Using administrative data for statistical modeling: an application to tax evasion

Administrative data, gathered by public authorities with a general aim of control, are very precious sources of information because they allow to study phenomena that would remain otherwise unknown. On the other side, administrative data strictly contain the information they were collected for, and to be used for statistical purposes they need to be integrated. This work shows the potentials of the integration of three data sets for statistical modeling: the audits carried out in Italy in 2005 by the National Institute of Social Security on building and costruction companies, the ASIA archive of Istat and the ”‘Studi di Settore”’ of the Italian Revenue Agency.