Advances in the productivity with which food is produced around the world h
ave been made possible through the intensive use of industrial inputs that
have important environmental impacts. Like standard measures of macroeconom
ic performance, however, commonly used measures of agricultural efficiency
and productivity account only for marketed commodities and inputs, but igno
re the environmental effects of these production processes. A more complete
analysis of trends in the sector's productivity requires the use of models
that incorporate these environmental effects to provide better measures of
the contributions of the sector from the social point of view. This paper
compares the conceptual merits and empirical performance of alternative app
roaches that can be employed for this purpose: input distance functions, ou
tput distance functions, nonparametric methods, and index number approaches
. Each of the methods has relative strengths and weaknesses. The methods ar
e empirically illustrated using data from the Canadian pulp and paper indus
try. (C) 2001 Elsevier Science B.V. All rights reserved.