Evaluating the Ability of Interval Fuzzy and Robust Data Envelopment Analysis Models to determine the efficiency of Broiler Chicken Breeding Units in Khuzestan Province

Introduction Chicken breeding production is one of the most agricultural subsectors in Iran and has a special position in the production and employment of the agricultural sector. According to the importance of chicken as a strategic protein product, investing in this field is inevitable. Due to high growth rate of birds compared to other livestock, the possibility of producing in all weather conditions and rapid return on investment, the poultry industry has a high priority to other industries. Optimum use of inputs and existing possibilities lead to an increase in production and reduce the fixed price, thereby increasing the Country's competitive and export potential and also an increase in the0 welfare of the society. Since increasing production by inputs development and technical changes has many limitations in developing countries, improving technical efficiency has been emphasized as a suitable approach. So investigating the efficiency of chicken breeding units and trying to improve this efficiency and optimum use of inputs has been a matter of importance. The main objective of this study is to measure technical efficiency of broiler breeding in Khuzestan Province by utilizing the models of RDEA and FIDEA model. Also, the sub-objectives such as estimating the optimal use of inputs in inefficient units and comparing the capability of RDEA and FIDEA models when deals with uncertain data are obtained. Materials and Methods . These models also suffer from lack of point solutions, awareness of data distribution and ignoring the information of coefficient uncertainty respectively. To clear up the mentioned defects, the Robust Data Envelop Analysis (RDEA) model was developed in the late 1990s. In this study, FIDEA and RDEA models are applied to calculate the efficiency of decision making units under the condition of uncertainty. Moreover, Mont Carlo simulation is used to evaluate the capability of RDEA and FIDEA models against uncertain data and investigate their ability to deal with possible changes in input and output data. The necessary data were collected from the population of poultry production units in Khuzestan Province, by surveying a questionnaire of 105 production units that were selected using simple random sampling method. Results and discussion Based on the RDEA model output, the average technical efficiency of all production units in three probability levels of 0.1, 0.5 and 1 are 0.88, 0.91 and 0.93 respectively. The results also indicate that in three probability levels, the average efficiency of semi-industrial poultry production units is more than industrial ones. Calculating the technical efficiency by FIDEA show that in all levels of optimal use of resources, the high and low range of efficiency in semi-industrial poultry production units are more than industrial ones. Applying the inputs of drug, electricity and water cost, and area in industrial units and inputs of labor force and area in semi-industrial units are technically inefficient. The results of Mont Carlo simulation indicate that in RDEA model, by increasing system protection against uncertainty (decrease in p value), the ranking compatibility percentage is increased. The average ranking compatibility percentage for simulated data in all uncertainty scenarios in RDEA model is higher than FIDEA. Suggestion  According to the results, since industrial poultry units are less efficient at using inputs than semi-industrial ones, these units must be the references for inefficient units. Moreover, since the cost of drug, affecting the inefficiency of poultry units, is one of the most important inputs, it is recommended that these units should be given the necessary training to use this input. Furthermore, since the RDEA model has more capability to protect the system against uncertainty, it is suggested that all corrective actions should be planned with respect to the outputs of this model.  JEL Classification:: D61, Q1, C61, D81