Fast and reliable computation of generalized synthetic controls

Given that existing implementations of synthetic control methods are plagued by serious weaknesses, new methods are offered for calculating synthetic control units. In particular, it is shown how to detect and handle important special cases that have yet to be addressed. Numerical methods for fast and reliably solving the nested optimization associated with the standard case are also discussed. An open source implementation of the presented methods is provided with the R package MSCMT, which can also be applied to generalizations of 'standard' synthetic control methods.