中国産業連関表を地域化するのにFLQは有効か?

抄録

This paper examines the use of location quotients （LQ） for constructing regional input-output tables from national input-output tables. The focus is on the modified FLQ formula proposed by Flegg and Webber （1997） and augmented version of the AFLQ formula proposed by Flegg and Webber （2000）.
China consists of 31 municipalities, provinces and autonomous regions （excluding Hong Kong and Macau）. China regions range in size from very small （Qinghai and Ningxia, 0.3% in GDP） to relatively large （Guangdong, 11.3% in GDP）. It is very useful to verify the effectiveness of regionalization method from national input-output table to regional input-output table, for constructing small regional input-output table at the city or prefectural level in future analyses.
This paper tests the regionalization method with location quotients such as Location Quotient （LQ）, Cross Industry Quotient （CIQ）, Flegg’s Location Quotient （FLQ） and an augmented FLQ （AFLQ） by using China national input-output tables, and examines the accuracy of the output multiplier derived by the estimated regional input-output tables when compared with survey-based Provincial input-output tables.
The results showed that the LQ and CIQ of 10 provinces （one third of total region） are better than FLQ, and AFLQ gives better estimates in all regions under the appropriate δ value than LQ and CIQ. However, a reasonable value for δ as discussed in the previous research would be 0.05-0.075 for FLQ and 0.15 for AFLQ based on the average from my empirical evidence. The relative low value of δ and its valiance implies that regional size may not be very important.
In conclusion, AFLQ should be used to construct small regional input-output tables at the practical level.

JEL Classification:O53, R10, R15