Ralf Martin (LSE and CeRiBA)

Measuring the true spread

How much of the observed labour productivity spread is real? This paper develops a novel framework to calculate productivity differences between plants which are due to differences in TFP and/or idiosyncratic demand shocks and not due to measurement error in variable factors or substitution between labour and other factors. The framework simultaneously accounts for imperfect competition, variations in output prices across plants and endogeniety of factor inputs. For UK manufacturing as a whole I find that on average 58 percent of the labour productivity spread is explained by TFP and demand shocks. Measurement error accounts on average for 9 percent of labour productivity spreads. This masks considerable heterogeneity across 3 digit sectors with measurement error accounting for as much as 26 percent in some sectors.

Session: 6a   Auditorium   Category: Evidence Based Policy Fund Productivity Microdata Project

Paper to follow