Measuring and Decomposing Productivity Change
提出用随机投入距离前沿测量生产率变化，不限制规模报酬，并推导出与技术变化和效率变化的分解方法，适用于投入产出可能内生的情况。
Measuring productivity change with Malmquist indices has become common practice, because they are easily computed using nonparametric programming techniques and can be readily decomposed into technical and efficiency change. However, this approach is nonstochastic and requires a constant returns to scale assumption to construct the reference technology. We propose estimating productivity change using a stochastic input distance frontier, imposing no restrictions on returns to scale. We derive the analogous decomposition of productivity change and develop a generalized method of moments strategy in which outputs or inputs may be endogenous. We compare two methods in an application to electric utilities.