Analysis of port efficiency using imprecise and incomplete data

Port efficiency assessments based on data envelopment analysis (DEA) usually assume that the inputs-outputs are measured using known and crisp data values. However, the data accuracy, the imprecision, and the missing values are common problems in many port efficiency assessment applications, particularly when the data are drawn from various heterogeneous sources. In those cases, common practice so far was to use either approximated values or to completely exclude these ports from the analysis. This paper proposes Imprecise DEA (IDEA) to assess the efficiency of ports when imprecise and missing data appear in a port assessment problem. In this approach, the missing or imprecise data are replaced by interval or ordinal data, properly estimated using auxiliary data. In a post-DEA analysis stage an iterative procedure is developed with the purpose of estimating new interval bounds that turn non-efficient ports into efficient. This methodology is put into practice through an application that assesses the efficiency of a port sample with missing and imprecise data in the year of reference.