Or, put another way, marijuana is not a gateway drug but a replacement drug that — when properly regulated — could, on measure, lead to a net improvement in public health.

Most marijuana is consumed by people with a high school degree or less, and therefore relatively little earning power; that means they are significantly more responsive to the price drops that usually accompany decriminalization. That means a broad shift toward legalization is very likely to lead to a significant increase in pot consumption.

But according to the academic blog, JSTOR Daily

Even if smoking increases, there are intriguing suggestions that this means drinking will decline. That change alone could have a huge impact on public health. “The social costs of alcohol are pretty darn high,” said D. Mark Anderson, an economist at Montana State University who has conducted several studies on public health and marijuana policy. His own research suggests that alcohol and marijuana are substitutes for each other, not complements: in other words, if marijuana becomes easier to acquire, people may use it instead of alcohol, not on top of it. “If the substitution effect is large, then increasing marijuana consumption could really provide public health benefits,” Anderson said in an interview.

Though some lab experiments have suggested that people under the influence of THC, marijuana’s psychoactive component, are likelier than drunk people to overestimate the degree to which they are impaired and take more precautionary measures, such as driving more slowly, Anderson is careful to say that this research does not imply that driving under the influence of marijuana is necessarily safer than drunk driving. But even if “high driving” is just as dangerous as drunk driving, it could simply be the case that people using marijuana are likelier to stay home and smoke, instead of driving from bar to bar….In a study published last year in the Journal of Law and Economics, Anderson and two co-authors found that in the 19 states with medical marijuana laws, the first full year after the laws went into effects was associated with a drop in traffic fatalities between 8 and 11 percent. The estimated effects of the laws were largest at night and on weekends, when most drinking occurs.

Then there’s the fact that alcohol abuse …. causes between three and four times as much dependence as marijuana, but more than 10 times the violence and crime.

This research comes with important caveats — most notably, we we won’t know the full effects of legalization until decades in the future. Substance use and abuse patterns tend to be established in the teen years; the fear is not that the entire population will start smoking daily all at once, but that more young people will establish pot habits that will stick with them for a lifetime. “We’re not going to find out if legalization will lead to heavier use by 50-year-olds until today’s 15-year-olds are 50-year-olds,” says Jonathan P. Caulkins, a professor of Operations Research and Public Policy at Carnegie Mellon University, who has written frequently about marijuana and public health.

“If a million more kids and young adults using marijuana continuously — four times a day every day — for several years knocks down learning and labor productivity, those are outcomes that by and large the [National Institutes of Health] doesn’t focus on,” he said.

For his part, Anderson is optimistic. “It certainly doesn’t sound like there’s been an apocalypse and massive catastrophe in Washington and Colorado. And we may not necessarily need 40 years to tell.”