Income Lost from Snow Days*
by Guest Contributor Stas Kolenikov
“I’d like to thank my friend and colleague Dr. Stas Kolenikov for contributing to this month’s guest blog. Dr. Kolenikov is a Senior Associate at Abt Associates Inc. He publishs papers in academic journals related to economics, psychology, sociology, and survey methodology. In other walks of life, he bikes the roads and hikes the trails near Columbia, MO, where he lives with his family, and enjoys the company of his wife sharing ballroom dancing and acro yoga with her.”
We have hit that time of the year where we hope to have moved past the snow storms but the memories remain. How many times did you look at one inch of snow on your driveway, and wonder “Why did the school district have to close?” You then also wonder if parents who have to scramble for child care arrangements or simply miss work could just collectively bribe the local schools. Heck, it feels like we could just buy them a couple of those special trucks with the snow shovels and gift them to the school district, under a strict promise they don’t cancel anymore because of snow!
This leads to an interesting question: what does a snow day really cost? In this original analysis, I am using data from publicly available sources to try and do just that.
I have an extract from the Current Population Survey Annual Social and Economic Supplement (ASEC), known in the social scientist jargon as the “March Supplement” for the month it is administered. To be more specific, I am using a data set downloaded from CPS-IPUMS website; IPUMS stands for “Integrated Public Use Microdata Series,” and it is a data service operated by the University of Minnesota-Twin Cities Population Center that provides neatly formatted data made available from the US Census Bureau. Here I present a step-by-step look at how to use this type of data to try to answer important questions of interest. I’ve done this analysis in Stata, and if you are a Stata user, you can reproduce these same results for yourself by downloading the data from the website above and following the links to my code that you see in purple.
The version that I have is the March 2015 ASEC supplement data for the state of Massachusetts. It has 3,253 observations nested in 1,248 households. You can use this code to describe the data. We can start by identifying families with a single working parent or with two working parents. These are the families that are the hardest hit by the school snow day closings; families with a parent who stays home may be less affected. How do we identify these families? There are two components to that: identifying the relations between adults in the household, and identifying the labor force status of these adults. What we want are households with school age children with no siblings old enough to serve as babysitters. Beyond that, in households with more than one person, we will assume that the person with the lowest identified household income will be the one who stays home with the children. There are many assumptions that we have to make on the available data and I have a complete walkthrough with reproducible Stata code available here. As with any analysis, there are a slew of limitations and assumptions, which I discuss further here.
Summarizing the results, there 1,785 observations from adults with school age children; of those we have 141 single-parent observations and 611 observations from families with two working adults. We have to use survey weighting to get population level estimates, but after reworking the income data to account for individuals who are not working, I find that these families typically lose, on average, 35.9% of their income on a given snow day. Since the number of such families varies from location to location, I estimate that a snow day costs every adult in an affected region an average of $10.33 each. So in a city with population of 100,000, the scale of a typical college town like the one I live in, the parents will lose a total of 743 thousand dollars per snow day.
How many snow trucks can you buy with that??
Pingback: The COVID Denominator - JPHMP Direct
Pingback: The Opioid Data Crisis by Jason Brinkley -- JPHMP Direct