Interestingly, a paper in the British Medical Journal (BMJ) from late 2011 shows that musicians (band members and solo artists) with a number-one album in the UK between the years 1956 and 2007 have consistently higher death rates than do comparable cohorts from the general UK population. For musicians in their 20's and 30's, the death rate is over twice as high. Now I have not read this paper really carefully, but from what I can gather its results are puzzling, because they seem to go against basic human capital theory.
Acquiring more human capital (skills and whatever makes what one can do more valuable to others) should result in a greater willingness to take care of your health. This is simple enough; living longer with intact abilities gives you more time to enjoy the extra income generated by the higher level of human capital. (Finding out that you are going to live longer should also increase your propensity to acquire more human capital, since the benefits will accrue for a longer time.)
So, people of indubitably high levels of human capital should, all else equal, live longer than people of indubitably low levels of human capital. In samples large enough that chance events may be neglected, successful musicians should live longer than do people on average.
What, then, could explain the findings of the BMJ article? A few tentative suggestions:
1. The musicians sampled in the BMJ article are not necessarily the ones who wrote the songs. Writing songs and performing them are distinct skills and it may be that singers, even singers of hits, simply are not generally that important to a song's becoming a hit, so their human capital is not as valuable as one might first think. Maybe the people with really valuable skills are the song writers and maybe (I don't know) they actually do live longer?
2. Some of the commercial success of singers and bands may be due to their leading a dangerous existence, so a number-one album might not happen if the hard-core rock-'n'-roll image is absent. In other words, perhaps a dangerous lifestyle complements musical ability for commercial success?
3. For painters, it is often said that the value of their art increases greatly upon their demise. If art purchasers rationally anticipate the death of living painters, however, they should, ceteris paribus, be willing to pay more for paintings by artists who do not take good care of their health. I haven't seen any data on this, but is it possible that something similar is going on with musicians? The prices might conceivably be higher for tickets to concerts ("last chance to catch a live performance") and for certain memorabilia. Yet, music and paintings are rather different art forms and if this is a significant incentive for one group of artists, it seems it should be painters rather than musicians.
4. I suppose it is also possible that some sort of culture exists in (parts of) the music business, such that there are social pressures to engage in risky activities. If successful musicians (of vast human capital) associate with terrible colleagues who will never amount to anything (negligible human capital), the latter, having less of an incentive to live long lives, could exert pressure on the former to drink, do drugs and smoke and other such activities which put good health at significant risk. But why should successful musicians have such strong ties with unsuccessful musicians? It would make more sense if they associated mostly with other high-quality performers and in such circles bad habits cease to exist, not the people constituting the circle.
I doubt the validity of 3 and 4. Maybe 1 and 2 are worth exploring, though I am not really happy with them either. Either I am deficient in insight, or this is tricky stuff.