Andrew Leigh has pointed me to a recent study of US bankruptcy (paywalled, but the abstract is over the page). which concludes that the increased variability of income, and exposure to expense shocks such as medical expenses are not important factors in explaining the dramatic increase in bankruptcy rates since 1970. (I’ve seen a blog link to this also, but can’t find it now).
Count me as unconvinced. The main reason for rejecting income shocks is an explanation of bankruptcy is that, in the model of the paper, households should respond to increasing variance of permanent shocks by increasing precautionary savings. This appears to impute to households a much higher level of ex ante information about future income shocks than they actually possess, and also to rely critically on strong assumptions about rational planning. The whole credit card business is centred on the fact that lots of people (about half the population) don’t pay their monthly balances down to zero and therefore carry semi-permanent debt at very high interest rates. It’s hard to imagine that people who have trouble managing their credit cards are computing, in advance, the income risk they face and making precautionary savings to offset this.
That’s not to discount the importance of the ‘supply side’, in terms of easier access to credit, which has assisted people in managing increasingly risky income and expenses, at the cost of steadily increasing debt-income ratios. But you have to look at both sides of the story, and this paper rules out one side by assumption.
Accounting for the Rise in Consumer Bankruptcies by Igor Livshits, James
MacGee, Michele Tertilt – #13363 (EFG)
Abstract:
Personal bankruptcies in the United States have increased dramatically,
rising from 1.4 per thousand working age population in 1970 to 8.5 in 2002.
We use a heterogeneous agent life-cycle model with competitive financial
intermediaries who can observe households’
earnings, age and current asset holdings to evaluate several commonly
offered explanations. We find that increased uncertainty (income shocks,
expense uncertainty) cannot quantitatively account for the rise in
bankruptcies. Instead, the rise in filings appears to mainly reflect
changes in the credit market environment. We find that credit market
innovations which cause a decrease in the transactions cost of lending and a
decline in the cost of bankruptcy can largely accounting for the rise in
consumer bankruptcy. We also argue that the abolition of usury laws and
other legal changes are unimportant.
{ 22 comments }
Mrs Tilton 09.18.07 at 7:59 am
I’ve seen a blog link to this also, but can’t find it now
Here, I think (pdf warning).
Andrew Leigh 09.18.07 at 11:04 am
The New York Times paywall is now being lifted. I’m giving credit to your statistical analysis of October 2005.
Slocum 09.18.07 at 11:30 am
Count me as unconvinced.
Do you really think that the frequency of bankruptcy-inducing ‘income shocks’ has increased six-fold since 1970?
Count me as unconvinced about that thesis. None of the measures of shocks or volatility seem to show anything like changes of this magnitude since 1970. It does seem more probable to me that ‘changes in the credit market and decreased costs of bankruptcy’ (including, particularly, the decreasing social costs of bankruptcy), would be more likely to account for such large effects.
And this seems relevant:
“Canada is a country with universal health care coverage. Hence, catastrophic medical expenses are unlikely to be the main cause of bankruptcies in Canada, which is consistent with the lower level of bankruptcies relative to the U.S. However, Canada experienced a very similar increase in bankruptcies (see Figure 1), which suggests that a factor common to both countries is primarily responsible.”
John Emerson 09.18.07 at 12:03 pm
The main reason for rejecting income shocks is an explanation of bankruptcy is that, in the model of the paper, households should respond to increasing variance of permanent shocks by increasing precautionary savings.
The assumption of rationality seems to slip back and forth between predictive and normative. Sometimes you assume rationality in order to build a model of whatwill actually happen. Other times you just use it to blame people for not being rational, and explaining that their problems are their own damn fault.
It would seem that increases in various sorts of employment insecurity and income uncertainty would be the relevant shocks here.
Going further, “consumer confidence” is a major index of the health of the economy. If consumers become more cautious and more rational, the economy suffers. Thus, bankrupts are martyrs of prosperity, sacrificing their own well-being for the good of all.
Marc 09.18.07 at 12:10 pm
Yes, Solcum. In the 1970s corporations had some reluctance to fire their employees en masse to afford extravagent payoffs to their CEOs. This has changed. People really did have jobs with the same company for decades. The 1970s also was the era where the divorce rate skyrocketed.
The idea that none of these factors matter because people aren’t saving more money is, well, ridiculous. An actual experiment would look at the employment histories of people who filed for banckruptcy. Illness, divorce, and unemployment are the predominant causes…
abb1 09.18.07 at 12:13 pm
None of the measures of shocks or volatility seem to show anything like changes of this magnitude since 1970.
May I recommend Going Postal by Mark Ames.
Matt Weiner 09.18.07 at 12:55 pm
It seems to me that John Q’s claim is that income shocks and volatility account for some of the increases in bankruptcy, not necessarily all; while Livshit’s et al.’s claim is that they don’t account for much of it at all. (Hence John Q’s claim that you need to look at both sides of the story.) So John doesn’t need to find a sixfold increase in uncertainty even if bankruptcy has increased sixfold. I don’t think anyone denies that the scattering of credit cards like blow-in subscription cards in magazines has had an effect.
lemuel pitkin 09.18.07 at 1:08 pm
The whole credit card business is centred on the fact that lots of people (about half the population) don’t pay their monthly balances down to zero and therefore carry semi-permanent debt at very high interest rates. It’s hard to imagine that people who have trouble managing their credit cards are computing, in advance, the income risk they face and making precautionary savings to offset this.
Does “hard to imagine” here really mean “logically impossible”?
