I was going to cite job loss as a major cause of bankruptcy in the previous post, which is the conventional wisdom. But this paper argues it isn't true:
This paper utilizes the population of personal bankruptcy filings in the state of Delaware during 2003 and finds that household expenditures on durable consumptions, such as houses and automobiles, contribute significantly to personal bankruptcy. Adverse medical conditions also lead to personal bankruptcy filings, but other adverse events such as divorce and unemployment have marginal effects. Over-consumption makes households financially over-stretched and more susceptible to adverse events, which reconcile the strategic filing and adverse event explanations.
According to Zhu, having a serious medical condition makes you 50% more likely to file for bankruptcy, but not because of medical bills; medical bills are only a very small percentage of the overall debt of bankrupts, and are not significantly correlated with higher credit card debt, which one would expect if people were keeping down their medical bills by charging them to Visa. Presumably it's the income effect of disability or caretaking responsibilities.
Job loss may precipitate bankruptcy, but bankrupts don't report being laid off at a significantly higher rate than the control group. The difference is, the control group had savings to cover its financial emergency.
The paper seems to have covered most of the ways I initially suspected it had gone wrong; for example, I thought they might have missed people who had had an adverse income event like being forced into a lower-paying job, but length of job tenure was actually higher for bankrupts. I still have the lingering suspicion that it overstates its case, but it seems pretty robust--unlike the more widely quoted Warren study, which had to use a tenuous definition of causation to make its sensational claim that 50% of bankruptcies were due to a medical event--which turned into the even more sensational claim that 50% of bankruptcies were due to high medical bills in the hands of innumerate activists and journalists.
It's also worth noting that it's harder to save on $25,000 a year than $75,000, and bankrupts as a group tend to be poorer, which means they have little shield from adverse events. On the other hand, the bankrupts were consuming at levels comparable to the wealthy controls. Spending as much money as those who are much richer than you is pretty much definitionally a recipe for disaster.