Researchers interested in estimating productivity can choose from an array of methodologies, each with its strengths and weaknesses. Methods differ by the assumptions they rely on and imply very different calculations. I compare five widely used techniques: (a) index numbers, (b) data envelopment analysis, and three parametric methods, (c) instrumental variables estimation, (d) stochastic frontiers, and (e) semi-parametric estimation. I compare the estimates directly and evaluate three productivity debates using a panel of manufacturing plants in Colombia. The different methods generate surprisingly similar results. Correlations between alternative productivity estimates are invariably high. All methods confirm that exporters are more productive on average and that only a small portion of the productivity advantage is due to scale economies. Productivity growth is correlated more strongly with export status, frequent investments in capital equipment, and employment of managers than with the use of imported inputs or foreign ownership. On the debate whether aggregate productivity growth is driven by plant-level changes or output share relocation, all methods point to the importance of plant-level changes, in contrast to results from the U.S.
MotivationProductivity is used and discussed widely. Ever since Solow (1957) decomposed output growth into the contribution of input growth and a residual productivity term, the concept has increased in popularity. Productivity has generated a lot of interest in its own right and is used as a benchmark to rank firms or countries. Such rankings gained credibility once other studies documented that productivity is correlated with other indicators of success such as employment growth, export status, or technology adoption. Low productivity has also been found to predict exit, the ultimate performance standard. Its importance can also be gauged from the attention it receives as a criterion to evaluate policy interventions or firms' decisions.In industrial economics, for example, a large literature investigates the effect of R&D on productivity and the resulting impact on industry structure. In international economics, efforts to evaluate the impact of trade liberalization has turned from estimating changes in price-cost margins to productivity changes.Fundamentally, the objective of productivity measurement is to identify output differences that cannot be explained by input differences. Because the production technology of each firm and the input tradeoff it allows, is not observed, our ability to control for input substitution is subject to error. In addition, inputs and outputs are likely to be measured with error, certainly in less intensively used data sets from developing countries. Methodologies for productivity measurement differ vastly in their sensitivity to measurement and specification error.
1I evaluate five widely-used methodologies, which fall in three broad classes. The first two, index numbers and data envelopment analysis, are flexible in the specific...