With the government shutdown in its second week, everybody is offering opinions on whom to blame, what the costs will be, and what policy concessions it will take to end the shutdown.
Today we consider a different angle—do the stats on past government closures tell us anything interesting about the current machinations in Washington. Does the math have any predictive power as to how long this drama will last? Your guess is as good as ours, but here’s what the numbers say.
Since this revised budgeting process began in 1976, we saw 17 previous shutdowns. This chart shows us the length of each and the year it took place. The longest was the December 1995—January 1996 shutdown that lasted 21 days.
We can go further than that. We can deploy a lot more data to build a much more detailed story. Consider that each shutdown had a different composition of government: the party controlling the presidency, house, and senate can vary. Are Congress and the White House run by the same party, or not? We can put all this together into a much denser chart:
This shows us in one place, the party of the President, the House, and if those two parties matched. Fun fact: we ended up ignoring the makeup of the Senate in this analysis because after going through all the data, we found that the party of the House, rather than the Senate, was a much more significant factor in affecting shutdown length.
We can take all this data to calculate which factors matter the most to predicting the length of a shutdown. The simplest, most efficient formula we found to explain the length is:
Shutdown Length = 1.9 – 1.8 * HP-same + 11.1 * Dem-Pres
Breaking down the equation, here’s what we see:
• Start with 1.9 days as the typical length of a shutdown, no matter what.
• Subtract 1.8 days if the house majority and president are from the same party
• Add 11.1 extra days if the president is Democrat.
That’s right: The single most important factor in predicting how long a shutdown will last is if the president is a Democrat. The stats say no other factor has a greater numerical effect.
You can see it in the chart: The blue shapes (Democratic presidents) reveal much longer shutdowns than the red shapes.
We evaluated several further factors that didn’t end up moving the needle:
• Party makeup of House and Senate (in percentage terms)
• If the house majority party matches the senate’s
• If the president’s party matches that of the senate majority
• Year of shutdown
• Number of prior shutdowns
• Length of previous shutdown
These factors are not significant. The only two that matter are the president’s party and whether the house majority party and president’s party matches up.
Based on this analysis, and given the current makeup of a Republican house and a Democratic president, the numbers suggest we should expect about 13 days of shutdown. (That’s 1.9 from the constant, plus 11.1 because the president is a Democrat.)
We know one additional thing to be true: Washington doesn’t operate with the same simplicity and logic as do basic mathematical equations.