Nonparametric measurement of the overall shift in the technology frontier: an application to multiple-output agricultural production data in the Brazilian Amazon

摘要

This article develops a simple, but informative, approach to measure technical changes by applying the Malmquist index framework using data envelopment analysis (DEA). A set of directional vectors is used to capture complete information regarding the overall shift in the technology frontier; we then visualize the DEA frontiers in the output space. This approach deals with the problems of sample dependence and non-circularity of technical change measures. It allows us to measure output-by-output technical change. Furthermore, the geometric mean of the technical change measures derived using this approach satisfies circularity. The application of Malmquist indices to the panel datasets of agricultural production in the Brazilian Amazon during the period 1975-1995 indicates a non-Hicks-neutral technical change, with intersection of frontiers during both the 1975-1985 and the 1985-1995 sub-periods. We find moderate progress in cattle production and annual crops, and a decline in perennial crops (bananas). Technological progress is a modest 0.7 % during the period 1975-1995 on average. In addition, comparisons with the sample-dependent measures in the preceding studies demonstrate that these indices are usually sensitive to the distribution of actual sampled data points.