Forecasting U.S. bankruptcy filings for this year was a little more complicated. In a comment to my post about the total 2015 bankruptcy filings, Erich Fabricius made the astute observation that December 2016 saw the introduction of new bankruptcy forms and that could explain my befuddlement at the abnormally large 14.8% decline for December in terms of year-over-year daily filing rates. November, in contrast, saw almost no decline in the year-over-year rate, which is also unusual. The relatively stronger numbers in November suggests that attorneys were perhaps trying to beat the deadline before the new forms went into effect. The effect would not have to be huge -- shifting 5,000 filings from December into November would have been enough to create this effect.
What that means is my preferred mathematical model to predict bankruptcy filings for the next year has to start with the immediately previous two months being untypical months. If I run the model with the actual data, I get 800,000 filings. If I "correct" the numbers for November and December to what would have been expected had recent monthly trends continued, the model predicts 782,000 filings. The model controls for the amount of outstanding consumer credit and national personal income and has proven accurate in the past.