I wonder what good money paid for the study?
Of course the early adopter states to 21 would see a reduction, as the 18-20 year olds went into neighboring states where they could drink legally. Once every state was forced to adopt then rates evened out again.
Studies from archival data of that sort are essentially free: some professor got a publication by doing the study at the behest of his or her university president, and as a result got a slightly higher raise than he or she would have gotten without the study, and that, and some paper and computer time was the entire cost of the study.