Abstract
Taking an approach with Input-Output （IO） models to analyze local socio-economic effects induced by implementation of technologies and policies requires localized IO Tables （IOTs） that better reflect regionally-specific economic structures. Because a complete localization is costly, it is necessary to prepare them with various assumptions on localization. Here, we examine discrepancies caused by variants of IOTs, with a case study of an envisioned transformation to decentralized energy system on Tanegashima. Three variants of IOTs were prepared; local scope （Case I, benchmark）, prefectural scope （Case II） and a hybrid scope that is localized for a limited number of sectors while others taking the prefectural scope （Case III）. The increase in the total output induced by the transformation of energy system was 27.2% underestimated in Case II compared with Case I. This was caused by incorrect input structure in small number of sectors that supply resources to decentralized energy sectors. Therefore Case III which improved the input structure successfully reduced the discrepancies （+7.4% to Case I）. For analyses on implementation of decentralized energy system, a hybrid approach that elaborates on resource-supplying sectors can provide a reasonably good forecast to local decision makers with a significantly less cost than preparing completely localized IOT.
