The degree of localization of manufacturing, defined as the excess geographic concentration remaining after correcting for both sectoral concentration and the agglomeration of overall economic activity, has recently gained new techniques of measurement in the form of indexes and numerical methods. These techniques are illustrated, compared, and theoretically discussed to show that they are consistent advancements, but none is yet able to satisfy all desirable properties. The measurement of localization would be more interesting if it could be related to economic processes and not limited to simple observational aspects. In achieving this, there are a number of complications, which are reviewed in the paper. Among these, the choice of the right sectoral scale at which these measures have to be applied is normally given too little importance. The second part of this paper advocates that this is a complex issue, since no digit level is appropriate for all industries. Moreover, different aggregations of narrowly defined subsectors may be more suitable than the standard industrial classifications. This is supported by an exploratory analysis, in two British sectors, based on the investigation of possible localization reasons identified in recent taxonomies. As the many localization measurement techniques have become available, it is recommended that they be applied to sector studies, and not confined to countrywide extensive explorations of all sectors.
