Estimation of Multi-output Production Functions with Incomplete Data: A Generalised Maximum Entropy Approach.

One problem commonly encountered when analysing multiproduct-multifactor firms is the lack of activity-specific input data. The present study contributes to the empirical literature on production functions by presenting an information theoretic approach to estimating production functions in the absence of activity-specific input allocations. The procedure relies solely on production data, so that behavioural hypotheses (e.g., profit maximisation) are not required and estimation can proceed in the absence of price data. In addition, the method can be applied to ill-posed problems. The proposed procedure is demonstrated with simulated production data and exhibits favourable statistical properties. Copyright 1998 by Oxford University Press.