In observational studies for policy evaluation, matching is used in service of causal inference to simulate randomization and thus reduce selection bias that might occur when treatment assignment differs systematically. ...