Bankruptcy rates vary alot across states. With a fairly simple statistical model, Lars Lefgren and I explain about 70% of these differences in a paper just published in the Journal of Law and Economics. For cross sectional work using survey data, where you are looking across states at a point in time, explaining 70% is pretty darn impressive.
The leading indicators (and likely causes) are:
1. Wage garnishment laws. Allowing creditors to garnish wages is a very strong predictor of your bankruptcy rate, because once the garnishing starts, the debtor heads to the bankruptcy lawyer. If your state makes garnishing hard or effectively impossible, then the debtor still doesn’t pay his debts, but he need not file a bankruptcy. If you wanted to know the one thing to best predict a state’s bankruptcy rate, it would be their wage garnishment laws.
2. Chapter choice. One can file a chapter 7 bankruptcy and be done with it, or one can file a chapter 13 bankruptcy that lets you keep your assets but requires you to pay back some of your debts over the next several years. People filing bankruptcy, though, tend to have trouble completing these plans, and so 70% of chapter 13’s are not completed. This can send the filer back to court for a _new_ bankruptcy filing — meaning that the same debt creates multiple bankruptcy filings. This then drives up the reported number of “bankruptcies”. Some states’ lawyers and courts funnel debtors into 13’s. Some encourage 7’s. This creates variation in the number of bankruptcy filings.
3. Demographics. Bankruptcy filers tend to be low to middle income and in their late twenties. This is enough time to get some consumer debt and get in trouble, but before you have substantial assets that you would want to protect from a bankruptcy. Student loans, as I recall, are not generally dismissable as part of bankruptcy. Also, family structure matters– like being a single mom or divorced with kids.
The abstract is here , the paper is freely available from the JLE here , or you can read the press release. All in all, the takeaway message is that bankruptcy rate differences across states are more policy noise than a revelation about credit abuse among different state populations. Low bankruptcy rates do not necessarily imply that people are paying back their debts any more than those in high bankruptcy states. In fact, the wage garnishment evidence would suggest the opposite.
More broadly, differences across states in some social outcome may not be telling you what you think they are.
“differences across states in some social outcome may not be telling you what you think they are”
What do we think they are?
I haven’t read the whole thing yet, but when I got to the equations, my mind was drawn back to the last encounter I had with the term “Garn”–this passage from Wikipedia:
This is severely hampering my ability follow your reasoning. :-)
Generally speaking, differences across states (or countries, or cities…) may not reflect people differences, but rather policy differences. Or vice versa. Or something else entirely. The reader should beware of easy inference.
Equations do that to me too (unless I wrote them).
Frank M, thanks for your continuing great work. Economists are the heroes in making sense of many societal trends these days, and they do it with facts and figures rather than emotional mumbo jumbo.
As an economist, I dispute your claim that I am not driven primarily by emotional mumbo jumbo. The true skill of economists lies in _hiding_ the emotional mumbo jumbo in mathematical mumbo jumbo.
Frank, I’m no economist but I’m pretty sure your #3 is a total non-reply. Your phrase, “differences across states in some social outcome may not be telling you what you think they are,” seems to imply that you think we’re thinking something erroneous. Please clarify your assumptions about what we’re thinking.
BTW, if it helps, I’m not being sarcastic and my question’s not a trap. I just want to understand what misconceptions you think you’re clearing up.
OTOH, if you’re just saying, “be careful with jumping to conclusions about social sciences, because the data can be trickier than we think,” then s’cool.
Four in a row!
Steve, since it is a general point, it is not clearing up any particular misconceptions (hence the “more broadly” at the front of the sentence that you clipped off). Rather I am pointing out a general methodological misconception people make with data. Here are two examples of this _general_ point:
1. This study. People often informally attribute differences in bankruptcy across states to differences in people’s willingness to repay creditors or their financial management skills in those states. This is probably a bad idea, for the reasons cited in the paper.
2. From the other direction, people often think that a school is “good” (i.e., high test scores) because of good teaching or school polices. But much of the difference in test scores (way more than half) appears to actually be driven by demographics, not policy differences. Thus here, it is the people, not the policies.
Does that help?
I was saying #9.
Let me distance myself from Scott’s claim. I, at least, lack the emotional range to be driven by “emotional mumbo jumbo”. I have maybe 3 emotions and that is only if you count “sardonic” as an emotion.
Alas, I cannot speak for economists generally one way or another.
Frank & Geoff,
I apologize. Frank is right. I thought we were talking about emoticons. I have lots of those.
Thanks for the reminder about being skeptical of statistical comparisons of states that try to draw broad conclusions from variations involving many factors. Another example was the “study” by a mental health professionals association that announced that Utah was the “most depressed state”. A close look at the numbers showed that the variation from state to state was minor, and the several measures that went into the “average” were mostly subjective. The whole point of that “study” was to promote government funding for mental health programs that employ members of the associaiton that put out the press release. It didn’t tell us anything useful about how to have better mental health.
It is clear that a lot of the “information” being thrown around to justify a massive overhaul of the health services industry in America is similar meaningless baloney.
Then there is the saga of Global Warming. Back in 1997, the advocates of the Kyoto Protocol (which did not include the Clinton-Gore Administration–they never sent the treaty to the Senate for ratification) trumpeted the rising trend in global “average temperature” over the previous 20 years. However, since 1998, the global average temperature has been level or falling, especially coming down the last two years. You NEVER see the temperature chart for the last 10 years reported in the mainstream news media. Instead we get pictures of glaciers, that have retreated over the course of the 20th Century–including the 40 years from 1935 to 1975 when global temperatures were FALLING.
In light of the persistence in the public mind of so much consummate yet obvious baloney, the persistence of misinformation about the Church and its members, despite our best efforts to make the truth known, should not surprise us. Critics of Mormonism passionately put out the misinformation, and most of the rest of mankind doesn’t care enough to bother questioning it.
