Beware of the selection bias: even if veterans show more productivity, it could just be because the military training has selected those with higher discipline
Data from periods of forced conscription would correct for that bias, but would introduce the new bias of a 4-F control group. Is there a fancy statistical trick to combine the data and eliminate both biases?
Draft lotteries provide randomized experiments. The most famous one is the Vietnam draft analysis by Angrist 1990 where he finds a 15% penalty to lifetime income caused by being drafted. There’s also little evidence of any benefit in Low-Aptitude Men In The Military: Who Profits, Who Pays?, Laurence & Ramberger 1991, which covers Project 100,000 among others, which drafted individuals who would not otherwise have been drafted either as a matter of policy or, in the ASVAB Misnorming, unrealized accident. (As Fredrik deBoer likes to say, if there’s any way it can possibly be a selection effect, then yeah, it’s a selection effect.)
This suggests we might look at data from countries like Israel and Singapore with mandatory military service (although it’s complicated because I think both have alternative civil service for objectors and exceptions for certain groups), and look at how results of veterans there compare with similar populations to hold other cultural effects on discipline constant.