Confidence Intervals Estimation for Revenue Malmquist Index

The result of non-parametric Malmquist index and its decompositions would not be always precise enough when the number of decision making units (DMU) was limited. To decrease its influence, this paper adopts the kernel method to generate large number of consistent bootstrap pseudo samples, and estimates the confidence intervals of the revenue Malmquist index. The result indicates that the efficiency evaluation method, which is based on the production frontier, can effectively lower the influence derived from the number of DMU. As a result, this confidence intervals estimation method enriches the Malmquist index research a lot.