Abstract:
This study examines productive efficiency of each of the 49 commercial banks of Bangladesh for the period from 1999 to 2005 using two techniques, Stochastic Frontier Analysis (SF A) and Data Envelopment Analysis (DEA). Pure technical, scale and overall technical efficiencies are derived from solving input-oriented and outputoriented DEA models. The two techniques, SF A and DEA are compared on the basis of technical efficiency estimates. Based on the stochastic frontier, the average technical efficiency estimates of the commercial banks of Bangladesh for 1999, 2000, 2001, 2002, 2003, 2004 and 2005 are 75, 77, 82, 70, 76, 67 and 77 percent respectively. Based on input oriented DEA, the average overall technical efficiency, average pure technical efficiency and average scale efficiency of these banks for 1999 are 83, 93, and 89 percent; for 2000 are 89, 94 and 95 percent; for 2001 are 88, 95 and 92 percent; for 2002 are 83, 90, and 93 percent; for 2003 are 65, 88, and 74 percent; for 2004 are 74, 94 and 79 percent; and for 2005 are 73, 88 and 84 percent respectively while based on output-oriented DEA, the corresponding values are 81, 91 and 89 percent for 1999; 93, 97, and 96 percent for 2000; 85, 93, and 93 percent for 2001; 84, 88, and 95 percent for 2002; 74, 90 and 81 percent for 2003; 79, 94, and 84 percent for 2004; and 86, 97, and 89 percent for 2005 respectively. The technical efficiency estimates obtained from the stochastic frontiers are significantly lower than those obtained from DEA. Stochastic frontier results show constant returns to scale, whereas DEA results show both constant and variable returns.
To facilitate intra-category comparison all the banks are grouped into five categories, for example, Nationalised Commercial Banks (NCBs), Private Commercial banks (PCBs), Specialised Commercial Banks (SCBs), Islamic Private Commercial Bank (IPCBs) and Foreign Commercial Banks (FCBs) according to ownership forms and objectiv principles. NCBs are found the most efficient and the 'best practice' performer. SCBs are second best performer. Then appear the rank of FCBs. PCBs are found more efficient than IPCBs in case of private banking. However, their efficiency scores fluctuate over time and in their intra-category comparison. Given current output levels, NCBs would reduce 4-29 percent, PCBs 20-34 percent, SCBs 22-33 percent, IPCBs 17-38 percent and FCBs 15-36 percent of their input expenditures if they all achieved full efficiency according to SF A. If current resources were utilised at full efficiency, NCBs would increase their output by 25 percent, PCBs 41 percent, SCBs 38 percent IPCBs 39 percent, and FCBs 26 percent. According to outputoriented DEA, this study indicates that the banking sector of Bangladesh is potential for increasing banking output through improvement in efficiency.