How to detect outliers in data envelopment analysis by Kourosh and Arash method

TL;DR

This paper introduces a new, simple method using the Kourosh and Arash Method (KAM) to detect outliers in Data Envelopment Analysis (DEA), addressing previous computational challenges and providing a clear definition.

Contribution

It proposes a novel, computationally efficient approach to identify outliers in DEA using KAM, with a clear definition and practical example.

Findings

KAM effectively detects outliers in DEA data.

The method avoids high computational complexity of previous techniques.

Results challenge the belief that DEA cannot detect outliers.

Abstract

One of the concerns about using non-parametric estimators such as Data Envelopment Analysis (DEA), is the presence of outliers. There are a good number of studies that mention this assessment in the literature of DEA, however, there is no clear definition to identify what outliers are in DEA. Moreover, most of the studies have used additional procedures which have high computational complexities. This paper proposes a suitable definition to identify outliers as well as a simple methodology to illustrate how DEA, by using Kourosh and Arash Method (KAM), is easily able to detect outliers without using additional technologies and their computational complexities. The methodology of detecting outliers by KAM is represented with an example which was used in previous research to depict DEA's weakness of detecting outliers. The results clearly reject this claim that DEA is not able to detect outliers.