Over the last twenty years an accelerating number of studies have relied on the standard definition of the Malmquist-Luenberger index proposed by Chung et al. (1997) [J. Environ. Manage., 51 229-240], to assess environmental sensitive productivity change. While recent contributions have shown that it suffers from relevant drawbacks related to inconsistencies and infeasibilities, no one has studied systematically the performance of the original model, and to what extent the existing results are unreliable. This paper introduces the optimization techniques that allow implementing the first model solving these problems, and using a country level database including air pollutants, systematically compares the results obtained with both approaches. We discuss the relative number, magnitude and significance of the disparities that researchers should expect if resorting to the original model. Results show that inconsistencies and infeasibilities in the original model are increasing in the number of undesirable outputs included, reaching remarkable values that seriously question the reliability of results, and compromise any policy recommendation based on them.

APARICIO Juan;  BARBERO JIMENEZ Javier;  KAPELKO Magdalena;  PASTOR Jesús T.;  ZOFÍO José L.; 

2016-12-01

Publications Office of the European Union

JRC104083

978-92-79-63966-1,   

1831-9424,   

EUR 28246 EN,    OP LF-NA-28246-EN-N,   

https://publications.jrc.ec.europa.eu/repository/handle/JRC104083,   

10.2791/173984,