As Boston considers how to promote more late-night activity, the city shouldn’t play it safe — nor should it expect to get every policy right on the first try. The MBTA is once again testing out late-night service, Boston is piloting late-night food trucks, and Mayor Walsh’s Late Night Task Force is designing a trial program to allow some bars and restaurants to stay open after 2 a.m. These initiatives are delightful, but they raise a thornier question: Where should all this activity in the wee hours be allowed?
Fortunately, the task force’s work doesn’t just create an opportunity to make Boston livelier. It also creates an opening for a more systematic, data-driven approach to regulating city life. Walsh’s administration should resist the natural temptation — to minimize conflicts with neighbors by extending hours only in one or two non-residential areas, such as the Seaport District, where there is little likelihood of anything going wrong.
As a practical matter, designating a single area as the extended-hours district will also lead to alco-commuting, in which late-night revelers would travel after midnight to get to the city’s one late-night spot. Extra late-night travel seems like the last thing we want.
Beyond that, though, we don’t learn anything unless we try options with uncertain results. By choosing a single low-risk area, the most that we can ascertain is that late licensing can work out well under ideal conditions. We would learn little about how later hours would affect other areas outside the emerging Seaport District. Would that mean more crime, drunken driving, or neighborhood nuisances in older areas? Will better nightlife really make Boston more of a draw to young innovators? These are the things that we need to figure out, and to do that we need a rigorous experiment.
We already have some data on which neighborhoods would be most and least suited to later hours. Liquor licenses are sufficiently associated with reported crimes for us to take safety seriously, but not strongly enough for us to have much confidence about how much crime will rise if bars stay open later. The map above shows a plot of all non-airport liquor licenses and reported weekend street crimes, such as assault and robbery, between the hours of 12 a.m. and 3 a.m. between 2010 and 2014. (I obtained these data from Benjamin Weinryb Grohsgal, a member of a Kennedy School student team that supported the Late Night Task Force with mapping and quantitative analysis.) More liquor licenses can mean somewhat more crime, but there are some well-licensed areas, like the North End, where crime is relatively low, and there are other areas with few licenses and abundant crime.
Not every neighborhood needs to be considered for later hours. As the map shows, there are great swaths of residential Boston with extremely low levels of liquor licenses, and we may never want to lengthen hours in highly residential areas with no T access. So let’s knock them out of the pilot. One approach would be to draw an arc from Northeastern University to the Reserve Channel south of the Innovation District. All areas south of that line would be off limits, and residential areas north of the line that strongly object to later hours can also be excluded.
Within these constraints, we could conduct a randomized trial. The remaining area — the T-connected northern part of Boston — would then be subdivided into a checkerboard of mini-neighborhoods. The city could then identify 10 or so matched pairs of area with similar characteristics — and then flip a coin to decide which area gets longer bar hours over the summer. At least some of those locations need to have residential neighbors. By comparing the two groups and surveying local residents, we could figure out the impact of later licenses on reported crime and neighborhood satisfaction.
I know: Setting rules by flipping a coin sounds awful. Who likes the sound of randomized government? But blind adherence to the status quo is a terrible approach, and making change without solid data isn’t much better. Governments need to benefit from the same march of information that routinely gives us better computers and medicine. Doctors never prescribe drugs until they have gone through randomized trials. Governments should follow the same approach.
With a proper treatment and control sample, the city would be able to measure the impact of later nights on crime and neighborhood disruptions. Local residents can be surveyed on the Web while the experiment is ongoing. The Police Department could step up spot checks of late-night drivers to see whether there is an increase in drunken driving. Heat maps of Twitter can measure if late-night areas see an upsurge in tech-savvy night owls.
Conducting a policy experiment on this scale may seem difficult to reconcile with complex licensing rules. Yet the state Senate has already taken a step to alter the legal machinery by adopting, as an amendment to the 2015 budget, a provision that would eliminate the statewide ban on selling alcohol between 2 a.m. and 8 a.m. “in a city or town that is serviced by the Massachusetts Bay Transportation Authority’s late-night service.” Even if this change is accepted by the House, it will only empower the city to change the rules — not mandate any late-night service. Local licensing officials could presumably be selective, granting later hours in some areas and not others.
Many people prefer a greater level of predictability from government. Ideally, we’d have a government that doesn’t make mistakes and always knows what it’s doing. But recognizing the limits of our current knowledge is the first step towards wisdom.
Smart entrepreneurs don’t expect new products to just pop out of their brains fully formed, like Athena from the head of Zeus. New ideas need testing and tweaking. This is even truer in government than business, because everyone suffers if public policies fail. Ultimately, every policy change is an experiment of sorts — so we might as well design the experiment right.