Kieran Healy 09.18.07 at 1:30 pm
No, it means “consistent with other things we know about consumer behavior.”
SamChevre 09.18.07 at 1:50 pm
I tend toward the “it’s the credit market” explanation. From what I’ve heard, it was very difficult to get unsecured loans in 1970, unless you had substantial assets. (I wasn’t born yet, so corrections will be accepted graciously.)
The fundamental change to bankruptcy law was to make it more parallel to the credit market’s assumptions; expected future income is an asset.
Slocum 09.18.07 at 2:43 pm
“In the 1970s corporations had some reluctance to fire their employees en masse to afford extravagent payoffs to their CEOs. This has changed.”
The 1970’s were also an era of both high unemployment and high inflation (remember Jimmy Carter’s and the ‘misery index’:
http://en.wikipedia.org/wiki/Misery_index_(economics)
The attempt to recast the 1970s as some kind of economic golden age is pretty comical to anyone old enough to remember them.
mpowell 09.18.07 at 4:24 pm
I guess I should read the paper, but if the primary reason for rejecting income volatility as a reason for part of the increase in bankruptcy filings is that people ought to save more money in uncertain times, then that’s kind of funny. I mean, isn’t that just starting with the assumption that you’re right? Do they actually measure the personal savings people accrue as a function of the income volatility that they face to test this hypothesis?
Otherwise I guess you could say that they’re arguing that the failure to save is the cause of the increase in bankruptcy. But then John Emerson would have it right: just blame the people for not being rational.
Doug K 09.18.07 at 8:12 pm
a ‘decline in the cost of bankruptcy’ ?
That no longer applies, since the bankruptcy bill of 2005. Have bankruptcy filings declined since 2005 ? No, they have not:
“May 26, 2006 — The number of bankruptcy cases filed in federal courts rose 12.8 percent in the 12-month period ending March 31, 2006, according to statistics released today by the Administrative Office of the U.S. Courts. Bankruptcy cases totaled 1,794,795 for that period, compared to 1,590,975 bankruptcy cases filed in the 12-month period ending March 2005.”
http://www.uscourts.gov/Press_Releases/bankruptcyfilings052606.html
That undermines their diagnosis, rather.
They go on to say, “Based on experiment 4 in Table 3,we conclude that medical shocks can account for less than 20 percent of the rise in bankruptcies..’
Actual data, rather than experiment, shows that in 2001 more than half of bankruptcies were due to medical bills.
http://www.consumeraffairs.com/news04/2005/bankruptcy_study.html
This doesn’t give me much confidence in their experiments, either.
SamChevre 09.18.07 at 8:16 pm
Doug,
Read the study. It’s a totally bogus headline.
Most critically, they counted ANY bankruptcy with medical debt as a “bankruptcy due to medical expenses”, even if the medical debt was only a minor portion of the overall debt. Also, they counted addiction (gambling, alcohol abuse, drug use) as a medical problem.
SamChevre 09.18.07 at 8:17 pm
Doug,
Also, the time period you reference includes both pre-change and post-change dates. Look at the quarterly values and you can see the effect of the bankruptcy bill.
mq 09.18.07 at 9:05 pm
Canada is a country with universal health care coverage….However, Canada experienced a very similar increase in bankruptcies (see Figure 1), which suggests that a factor common to both countries is primarily responsible.â€
Slocum, if you look at the actual Figure 1 in their paper, it does not support this statement much at all. Not for the last thirty years, anyway.
Jay 09.18.07 at 9:56 pm
You will recall from Keynes that, when a large segment of the population attempts to save at once, the result is a decrease in the velocity of money and a recession. Central banks, having the avoidance of recessions as a major goal, traditionally respond with interest rate cuts. The result is an economy in which savings are discouraged and debt encouraged, even if underlying income volatility is rising.
To my mind, theory matches observation rather well in this case.
save_the_rustbelt 09.18.07 at 11:32 pm
I’m not convinced either.
Out here in the real world medical crises, job losses and divorce (or some combination thereof) are the reasons for a large number of bankruptcies (some estimates of put medical bills at about 50%, probably not that high).
I regrettably became quite expert in bankruptcy accounting issues during the ’81-’82 recession, and we are now in a longer cycle (six years) of high bankruptcy levels, albeit fewer businesses but more individuals. The cycle seems quite endless, as middle class jobs evaporate and medical insurance is harder to acquire.
Drop down to the courthouse and read the files, it will hook you up to reality.
wood turtle 09.19.07 at 2:29 am
A lot of medical treatments, especially for cancer, were not available in the 1970’s. So maybe the reason there were fewer bankruptcies then was that they were dead.
delagar 09.19.07 at 5:00 pm
Samchevre — as someone who recently went through a bankruptcy due to medical debt, I’ll mention that medical debt can cause debt in other areas — you’re paying off the hospital, the doctors, the pharmacists, well, that leaves no money to buy groceries, or pay rent, or save up for emergency car repair, and when the car breaks down, what do you do? You put it on the Visa, that’s what. It might look like you owe sixty thousand dollars for “other” debt; but that money would not have been on the Visa if you hadn’t owed the other money for the hospital bills.
SamChevre 09.19.07 at 5:10 pm
If you said, “medical emergencies cause many bankruptcies,” I would not disagree–but I strongly suspect the big effect is lost income, not medical bills.
Jon H 09.21.07 at 12:13 am
“Look at the quarterly values and you can see the effect of the bankruptcy bill.”
Take some care with that – there was a rush to file before the bill took effect, which would distort things a bit if you simply compare the quarter when it took effect versus the prior quarter.
Comments on this entry are closed.