A0763
Title: Location powered quotient: A compositional data analysis-based approach
Authors: Takahiro Yoshida - The University of Tokyo (Japan) [presenting]
Daisuke Murakami - The Institute of Statistical Mathematics (Japan)
Hajime Seya - Kobe University (Japan)
Abstract: A typical measure of industrial concentration is the Location Quotient (LQ), which is simply calculated as the regional and national ratios of employment in each industrial sector. However, its calculation focuses on a single sector and thus ignores relationships with other sectors. Therefore, we propose an alternative version of LQ based on compositional data analysis, which is commonly used and developed in geosciences. The proposed index, Location Powered Quotient (LPQ), has the following properties. (1) LPQ is derived from the powering operator in Aitchison's vector space structure, (2) LPQ considers not only specialization but also the balance of composition, and (3) LPQ has a sign. We apply this LPQ to an analysis of Japanese industry data to examine how the LPQ is interpreted.