A lot of my work deals with conditional cash transfer, where payments in many Latin American countries are made to mothers on the condition that children go to school and go in for health checkups. Similarly in Florida now welfare payments are given on the condition that applicants pass a drug test. The analysis link suggested that after the cost is figured in of the drug test, with 96% plus of applicants passing the test the state is saving only $60,000 on a program that costs $178 million, so really it is in the rounding error.
Two things to consider about conditions.
1. Due they cost more to implement than the save? If the program in Latin America is designed to get kids to go to school, would more kids go to school if an unconditional transfer was given to more families than a conditional one is given to fewer? For these programs conditioning can make up 25% of the cost (link)
2. Perhaps conditioning is needed to build political support. Even if it makes the program more expensive and lowers coverage. It is possible people prefer a situation in Florida where welfare covers 100,000 drug free participants to one that covers 125,000 drug free participants and 2,000 drug users for the same cost.
Of course this assume rationality in Florida?