Efficiency analysis in uncertain operating environments: the problem with outliers

In operating environments where there is a great deal of uncertainty, there can be great diversity in terms of outputs of firms even if inputs are similar. Some firms may achieve relatively high levels of output due to environmental effects that are unrelated to their efficiency. These firms may be treated as outliers. This chapter implements an innovative approach to outlier detection in two separate case studies, using data for a fishery and an irrigated agriculture industry. The approach recognizes that a firm may be an outlier either in terms of the mix or scale of its input-output vector and provides distinct measures of dissimilarity relative to its peers. This distinction is particularly relevant to non-stochastic frontier methods such as data envelopment analysis (DEA), since it identifies the firms that exert the most influence on the resulting efficiency scores. Such identification allows either for an explicit adjustment of the scores for environmental factors, or the exclusion of the outliers – and hence an implicit adjustment of the scores that would otherwise have resulted.