One More Time, With Feeling

A Credit Slips reader pointed me to an article in the Atlanta Journal-Constitution pondering why the bankruptcy rate is falling. The piece is filled with quotes about the relevance of the economy and the cost of filing bankruptcy. Most of it is wrong. For example, it is right that the cost of filing has increased since the 2005 changes to the bankruptcy law, but there is no  evidence the cost has risen in the last year. Thus, the rising cost of filing bankruptcy helps to explain why bankruptcy rates have declined relative to pre-2005 levels but not why they have declined since last year.

Regular readers will know a piece like this just pushes my buttons. Outstanding consumer credit has the strongest statistical link to the short-term ups and downs of the bankruptcy filing rate. The relationship is counter-intuitive and paradoxical. As consumer credit rises, banrkuptcy rates tend to fall in the short term. As people borrow to stave off the day of reckoning, they postpone bankruptcy. When consumer credit tightens, people are less able to borrow to satisify their current needs and, as they run out of options, are more likely to end up in a bankruptcy lawyer's office. When it comes to the economy, the bankruptcy filing rate tells us very little about the overall health of the economy. The strongest reason why bankruptcy filing rates have eased slightly is that consumer credit has become slightly more available, according to the Federal Reserve's latest release.

A previous blog post discussed at length the link between consumer credit and bankruptcy filing rates. That post featured a monthly trend line for the past few years. The relationship is long-standing, however. The graph to the right shows how bankruptcy rates tend to move with  consumer credit since 1960. (Cllicking on the graph will open a larger image, and more detail about the calculations appear at the bottom of the post.) The consumer credit axis is inverted (such that negative numbers are at the top) to make the relationship easier to see. In addtion to the "ocular regression" of just eyeballing the charts in this post and the previous post, more rigorous stastical testing verifies the relationship. (Lawless, Robert M. (2007). "The Paradox of Consumer Credit," University of Illinois Law Review, 2007:347-74.) In making a forecast of this year's bankruptcy filing rate, I relied heavily on the trend in the available of consumer credit, and the forecast has been pretty close to spot on. And, in a self-congratulatory reference (humor me by pretending the rest of this paragraph is otherwise), it is not as if these points have not been made in the New York Times instead of just this small corner of the blogosphere.

This story on bankruptcy filing rates came out just as I was preparing for our Empirical Methods course. In the first week, my co-teacher, Jen Robbennolt, and I discuss how we all like to find patterns that don't really exist. Bankruptcy attorneys might see people come into their office who are unemployed and come to the conclusion that unemployment drives the bankruptcy rate. During economic downturns, the topic of bankruptcy becomes more salient, and we are more likely to remember stories about bankruptcy. Thus, we tend to associate higher bankruptcy rates with the economic downturn.

There is at least one way, however, that all of my data analysis is wrong. The data suggests there is a tendency for consumer credit to affect bankruptcy rates. Technically, all the data suggests is that there is a relationship--causality could be moving in the other direction although I think that is unlikely. The data describes the tendency--the average effect--and by so doing fails to capture any individual case, which is both the strength and weakness of the analysis. Also, the analysis does not explain all of the variation in bankruptcy filing rates. Unemployment, economiic downturns, and other factors such as foreclosures undoubtedly play a role, but it is just that their role pales in comparison to the role of the outstanding amount of consumer credit.

Notes on the graph: The graph shows the percentage change from year-to-year for total consumer credit outstanding and total bankruptcy filings. The right axis for consumer credit is inverted to allow a clearer understanding of the inverse relationship between the two data series. Total consumer credit, which includes both revolving and nonrevolving credit, is taken from the Federal Reserve Statistical Release G.19 on consumer credit. The bankruptcy filing data are from the Administrative Office for U.S. Courts. Because of limits on the availability of old bankruptcy filing data, all data are for the twelth months ending June 30 for the year shown or as of June 30 for the consumer credit changes.