The ability of current statistical classifications to separate services and manufacturing

The paper explores how well our statistical classification systems perform in classifying firms and in particular how they distinguish firms doing services and/or manufacturing. It is found that a large share, almost 20%, of firms can be said to be misclassified based on their statements on activities compared with the assigned industry codes. The result is robust to analyses on different levels of aggregation. The paper contributes with a quantification of misclassifications in contrast to previous literature. The paper provides explanations as to why grouping firms using only industry codes is not precise. The increasing complexity of production and the fact that manufacturing products and services are increasingly interwoven are some of the primary explanations.