There is only an abstract available, so I can't be entirely sure, but this paper on the economics of drinking argues that

drug consumption, most commonly alcohol drinking, can be a technology to give up some control over one's actions and words. It can be employed by trustworthy players to reveal their type. Similarly alcohol can function as a "social lubricant" and faciliate type revelation in conversations. It is shown that both separating and pooling equilibria can exist; as opposed to the classic results in the literature, a pooling equilibrium is still informative. Drugs which allow a gradual loss of control by appropriate doses and for which moderate consumption is not addictive are particularly suitable because the consumption can be easily observed and reciprocated and is unlikely to occur out of the social context. There is a trade-off between the efficiency gains due to the signaling effect and the loss of productivity associated with intoxication.

So it's a good thing, right? That's what I'm going to take away from it. Thanks, Science! [Via]