Posted earlier in the week on possible errors in bankruptcy statistics because of the way software was designed. I just stumbled on a reference in an economics book, "Does Atlas Shrug", a compilation of articles on the effects of taxing the rich, that may also be pertinent. In 1986 the Bill Bradley/Rostenkowski Tax Reform Act passed (with minor assistance from what's his face, the guy at 1600 PA). In the introduction to the book, the editor says there's research showing that entrepreneurs switched the form of their business as a result of the changes in tax treatment. Apparently the big switch was from corporation C (never heard of it, I assume it's just the straight corporation form) to corporation S (which I have heard, where the business is a corporation but for tax purposes all the income is attributed to the individual). Note that I don't know for sure that such changes might affect bankruptcy comparisons, but it seems possible and 1986 is the date mentioned in the previous article.
It raises a bigger subject--the idea that there's a discrepancy between what is presented by lawyers and accountants under the tax laws and what the economic reality is. That's a discrepancy my old agency had a lot of experience with--farmers would change their organizations so as to limit the impact of payment limitation laws and regulations, but the economic reality wouldn't change.
The bottom line is that statistics can be tricky things.