Production function is used for optimizing purposes. Such functions are established by parametric or non-parametric approaches. In this study non-parametric method i.e. Data Envelopment Analysis (DEA) has been used to optimize the applied energy in wheat production system in Punjab. Two basic DEA models i.e. input-oriented CCR (Charnes, Cooper and Rhodes) and BCC (Banker, Charnes and Cooper) models were subjected on operation-wise and sourcewise energy inputs. BCC model could classify the inefficient farmers to different categories of returns-to-scale (RTS), namely, increasing, decreasing and constant. Results revealed that the yield was mostly dependent upon tillage, irrigation and sowing, respectively, with the mean share of 23, 21.6 and 17.4% for CCR efficient farmers and 26.4, 19.2 and 18% for BCC efficient farmers in all zones. Also the yield was sensitive to human, diesel and seed energy sources with mean shares of 32, 18 and 18.6% for CCR efficient farmers and 37, 23.6 and 12.4% for BCC, respectively. Weeding and fertilizer application used energy in excess by 46% and 31.3% for CCR inefficient farmers and by 34% and 29% for BCC, respectively. Electricity and machinery possessed the highest use among all energy sources. It was 43.8% and 21.7% for CCR inefficient farmers and 33.6% and 15.5% for BCC ones, respectively. Trend of excessive use of energy was the same for both the models.
