“…Some observations can be located on the efficient frontier in the deterministic DEA, while some stochastic inputs and outputs are by definition allowed to be around the efficient frontier can be allowed with the aim of conceptualizing the stochastic nature of the data into the model to adapt the measurement and specification errors. Stochastic input and output variations in DEA have been studied within various input-output DEA contexts by many scholars (see e.g., Olesen andPetersen, 2015, Olesen andPetersen, 1995;Huang and Li, 1996;Cooper et al, 1996Cooper et al, , 1998Cooper et al, , 2002Cooper et al, , 2004Land et al, 1993;Morita and Seiford, 1999;Sueyoshi, 2000;Talluri et al, 2006;Olesen, 2006;Bruni et al, 2009;Wu and Lee, 2010;Tsionas and Papadakis, 2010;Udhayakumar et al, 2011). Land et al (1993) were the first to extend the chance-constrained programming (CCP) DEA proposed by Charnes and Cooper (1959), in order to compute efficiency in the presence of uncertainty in which inputs are assumed to be deterministic and outputs are jointly normally distributed.…”