Differences between real and integer production possibility sets in data envelopment analysis

TL;DR

This paper explores the differences between real and integer production possibility sets in data envelopment analysis, highlighting how integer PPS may include points not dominated by observations, and discusses mixed-integer linear programming approaches.

Contribution

It introduces the construction of the integer PPS and addresses issues in recent mixed-integer linear programming methods considering integer axioms.

Findings

Integer PPS can contain points not dominated by observations.

The paper defines mixed-integer linear programming for PPS.

It identifies shortcomings in current mixed-integer LP methods.

Abstract

The real Production Possibility Set (PPS) is completely generated by observations in the real approach, but generating the integer PPS may not be possible by only using observations in the integer approach. In other words, this phenomenon says that all points in the real generated PPS are dominated by a point of the linear combination of observations, whereas there might be some points in the integer generated PPS which are not dominated with the points of the linear combination of observations. This paper shows how the integer production possibility set is made and the mixed-integer linear programming is defined. The paper also addresses some shortcomings in the recent mixed-integer linear programming while the integer axioms are considered.