Good IT Design and Bankruptcy

Interesting article in the NYTimes (registration required) today, discussing bankruptcy. The claim it reports is that the new bankruptcy law was based on an apparent decline in the number of business bankruptcies, from about 20 percent to about 2, with a corresponding increase in individual bankruptcies. This led to the conclusion that individuals were abusing their credit and the bankruptcy system, therefore the law was tightened.

Now a Harvard study claims to have traced the change in statistics back to a software error:

"The Kauffman study discovered why the government had undercounted entrepreneurs to such a large degree: a computing error. In the mid-1980's, the bankruptcy court system's records began to be computerized, and the most common software programs defaulted to individual settings. The error often went uncorrected.

'The model that Congress used for writing the bankruptcy law - a growing number of consumers and a rapidly shrinking number of entrepreneurs and small-business owners - was simply wrong,' said Professor Warren. 'It's bad policy on lots of levels.'"

A caution--someone on Volokh, I think Zyblocki, thinks professor Warren does bad studies, so this isn't gospel. But taking it as written, a couple comments:

1. I wouldn't consider this an error. Good user design will almost always default to the most common choice, in this case presumably "individual" instead of "sole proprietorship". Defaults save time and aggravation. (Of course, in 1986 the data entry was probably through a green screen terminal. If so, doing defaults may have been more difficult than it is now with graphical (Windows-like) screens.
   
2. The key issue here is whether the distinction is only for statistical purposes, or whether it is real--does the difference make a difference in how a bankruptcy is processed? If it doesn't, then the person doing the entry won't care and you'd have to use a drop down box with "pick one" showing. If it does make a difference, then ideally the software does something that depends on the right choice--prints out forms showing: "Bill Harshaw, Sole Proprietor of FacelessBureaucrat" rather than "Bill Harshaw, Individual".
   

This is a reminder that statistics can and do lie, particularly when they don't evolve out of the business process but are a separate operation. We should also remember that the bankruptcy statistics before the process was automated may have been defective. I'm also curious whether the bankruptcy process was automated once nationally, or district court by district court.