INTRODUCTION
Wheat is the second most important cereal crop in Bangladesh. Average per capita per day calorie intake is 2240 kcal (estimated), of which a significant percent comes from wheat. About 5% of total cultivable lands are utilized for wheat production. Crop sector contribute about 14% to the country's GDP of which a remarkable portion is contribute by the wheat^{[1]}. In 20012002 crop year Bangladesh produced 1903 thousand tons of wheat whereas in the same year imported 2424 thousand tons of wheat, which is 56% of total demand for wheat^{[1]}. To meet domestic consumption of wheat. Bangladesh has to import increased amount of wheat every year, strikes our foreign currency as well as balance of trade. In the present economic condition, it is our striking need to increase total production of wheat to keep pace with the demand for wheat. Due to the continuous pressure on the demand for the wheat, the government of Bangladesh used to import wheat for neighboring countries. Sometimes we used to here that the imported wheat is not of good quality and some portion of it is not congenial for human consumption. The low quality of wheat jeopardizes human health. At the present context, we will have to increase wheat production several times more than the present volume of production. We will have to explore and use all avenues and growth promoting factors for sustainable growth of wheat. Production of wheat in general can be increased in different ways. Increasing cultivable area can increase first production of wheat. But increase in wheat production by increasing area is not possible since total cultivable area is decreasing day by day due to the increased used of land for nonagricultural purposes.
Second production may be increased from increased use of inputs. But farmers of Bangladesh face resource limitation. Improving the production technology without increased use of inputs can increase third wheat production. This improvement consists of improved package of inputs, such as improved water management, HYV seed, chemical fertilizer, agricultural credit integrated pest control and appropriate land tenure systems. But production technology of developing countries cannot be changed rapidly due to several factors ranging from institutional to economic and from physical to natural. Production of wheat cannot be advanced by adopting the technology for certain economic condition of Bangladesh.
Fourth, increasing the productivity of inputs by reallocating and combining
them optimally without changing total quality of inputs and technology can also
increase output. This technology is generally termed as efficient production
technology, which is the main concern of this study. That is increasing the
technical efficiency of wheat using existing can increase production. The possibilities
of economic growth solely through the more efficient use of existing resources
will obviously be exhausted when an efficient production technology is reached.
In others words, the process of increasing wheat output only by improving efficiency
can not continue indefinitely, since under perfect technically efficient conditions
the frontier output level will be reached.
Table 1: 
Distribution of the two alternative production frontier model
of wheat farmer in Dinajpur, Bangladesh 

SPFM: Stochastic Production Frontier Model, TEPM: Technical
Efficiency Predictor Model 
Thus, other growth promoting strategies need to be considered when it is not
possible to increase wheat output only through efficient utilization of existing
resources. The use of modem technology in agriculture to raise wheat output
per unit of input is one such strategy. A sound and realistic agricultural policy
is the one of the most important instruments through which agricultural production
can be increased.
The Dinajpur district was bounded on the south by Joypurhat zilla and north by Thakurgoan on the west Rangpur, Gaibanda, Kurigram district. The cluster survey sampling methodologyandy adapted for the selection of the sample. A total 2686 sample household from 10 clusters has been interviewed. A study of Dinajpur district survey can be seen in Table 1. We tried to fit the models mentioned earlier. The accuracy of the identification of the impact of different factors on the efficiency effects depends on factors including the functional form of the production.
Model speciflcation: In this study a CobbDouglas stochastic production frontier is specified with a composed error term, by Aigner, Lovel and Schmidt^{[2]}, which is given below:
Where, y_{i} and X_{ki}, indicates the Output and inputs, respectively
(i= 1,2, ............., N farms and k ......, n number of inputs); β_{0}
and the β_{k} are parameters to be estimated, v is a random error
term and u is nonnegative random variable assume to represent technical inefficiency
in production. To estimate the parameters of this model using maximum likelihood
one must select distributional forms for the two error terms (v and u). The
most commonly made assumption are those the random error term, v is independently
and identically distributed as N (0, σ^{2}_{v}) and the
nonnegative inefficiency random variable, u, is distributed independently of
the v and has a half normal. distribution. That is, it has a distribution equal
to the upper half of the N (0, σ^{2}_{u}) distribution.
The intuition behind the error component specification is that any deviation
from the frontier caught by the technical efficiency term, u is the result of
factors under the firm's control, efforts of the producer. employees and factors
such as defective damaged product^{[12]}. However the frontier itself
can vary randomly across firms due to the random error v. On this interpretation,
the frontier is stochastic, with random disturbance v, being the result of favorable
or unfavorable external events such as luck or climate. Moreover errors of observation
and on measurement of production constitute another basis for the presence of
v in the frontier model. Given the definition of the stochastic frontier production
function in Eq. 1, we note that the realizations of the %
are not observable. That is, following the estimation of the unknown parameters
of the model defined in Eq. 1, the residuals of the model
will be realizations of ε_{i}= v_{i}u_{i} not
of u_{i} . Coelli et al.^{[3]} observed that a best predictor
for u_{i} is the conditional expectation of u_{i }given the
value of ε_{i}= v_{i}u_{i}. That is, one may defined
the firmspecific technical efficiency predictor using:
The above model subsequently has an influence upon their technical efficiency level. In order to take into account this situation we consider two alternative approaches:
Case 1: Assume that environmental conditions or factors influence the shape of the production technology or assume to represent technical inefficiency in productionas mentioned above.
Case 2: Assume that environmental conditions or factors influence the shape of the production technology.
Case 3: Assume that environmental conditions or factors influence the firm's technical efficiency.
Case 4: In case 2 we consider that the environment has a direct influence on the production structure and model the technology by introducing by some representative variables aside the production factors. It is assume that in this case firm faces a different production frontier. In terms of Eq. 1 and assuming that M (firmspecific) factors representing the environment, Z_{j} enter in a simple longlinear way i8n the production frontier, we will have a modified production frontier:
where, the θ_{j} are parameters to be estimated. When Eq. 2 is used to define predictors of technical efficiency relative to the frontier model defined in Eq. 3 the technical efficiency measures obtained will be net of environmental influences. That is, this technical efficiency may be termed as net technical efficiency. One may also obtain measure of gross efficiency (i.e., inclusive of environmental influences) by reevaluating the technical efficiency predictors with.
Case 5: In other studies^{[4,5]} environmental factors are assumed to directly effect technical efficiency.
Then the underlying hypothesis is that all firms share the ame technology represented
by the production frontier (1) and the environmental factors have an influences
only one the distance that separate each firm from the best practice function.
When Eq. 2 is used to defined predictors of technical efficiency
relative to the frontier model in Eq. 1, the predicted technical
efficiency is usually termed as gross technical efficiency.
Application: Both the estimation of stochastic frontier model production function and the prediction of the technical efficiency models are applied to the data of Bangladesh for different gross and net technical efficiencies of wheat production in which the first stage involves the specification and estimation of a stochastic frontier production function or management factors (Table 1).
The efficiency effects are not identically distributed. A more appropriate approach involves the specofication of a model in which both relations are estimated in a single stage.