Portal de Periódicos da CAPES

A Monte Carlo Study of Alternative Estimates of the Cobb-Douglas Production Function

1963; Wiley; Volume: 31; Issue: 3 Linguagem: Inglês

10.2307/1909977

1468-0262

J. Kmenta, Malpeli Joseph,

Economic Growth and Productivity

THE MAIN critical comment of Hoch follows from his emphasis on the use of production function estimates for the purpose of testing for efficient allocation of resources.His claim that Klein's estimates are inappropriate in this case is, of course, correct.However, the article was not concerned with any specific field of application-its aim was simply to examine the small sample behavior of several available estimates.Klein's estimates were included because they can legitimately compete with other types of estimates in all applications other than tests for efficiency of allocation.Such applications were thought to be fairly common.Apart from models designed for prediction, economists may be interested in having "good" estimates of marginal productivity of a given type of input for the purpose of comparison with other inputs or other industries or regions.Those concerned with tests for optimality of resource allocation should, of course, limit their attention to estimation methods other than Klein's.The models used in the experiment were chosen so that, in the opinion of the authors, they would represent reasonably realistic situations.Our choice of the variances of the disturbances was based on the supposition that it is common for the "economic" disturbances to be more, or at best equally, spread out compared with the "technical" disturbance.The same contention was often made in economic literature and was also implied in the selection of combinations considered by Hoch in Table I of his 1958 article.Further, the sum of a, and a2 was made equal to 0.9 because it was believed that in a number of industries the returns to scale are fairly close to being constant.Subject to this condition, the value of a, was determined by expediency as mentioned in the text.As for the average firm, it was thought unlikely that it would lie a great distance from the optimum position, which in our models implies a profit of 10 units.Hoch's alternative suggestion of M$=