Durban, South Africa.
Zaim O, Cakmak E 1998. Efficiency in turkish agriculture:
Kalirajan K, Flinn JC 1983. The measurement of farm trend and comparative analysis. Agricultural
specific technical efficiency. Pakistan Journal of Structure and Employment in Turkey, Ankara (in
Applied Economics 2: 167-180. Turkish).