Area selection, data source and collection The study was based on primary data, were collected using a set of pre tested questionnaire in 2014. Based on preliminary information received, Cox’s Bazar, Chittagong and Bagerhat region were purposively selected, where large numbers of small scale marine fishermen are involved in marine fishing. In these areas, most of the fishermen use three types of gears or technology, such as Gillnet, Setbag net and Longline. Of them study considered only Setbag net fishing boats. A purposive sampling technique was followed for achieving the ultimate objectives of the study. A total of 100 samples of Setbag net fishing boats were selected, of which 40 from Cox’s Bazar, 40 from Chittagong and 20 from Bagerhat. Analytical technique- The translog production frontier functions were specified as the empirical model of this study. The Cobb-Douglas stochastic frontier model was found to be unsuitable to represent the data, while the translog model for Setbag net fishing provide better estimates. Translog function is very commonly used which is a generalization of the Cobb-Douglas function. It is a flexible functional form with less restriction on production elasticity and substitution elasticity. The stochastic translog frontier production function, described in equations 1 and 2, were estimated by maximum likelihood estimate (MLE) method using computer software STATA. All output and input variables, used in the production frontier analysis, were measured on a per trip basis as, except for trip days. While, all data on input variables were collected as per trip averages. Setbag net fishery involved multiple inputs, number of days at sea, crew size, fuel, ice, lubricant, number of net and other miscellaneous. However, for the purpose of this study, these inputs are aggregated into four categories; namely, trip days, crew size, total cost and number of net. In order to determine the effects of predetermine variable of marine fish catch by Setbag net fishing boat as well as the efficiency of resource used, the translog stochastic production function was estimated, which is given below: The linearised double-log form of (1) is InY = lnβ0 + β1InX1i + β2InX2i + β3InX3i + β4InX4i + δ1InL1i + δ 2InL2i + δ 3InL3i + Vi-Ui -----------(2); Where, Ln= Natural logarithm,Yi = Output (total catch kg/trip), β and δ are unknown parameter to be estimated, X1i=Fishing duration (days/trip), X2i= Number of crew (employed/trip), X3i=Total cost (Tk/trip), X4i=No. of net, L1i=Cox’s Bazar, L2i=Chittagong, L3i=Bagerhat, vi-ui= error term; Functional form of the technical inefficiency model: The empirical model of gear specific technical inefficiency is Ui= γ0+ φ 1Z1i+ φ 2Z2i+ φ 3Z3i+ φ 4Z4i+ φ 5Z5i+Vi- Ui ,- - - - (4); Where, γ and φ are unknown parameter to be estimated, z1i= Age of head maji’s(year), z2i= education of head maji’s (Years of schooling), z3i== Experience of head maji’s(year), z4i= Capacity of engine (Horse power), z5i=Boat used (Age of boat), Vi-Ui = error term. Age, education and fishing experience of head maji’s, which represents human capital, is generally postulated to have a positive impact on efficiency. Engine horse power and age of boat also have positive impact on efficiency. These common views the variables were select. It should be noted that the above model for technical inefficiencies in equation (2) can only be estimated if the technical inefficiency effects Ui are stochastic and have particular distributional properties (Coelli et al., 1996)