Data Envelopment Analysis (DEA) is a non-parametric method to estimate (in-)efficiency
scores of ex-post production activities of comparable Decision Making Units (DMUs). In
the BCC model – the case of variable returns to scale (RTS) – one can determine the RTS
values for all DMUs; these values then permit information for deciding whether the
activities should be up- or downsized. But sometimes the RTS values are – from a
decision-maker's point of view – abnormally high or low and hence impracticable.
Dellnitz (2016) identified such kind of DMUs – named RTS-mavericks – via a medianbased
outlier test, and proposed scale restrictions for handling this situation. However, in
exceptional cases this deployment of lower and upper scale bounds does not necessarily
lead to a revised RTS value. In this contribution we focus on this problem and, therefore,
develop a method that bases on cross-evaluation to cure this deficit. The new concept
will be illustrated by a small numerical example.