Too many confounding variables. I think a likely explanation is that nondrinkers are likelier fairly unhealthy to begin with. Someone on a lot of medications for various health problems is likely to not be able to drink -- or be strongly urged not to -- due to potentially dangerous drug-drug interactions. I think socioeconomic status may be a confounding variable too. At least in the US, low SES is correlated with extreme patterns of alcohol use -- either reckless abuse or strict abstinence. It also correlates to a harder and more stressful life, regardless of whether one drinks alcohol. Someone who manages a lifetime of moderate alcohol use is likely to be moderate and health-conscious in most other aspects of their life as well, and in an upward spiral, is also more likely to be successful.
A causal role could only be established if you ran a cohort study, whereby you cajoled half the nondrinking group to start drinking halfway through the study, and half the drinkers into quitting, and looked at which of the 4 resulting groups lived longest. But this study will never be run, because it's highly unethical.
I have serious doubts, in any case, if most abstainers would increase their life expectancies by taking up alcohol, even in modest amounts. Abstaining from alcohol is, for most of them I'd guess, a weight-bearing pillar in their lives that keeps everything else in balance, just as much as drinking modest amounts of alcohol is for most drinkers. In a society where alcohol is an accepted part of the social life of 2/3 of the adult population, someone who never touches it likely has a very good reason.
