Over the weekend, I wrote about Canadian policy experiments that could reduce alcohol consumption by setting a minimum price for beer, wine and liquor. Wonkblog reader and Cornell University economist Brad Rickard passed along new research he’s done on the economics of alcohol consumption, which suggests another, albeit counter-intuitive, approach that may curb negative impacts of excessive alcohol consumption: make wine cheaper and more available.

In a working paper for the American Association of Wine Economists, Rickard and his team start off by looking at the states that allow grocery stores to sell wine, versus those that limit such sales to liquor stores. The increased competition of grocery stores selling wine, unsurprisingly, correlates with both lower wine prices and higher rates of wine consumption.

There’s a surprising, public health benefit that grows out of that: States where wine makes up a larger part of total alcohol consumption tend to have lower rates of traffic fatalities. Here’s what that looks like in chart form:

“Overall, these results indicate that an increase in beer and spirit consumption, as a share of total alcohol consumption, increases traffic fatalities,” Rickard and his team write. “Whereas an increase in wine consumption as a share of total alcohol consumption decreases traffic fatalities.”

This makes sense in the context of previous research: Wine is the least common alcohol consumed in binge drinking, accounting for about 10 percent of binge-drinking episodes in a 2007 Journal of Preventive Medicine study. Beer, in contrast, accounts for two-thirds of binge-drinking episodes.

The research does not suggest that state governments should start handing out bottles of wine as a means of reducing traffic fatalities: An increase in the aggregate amount of alcohol consumption correlates with more traffic fatalities. But Rickard’s work does suggest that what matters isn’t just how much we drink, but also what we drink — and there, state policies that effect the availability of wine can have an impact.