In Larry's recent post on bankruptcy reform and time inconsistency, he linked to Todd Zywicki's and HedgeFundGuy's posts on the dubious statistic that 50% of bankruptcies are caused by medical expenses.

Both Zywicki and HedgeFundGuy raise solid objections to the 50% notion. However, I think they both miss the most obvious reason to be skeptical of the statistic--the tremendous variation in bankruptcy filing rates across states. According to a WSJ article on March 10 (subscription required), UT, TN, and GA have rates of one bankruptcy per 36.5, 38.7, and 42.4 households, respectively. By contrast, AK, VT, and MA had one filing per 171.2, 156.2, and 144.3 households. I think it implausible that there could be such a wide variation in bankruptcy rates if 50% of filings were caused by medical expenses. While medical expenses may vary across states for many reasons, I think it unlikely that the variation in medical expenses could lead some states to have bankruptcy rates 3-5 times greater than some other states.

But what about lack of health insurance? The uninsured rates for the six states are UT 14.3%, TN 12%, GA 17.6%, AK 20%, VT 12.2%, and MA 11.3%. (Source here.) Neither the pattern of uninsurance rates nor the range is consistent with the notion that uninsured medical expenses cause half of bankruptcies. A caveat--statistics on insurance coverage don't say anything about the generosity of the benefits; it's possible (though I'm skeptical) that policies in UT, TN, and GA are sufficiently stingy relative to policies in AK, VT, and MA to explain the difference in bankruptcy rates.