Learning by Doing While Remembering Forgetting, with Reminders from Pakistan Manufacturing Data
A.R. Kemal, Richard C. Porter · Deep Blue (University of Michigan) · 1981
The process of subsidized industrialization in many LDCs has been taking place for a sufficiently long time that it is becoming possible to test empirically whether these new and initially inefficient industries have been "learning by doing." Such tests are beginning to appear. Unfortunately, they are bedeviled by the difficulty of quantifying the concept, "learning." The surrogates usually used lead to the strange implication that learning is unbounded and hence that the merest iota of capital and labor is ultimately capable of learning how to produce infinite outputs (and doing it). Furthermore, the learning variable usually selected is highly collinear with time and hence is econometrically quite unsatisfactory. this paper explores these demerits and presents a new method that overcomes these problems by remembering forgetting. This new approach is employed to examine various learning hypothese for sixteen manufacturing industries and the total manufacturing sector of Pakistan.