Steve, I think the background subtext is “Utah has a high bankruptcy rate because Mormons are dishonest.”
JG, now we’re talking.
Fortunately, Steve, we can now move on to “Utah has a high bankruptcy rate because of policy choices made by legislators WHO ARE MINDLESS ROBOTS UNDER THE CONTROL OF THE MORMON CHURCH!!!”
Comments 16 and 18 are good examples of emotional mumbo-jumbo (don’t you just love that word?). As an economist, Frank M is so incapable of emotional mumbo-jumbo that he could not even bring himself to acknowledge where Steve was heading.
“You NEVER see the temperature chart for the last 10 years reported in the mainstream news media.”
I have been told that in many states (Utah, for example) the interpretation of the chapter 13 income rules is so strict that it is very difficult for most filers to actually complete the bankruptcy plan, usually leading to a subsequent chapter 7 filing.
The federal rules and regulations that create such an effective impossibility don’t sound very well written to me. They do not appear to be in the best interest of either the debtor or the creditor. The IRS is much worse.
Bill, that is not a graph of the last ten years and it is ridiculously exaggerated for effect (“basis points” of 0.01 degrees). Worse it is a graph of land based surface temperature measurements, which have notorious quality problems due to the heat island effect and more serious sensor placement and data collection issues.
If you want a more realistic graph look at satellite based lower tropospheric temperature measurements over the past thirty years. Temperatures for the last ten years have been flat or declining. Please also note that temperatures declined for a thirty year period from 1940 to 1970 despite rapidly increasing CO2 emissions during the period.
Or another background subtext he might addressing is “Utah has a high bankruptcy rate because UTAH Mormons are dishonest.” Let’s see, we’ve got plastic surgery-loving, Zoloft pill-popping, AND now credit card-abusing. Those darn Utah Mormons!
FYI, I ran a county level regression on Utah just for curiosity’s sake while I was doing this paper. The more Mormons in the county, the fewer bankruptcies.
Check here for your concerns:
Frank, re: #24, so does Mormonism have any explanatory effect on the bankruptcy rate or not? Supposing that your paper did successfully identify institutional/legal factors that can explain the bankruptcy rate, does your comment mean that, in addition to these, Mormonism may in fact suppress the bankruptcy rate? If high Mormonism in a given county correlates with less urbanism, might other institutional factors (fewer bankruptcy lawyers per capita, lower home value to income ratios) have more explanatory effect than mere “Mormonism”?
“Let’s see, we’ve got plastic surgery-loving, Zoloft pill-popping, AND now credit card-abusing.”
Don’t forget that they loves themselfs some porn too!
Bill, That is one of the better sites I have seen on the subject. Thanks for listing it. The “Four Series” temperature graph on the page you refer to shows flat temperatures for the last ten years, completely disregarding the 1998 anomaly. This is wildly off topic though, so perhaps another time.
Hmm, Jimbob. Now that you say it, actually plastic surgery, pills, credit cards and p*rn doesn’t sound all that bad. I’m movin’ to Utah!
All this kidding aside, I applaud Frank’s efforts to set this record straight. It’s an important issue. When I was in private practice, it seems I was forever trying to explain the disconnect between a person’s bankruptcy filing and their supposed moral status.
The regressions in question control for things like income and urbanization, so i don’t think what we are picking up as Mormonism are those effects. Frankly I am suspicious that being Mormon is very strongly related to bankruptcy rates. But if it is related, my best guess based on what I’ve seen is that being an active Mormon would decrease the probability of credit default (both informal and through bankruptcy).
Frank, this is awesome! It’s rare enough that something argued in the Bloggernacle gets illuminated by actual published research. And here you’ve gone and published your own research to answer an issue that’s been raised. This is a pretty high standard for the rest of us to live up to, so I hope you don’t mind if we just continue to argue without evidence. :)
Exemption rates, size of public safety nets, and payday loan regulations contribute virtually nothing to the cross-state variance in filing rates.
So dispensing with welfare will have no negative effect? Let’s do it!
Outstanding work, Frank.
Congrats on the publication! I’ve been citing this paper in bankruptcy and Mormonism discussions since we read it in Public Econ. It’ll be nice to have an easier way to cite it “than this cool study I once read in school.”
A quick question if you don’t mind. Where did you get county level religious concentration data? Does the census track that (I noticed you cite in the paper them for ZCTA demographic data)? I can never find anything on religion in the public data.
I honestly can’t remember, but I don’t think I tried too hard (as it was just a side question for personal interest, rather than part of the main study). I probably found some dataset through ICPSR (The Michigan data warehouse). Certainly not the census.
If you really need to know, I might be able to hunt through old files and figure out where I got it.
Today the AP has printed results from their study showing a relationship between bankruptcies and garnishment laws. Was that a parallel effort, or related to your study?
The press has a hard time getting economics right, but this time they may have stumbled onto valid conclusion.
I spoke with a guy from the AP a little while ago who wanted some advice with a bankruptcy project and how to do it. He knew about our paper, so it’s possible they only did their work after seeing ours. Which would not be surprising, as our paper has been circulating for a couple years. It is sort of funny that it came out right now, a week after ours got published.
It is also quite possible that they started working on it independently, and then saw our paper later. Once you see the connection, it’s pretty obvious.
The AP should have cited your paper, especially since they knew about it in advance.
The AP is showing a trend of becoming a self-interested maker of news, rather than a disseminator of news – more editorials, etc. I think that jeopardizes their credibility, as they are competing with their subscribers.
Thanks, Clair. To be extra fair, I don’t know if the guy I talked to was working on this bankruptcy project or a different one. Thus it is possible that the author of the article you cited didn’t know about our paper.