Markups and measurement errors in six EU countries

This study of markups, ie prices over marginal costs in manufacturing industries, builds on the work of Robert Hall and Werner Roeger.We analyze several methods used in estimating sectoral markups, and then apply them to empirical analysis of the industrial sectors of six EU countries (Germany, France, Italy, the UK, Sweden and Finland).We argue that measurement errors in the model variables, particularly in the rental price of capital, are likely to be a major problem in markup estimation, and show that due to measurement errors, the approach developed by Roeger is likely to produce markup estimates with an upward bias.Such biased results are particularly deceiving since the outcome tends to produce artificially good fits, high t-values, and markup estimates which are "sensible" in magnitude. We also introduce a "modified" model for markup estimation, which would, if all assumptions were fulfilled and variables correctly measured, yield results identical to those obtained with Roeger's model.Yet, in contrast to Roeger's model, in the presence of measurement errors the markup estimates produced by this model have downward bias.Comparison of these two sets of estimates enables us to assess the seriousness of the measurement problem.We found that the estimates produced by the two models differed in a systematic fashion which is symptomatic of measurement errors.All in all, our results provide strong support for our hypothesis that the markup estimates obtained with Roeger's method are likely to be artifacts created by measurement errors mainly in the rental price of capital.