Creation of municipality level intensity data of electricity in Japan

An assessment of residential CO2 emissions is typically performed through the intensity method, in which total energy consumption is estimated by multiplying floor space by intensity value. Although spatially detailed intensity data is required for an accurate estimation, the finer spatial resolution will result in a less stable estimated value due to the small sample size; hence, existing studies in Japan created intensity data at a regional or prefectural level. The objective of this study is to create municipality level intensity data via a statistical approach, using the household level micro data from the National Survey of Family Income and Expenditure, of the Ministry of Internal Affairs and Communications, Japan, by focusing on electricity. First, this study builds several (spatial) statistical models, where per household electricity expenditure is regressed on housing types (two categories), household types (seven categories), and other household specific variables. Second, by substituting averaged municipality level explanatory variables from official statistics into the model, it estimates municipality level intensity data. The obtained results suggest that conventional intensity data in Japan, created by the simple average of samples in each unit (prefecture), may suffer from an upward bias, suggesting a danger of overestimation of residential CO2 emissions.