COBB-DOUGLAS AND TRANSLOG PRODUCTION FUNCTION: A COMPARISON OF EMPIRICAL PERFOMANCE

The objective of this paper is to test the specification of functional forms for farm production, in particular the adequacy of the Cobb-Douglas form, within the context of Taiwan's developing agriculture. To that end, the Cobb-Douglas production function is initially assumed as a working hypothesis and estimated from the data. The translog function which encompasses the Cobb-Douglas as a special case is then estimated from the same data. Examination and comparison of estimates from these two functions provide a test for the validity of the Cobb-Douglas relation. Statistical measures and economic considerations are used as criteria of empirical performance. The functional form most frequently used in production studies for agriculture in developing economies is the Cobb-Douglas. One major reason is that it often gives a good fit to data, which in turn is used as evidence of its close approximation to the 'true' function. The choice of a functional from on a priori grounds without taking into account its inherent restrictions might result in the erroneous specification of a single-equation production model.