The Minimum Legal Drinking Age (MLDA) is one of the most well-studied behavior-control policies in the US. Prior to the 1970s, states set their own MLDA and while most states set it at 21, several states had a lower MLDA (usually 18 or 19). When the federal voting age was lowered to 18 in 1972, a number of states began lowering the MLDA to 18 as well. Thereafter, a number of states that had lowered the MLDA began to raise it back up due to problems (or at least the perception of problems)--primarily, increased traffic accidents and fatalities caused by drunk teenage drivers. In 1984 the Federal government entered the picture by making certain Federal highway funding contingent on states raising their MLDA to 21. By 1988, all states had a MLDA of 21 (where it remains today).
The existing variation in MLDA coupled with the lowering and subsequent raising of MLDA created a large number of natural experiments on the effects of different MLDAs and teenage drinking. Dozens of studies have been produced and, at least among the higher-quality studies, concensus seems to have emerged on two points:
- MLDA is inversely related to alcohol consumption (lower MLDA => more alcohol consumed), and
- MLDA is inversely related to traffic accidents (lower MLDA => more traffic accidents).
Based on this, I would have to conclude that a lower MLDA fosters a less responsible use of alcohol (at least in terms of not moderating consumption nor refraining from driving when drinking).
Nonetheless, I'm not convinced that a higher MLDA is justified. First, because I don't support a nanny state. Second, any benefits that can be attributed to raising the MLDA can be extended virtually up to any age until you end up with prohibition (which I don't support either).