Files

Abstract

The measurement of productivity fluctuations has been the focus of decades-long interest. In addition to broad structural forces driving productivity changes, there is more recent interest in measuring and identifying the heterogeneous forces driving these changes. A major force is learning-by-doing which is used by economists to describe the phenomenon of productivity growth arising from the accumulation of production experience by a firm. This paper proposes a bounded learning concept with the learning progress function characterized by the degree of efficiency and the specification of the learning progress as a logistic function capturing both the slow start-up and the limit in learning progress. The inter-firm learning inefficiency is defined as the inability of a firm to reach the optimal plateau relative to the ‘best practice’ firm from the set of comparable firms. We further differentiate learning efficiency from the technical efficiency. The key contribution of this research is to provide a measure the firm’s movement along the learning progress curve and explain the existence of firm-level heterogeneity in learning. The time varying technical efficiency is estimated based on stochastic production frontier methods and firm-specific learning efficiency is disentangled using the residual of the production frontier (productivity).The model is then used to decompose the factor productivity growth into components associated with learning, scale, technical efficiency, technological change and change in allocative efficiency. This productivity growth decomposition provides useful information and policy level insight in firm-level productivity analysis. The major econometric issue in production function estimation is the possibility that there are some forces influencing production that are only observed by the firm and not by the econometrician. With firm input use being endogenous, inputs might be correlated with unobserved productivity shocks. The measure of technical efficiency by estimating the production frontier directly in presence of endogeneity of input choice can be biased in the sense that the measure of efficiency favors the firms employing higher levels of inputs. The Levinsohn and Petrin (2003) approach is extended to overcome this simultaneity problem in stochastic production frontier estimation to generate consistent estimates of production parameters and technical efficiency. The model is applied to plant-level panel data on Colombian food manufacturing sector. The dataset is unique longitudinal data on firms in the sense that it has information on both plant-specific physical quantities and prices for both outputs and inputs. In contrast to most of the existing literature which measure productivity by deflating sales by an industry-level price index, these data eliminate a common source of measurement error in production function estimation. Plant-level productivity growth decomposition and the contribution of learning effect are explored by estimating the production frontier and firm-specific learning efficiency.

Details

PDF

Statistics

from
to
Export
Download Full History