News & Events

Print/PDF this page:

Print Friendly and PDF

Share this page:

Against (Discretionary) Snow Days—A Commentary by Ian Ayres ’86

The following commentary was posted on newyorktimes.com on February 2, 2011.

Against (Discretionary) Snow Days
By Ian Ayres ’86

I awoke last week to another foot of snow adding a third blanket of winter to our city of Elms. I am reminded of the joy I felt as a child waiting to learn if school was canceled. Something has been lost in our age of instant information — who can forget huddling around the radio, holding your breath while the radio announcer lists the seemingly endless roster of closings? Today, we received the decidedly less romantic robo-calls at 5:33 in the morning.

Indeed, from this perspective, predictive analytics is going even further in killing some of the fun of anticipation. My daughter pointed me to a Snow Day Prediction Widget that will calculate today the probability that your school will be canceled tomorrow. If you input your Zip code, it will automatically pull the latest weather report from the National Weather Service and combine it with 6 other factors (including a 0-3 hype factor: 1 = “Many Kids are Talking About It”):

My first reaction to the calculator was mixed. I loved the pragmatic application, but I worried that the widget was based on a pseudo regression — with the weights being pulled from intuition and experience. I’m a “show me the regression” kind of guy.

But my misgivings evaporated when I tracked down the brains behind the site, David Sukhin. It turns out that David is just a junior in high school in New Jersey. “I made the formula,” David emailed me, “about five years ago largely based on intuition but tested it using empirical data I collected. I did not do a ‘formal’ regression but I did design the curves based on data I had and how I believed schools would react to weather conditions. About two years ago, I made the calculator automatically get weather information from the national weather service.”

Instead of cranking out a science fair project which (might teach you something but) provides no benefits for others, David has created a site that provides useful information that helps others plan their lives. By the way, on January 20th alone, his site had over 150,000 hits, and this winter more than 7,000 unique Zip codes have been entered. College admissions committees take note. This is just the kind of kid I’d want at my school.

Kill Joy Economics

One of David’s factors might also help test one of my more depressing hypotheses. Holding other factors constant, the widget will predict a lower probability of a snow day if your school has already been cancelled several times previously this winter. In other words, if it snows one inch, you’re less likely to have a snow day if your school has already been canceled four times than if your school has only been canceled one prior time. This factor is interesting to me because it might play a role in testing whether cancelling school for small amounts of snow is a worthwhile social policy.

I haven’t tested it yet but I think most snow days are ill-advised from the perspective of public health and safety. I think that most children are exposed to more miles of driving on a snow day than they would be if they went to school. In our house, we often go out to breakfast on snow days, and to the movies and to play dates. There is something close to an iron law that the more passenger miles driven, the more injuries and deaths. And when kids aren’t driving around, they are engaging in more dangerous outdoor activities.

From an economic perspective, snow days externalize risk. Discretionary snow days don’t reduce risk (I hypothesize), they just take the risk off the school districts’ books and shift it to the private parents. If I’m right, we should expect to see more injuries to kids the first time there is one inch of snow (and the probability of a snow day is high) than the 4th time in a winter when there is one inch of snow (and the probability of a snow day is lower).

But wait. It gets worse. Discretionary snow days make families scramble for child care. I’d bet this disproportionately hurts working families that are already hustling to make ends meet. And unplanned child care is probabilistically higher risk. In sharp contrast to my happy childhood memories of snow days, I’d predict that discretionary snow days expose some kids to risk of abuse, neglect and/or negligent care when they are dumped last-minute at their uncle Ned’s. (And don’t forget the miles driven on snowy roads to get them there.)

Is Dunkin’ Donuts closed?

Of course, some snows days are justified. But here’s a simple test: is the Dunkin’ Donuts still open for business? The answer is usually yes. If workers with modest means can still make it in to their jobs at Dunkin’ Donuts (or Wendy’s or McDonalds), then teachers should also be able to arrive at their places of employment. If conditions are safe enough for city buses to be running, we might ask why school buses can’t safely operate.

Of course, we might be leery about letting corporate greed dictate public policy. But if we really think that it’s unsafe for workers to get to McDonalds, then OSHA might want to step in and mandate snow days to protect employees on their way to work as well. Or at least create a defense against retaliatory firing of workers who refuse to come to work — saying it was too risky for my kids to go to school, so I figured it was too risky for me try to come to work. But let me be clear: the fast-food industry test for snow days seems a lot closer to my guess of what would maximize public health and safety than the school board test. And harmonizing the two tests would provide the extra benefit that many working families wouldn’t need to scramble for childcare because the parents wouldn’t have to work when the kids didn’t have school.

The Dunkin’ Donuts test also provides a clue to what it might mean for a school to be open when it snows. Sometimes a store will be open but with reduced staff and services. The same idea could be used for schools. It is not essential that all teachers get to school on time or that normal classes be held. Instead, we might split the difference by still announcing snow days for light snow, but giving families the option of dropping their kids off at a school which will provide a safe environment even if it can’t muster the personnel to fulfill its usual education mission. Snow day school could even be re-imagined as a day of creativity and fun. Occasional warehousing is fine.

Even if it isn’t safe for school buses to operate, parents might have the option of getting their kids to school. Many parents would exercise this option.

This rant is more evidence that economics is a dismal science. I take one of our nation’s most beloved institutions and hypothesize that discretionary snow days increase traffic injury and child abuse. Policymakers who care about social welfare need to weigh these (as yet unproven) costs against the joy that millions of children experience when they hear that school has been canceled. Speaking of which, I woke this morning to learn that my kids’ schools are closed today. Notwithstanding the above, I look forward to putting our health at risk by sledding down Yale Divinity School’s famous “suicide” hill.