This paper uses the input-oriented Malmquist productivity index (IM) and the dual cost Malmquist productivity index (CM) to estimate and identify the sources of the productivity change in the Taiwanese banking industry. In this paper, a variable returns-to-scale production technology is assumed, and the CM productivity change is decomposed into pure technical efficiency change, technical change, allocative efficiency change, an input-price effect, and cost scale efficiency change. The proposed decomposition is a modification of the Maniadakis and Thanassoulis (2004) approach under the assumption of constant returns to scale technology. Empirical results show that between 1995 and 2003, the industry-wide IM were near 1.0, indicating no improvement in productivity in the banking industry. However, the estimate for CM shows a 5% productivity change, which is mainly due to cost saving reflected by the input-price effect and improvement in scale efficiency. This improvement in productivity in the Taiwanese banking industry would most likely be not identified were the analysis conducted using either the input-oriented (IM) or the output-oriented (OM) Malmquist productivity approach.