Published online by Cambridge University Press: 17 August 2016
A number of attempts have been made recently (see, a.o., Mulkay [1983], Artus and Muet [1984], Gérard and Vanden Berghe [1984]) to develop and estimate appropriate specifications for investment functions in a rationing context. Some of the specifications proposed are however clearly conflicting, not in the usual sense that they reflect conflicting theoretical premises (between which empirical testing might hopefully discriminate) but in the unacceptable sense that there exists some logical incompatibility between different specifications derived from the same theoretical framework.
In this note, we focus our attention on two specifications among the most significant ones proposed recently, show why they are contradictory and then explain what is the correction to be brought to one of them in order to reconcile both specifications. In the course of the discussion, we also seize the opportunity to point to common empirical practices which are clearly unwarranted and could account for some deceptive results encountered (even with a theoretically correct specification).
Université Catholique de Lille, and CORE, Louvain-la-Neuve.
This paper has been written in the context of a research project on investment behaviour in rationing models, which is jointly supported by the Commissariat Général du Plan, France and the Commission of the European Community. I would like to thank J.H. Drèze and two anonymous referees for very useful comments, but the usual disclaimer of course applies.