This paper studies procyclical productivity growth at the industry level in the U.S. and in three European countries (France, Germany and the Netherlands). Industry-specific demand-side instruments are used to examine the prevalence of non-constant returns to scale and unmeasured input utilization. For the aggregate U.S. economy, unmeasured input utilization seems to explain procyclical productivity. However, this correction still leaves one in three U.S. industries with procyclical productivity. This failure of the model can also be seen in Europe and is mostly concentrated in services industries.