“…Various selection criteria are available, but the most common method is to select the control group that minimizes the mean square prediction error (i.e., the pre-treatment deviance between the synthetic control group and the treated sector). Most studies use a donor pool from other countries (Castañeda and Vargas, 2012;Billmeier and Nannicini, 2013;Abadie et al, 2015;Karlsson and Pichler, 2015;Smith, 2015), other regions within the same country (Abadie and Gardezabal, 2003;Abadie et al, 2010;Leight, 2010;Bohn et al, 2011Bohn et al, , 2015Coffman and Noy, 2011;Bassok et al, 2012;Dorsett, 2013;Hankins, 2014;Ando, 2015;Dietrichtson and Ellegard, 2015;Munasib and Rickman, 2015), or other districts within a region (Bauhoff, 2014). One study uses a donor pool consisting of other persons (Chan et al, 2015).…”