I think it can be even worse than that. What I've often seen is that "failure" isn't even real business failure, but merely your boss having your idea shot down by the boss above him. The risk is mainly emotional. People train themselves to play an elaborate political game to advance their ideas.
When designing any system that involves people, it's vital to see what behaviours optimise your rewards. Usually there is some general process / overall plan you're supposed to follow:
Follow the plan | Don't follow the plan
Things go well | reward | unknown
Things go badly | no punishment | punishment
That is a setup which encourages people to follow the plan even if they think it's not the best one.
Something else which makes this quite difficult is that you never know what would have happened if you'd jumped columns.
Big companies are only prepared to give you X% of your annual salary as a bonus if you're good (with X < 15%). So the bigger the upside of your "innovation", the more you will find the company and its reward model unfair and leave.
So big companies lose the best employees and so have to buy back innovation by buying start-ups.
