a) Of course risk compensation is real. Usually it's the pro-helmet faction that argues that it's not (as counterpoint to those who say they would behave more dangerously if they had a helmte). Of course their very argument, that cycling is too dangerous without a helmet, proves that we are liable to increase our risk if we believe ourselves to be protected.
b) Can you be sure that a helmet helped you? That's the thing: if you hit your head and weren't injured, there's no evidence that had you hit your unhelmeted head you would have been hurt. And the converse is true. It's like saying I knocked on wood and haven't been run over yet, so clearly it really is lucky. You're an economist, you should know better. What would prove it is to have a well-designed, non-biased study that compared similar types of accidents between sufficiently large helmeted and unhelmeted populations and assessed efficacy based on that. Said study would need to control for cyclist ability, environment, age, sex, and probably a few other things that I can't think of.
It seems like they should, common-sense-wise--which is why I wear one--but the dramatic increase in helmet use rates has not had any discernable impact on fatality rates anywhere once you control for changes in cyclist populations.
Ken Kifer has a nice exploration:
http://www.kenkifer.com/bikepages/advocacy/mhls.htm