Substitution Elasticities in the Large-Scale Manufacturing Industries of Pakistan: A Comment (Communication‪)‬

In the Spring 1981 edition of this Review, A.R. Kemal presented evidence on 'substitution elasticities', estimated over the period 1960-70, in sixteen two-digit manufacturing industries in Pakistan [8]. After applying several variants of the basic 'production function' approach, the author argued that the elasticities of substitution in Pakistan were generally low and/or insignificant and therefore a manifestation of Pakistan's 'technological dependence'; that these results are 'consistent' with similar estimates for other developing countries; and that because of the low value of the elasticities, the removal of factor price distortions, while necessary to the attainment of a more labour-intensive pattern of development, needs to be supported by policies aimed at subsidizing the developing of indigenous technologies. While the above summary of the contents of A.R. Kemal's paper may seem innocuous enough, the actual reading of the paper has left me with a profound sense of unease and disagreement. This arises from several considerations : the use of a highly controversial data base without a warning to readers of its serious statistical pitfalls and without acknowledgement of three earlier published articles on the subject in this Review; the uncritical application of a 'production function' approach without setting out in clear, unequivocal terms the various assumptions that are crucial to sustaining it; the, unnecessary preoccupation with the actual task of measurement rather than focusing on the implications of, and the interpretations that might be attached to, the results; the use of a selective, purposive sample of other developing-countries estimates to argue for the 'consistency' of the results; the failure to provide the reader with sufficient evidence on the statistical properties of the data used; and the failure to match the results with such evidence as exists on the movement of relative factor income shares in manufacturing and the 'bias' in technical change. I should, in this comment on Kemal's work, like to elaborate on these, and related, issues.