How Much Credit Card Debt Per Household?--I've Lost $531 Billion

I don't like to disagree with financial columnist Liz Pulliam Weston over at MSNBC.com or our fellow bloggers at The Consumerist. I especially don't like to criticize the Federal Reserve lest the men with black helicopters show up to take me away. (It's still not really clear to me where the Federal Reserve fits in the power structure, and this didn't help.) All of them do good work, especially you Federal Reserve! Pulliam Weston wrote a column this week called "The Big Lie About Credit Card Debt" and The Consumerist also picked up on the story. 

Pulliam Weston rightly criticized a recent press release from CreditCards.com that claimed the average American household owed $9,300 in credit card debt. She pointed out that figures comes from dividing the total amount of credit card debt by the number of households. Averages are deceiving. As Pulliam Weston explains, "[I]magine that you and 17 of your friends were having dinner with Bill Gates and Warren Buffett. The average net worth of a person at that table would be about $5 billion. The fact that everybody else's personal net worth was a lot less wouldn't affect the average that much because Bill and Warren are so much wealthier than the rest of us." (Personally, I like the anecdote about the person with  one foot in a bucket of ice water and another foot in a bucket of steaming hot water, but who's OK, on average.) The same point would be true for a small number of households with a huge amount of credit card debt. They would skew the average amount of debt owed.

Instead of the $9,300 figure, Pulliam Weston cites statistics from the Federal Reserve's 2004 Survey of Consumer Finance that, for persons who carried a credit card balance, the median amount owed was $2,200. Half would owe more than the median, and half would owe less than the median--that is what a median is. Her point is that U.S. households are not as desperately in credit card debt across the board as the CreditCards.com statistics would have you believe, but she also emphasizes that a sizeable number of U.S. households are in dire financial straits. My problem is that I can get the numbers from the Fed's Survey of Consumer Finance to jibe with other consumer credit figures it releases.

The Fed last released the results for the triennial Survey of Consumer Finance from 2004, a summary of which is here. I'll therefore use 2004 figures for purposes of comparison. According to the U.S. Census Bureau, there were 112 million U.S. households in 2004, and the Survey of Consumer Finance says only 74.9% of U.S. households had a credit card and of households with a credit card only 58.0% carried a balance each month. The average balance for those who carried a balance was $5,100 (proving Pulliam Weston's point that a small number of large debtors skew the average). Other Fed figures, specifically Federal Reserve Release G.19 which catalogs consumer credit, show that there was almost $780 billion in revolving consumer debt at the midpoint of 2004, which I'll use a rough estimate of the average revolving consumer debt outstanding for all of 2004. "Revolving consumer debt" is basically credit card debt.

Doing some simple math, the 58.0% of the 74.9% of U.S. households who had a credit card and carried a balance in 2004 translates to a total of 48.7 million households meeting that description. According to the Survey of Consumer Finance, these 48.7 million households owed an average of $5,100 for a total of $248.1 billion. That leaves $531.9 billion unaccounted from the Federal Reserve's count of how much revolving consumer debt was outstanding in 2004.

What could explain this missing $531.9 billion? I see several possibilities, all of which together probably add up to the truth:

• The Fed's count of total revolving consumer debt would include amounts from consumers who pay off their credit card bills each month. These amounts, after all, are "outstanding" even if only for a short period. Again using the estimates from the Survey of Consumer Finance on the percentage of consumers who paid off their credit cards each month, the average monthly charge for these users would have to be over $15,000. That seems an implausibly high number even after considering that a small number of wealthy credit card users might skew the average monthly. Consider that the Survey of Consumer Finance found only $250 as the median monthly charge for those who pay their credit card bill each month. Still, the amount charged by monthly payers--so-called "convenience users"--undoubtedly accounts for some of the missing $531.9 billion.
• The Fed's definition of a "household" differs from the Census Bureau's definition, which the Fed does explain. But the definitions seem to overlap substantially, meaning that any discrepancies here have to be small. Still, this difference could explain a small part of the missing $531.9 billion as different denominators would be involved in my calculations above.
• "Revolving consumer credit" is slightly different from "credit card debt." An open, personal line of credit at your bank would be counted as "revolving consumer credit," but you probably would not answer on a survey that it was "credit card debt." This explanation would seem to be able to account for only a small part of discrepancy. In any event, from a policy standpoint, if the Survey of Consumer Finance does not treat this sort of debt as a type of credit card debt, it is giving a distorted view of the debt obligations of the U.S. household. Financially and legally, the two types of debt are virtually identical.
• Respondents to the Survey of Consumer Finance simply were not accurate in estimating the amount of credit card debt they owed. For my money, this explanation probably explains a fair bit of the discrepancy. Respondents may have been embarrassed to admit to the interviewer exactly how much credit card debt they owed. Even more likely, respondents probably don't want to admit to themselves how much they owe and systematically underestimate the total amount. Although respondents are allowed to consult documentation during the interview, many do not, and the documentation that is consulted tends to be a tax return, not credit cards statements.

If the underreporting hypothesis is right, the truth for the average credit card debt owed per U.S. household probably lies somewhere between the extremes of $5,100 and $9,000. Because the large debt households skew this average, the median figures are a better indicator of credit card debt in a typical household. Given the likely underreporting of credit card debt, the $2,200 median figure from the Survey of Consumer Finance (for households that carry a balance) is probably too low although it is impossible to say how low. Again, the Survey of Consumer Finance tells us that 42% of U.S. households don't even carry a balance on their credit card, although one wonders again about the accuracy of the self-reporting again. Nevertheless, Pulliam Weston's basic message is right--we're not in as bad a financial shape as some of the scare-mongering statistics would have us believe. At the same time, the U.S. household is not in as good a shape as other statistics might indicate.

I may be missing something in my analysis, which is entirely possible. I spent a little bit of time on this issue this evening, even trying to explain some of it to my son. (Be glad you were not raised in our family, although while I wasn't looking he did escape to catch an episode of Mythbusters.) I did not spend the weeks or months of time I would with one of my academic papers, and it's time for bed. Comments, as usual, are open.

Oh . . . and, go Federal Reserve!

Comments

[naiserie]

your links to the stories in the first paragraph are missing. they should be:

http://articles.moneycentral.msn.com/Banking/CreditCardSmarts/TheBigLieAboutCreditCardDebt.aspx
and
http://consumerist.com/consumer/lending/the-truth-about-credit-card-debt-284943.php

[Bob Lawless]

Thanks, naiserie. I've fixed the broken links.

[Marie Carnes]

I wonder if they considered department store credit cards in their factors. I mention this only because I've heard someone argue that these accounts are just credit and specifically not credit card debt. Although, to me, there's really no difference. If not, that may be part of the discrepancy.

Bob, re Mythbusters. I don't know if this in their realm, but maybe you could persuade them to take on the amount of credit card debt per household question. The guy from the show is a regular at MetaFilter and he's always on the look out for ideas.

P.S. Love that Fed!

[Ed Mierzwinski]

Bob-- You've done a comprehensive review of the possible reasons for the mean/median discrepancy. We have a similar analysis in the footnotes of our 2002 report "Deflate Your Rate" http://tinyurl.com/3d2tda and we (PIRG) believe that "under-reporting in survey answers" is a key part of it. Tamara Draut of Demos has a 2005 report The Plast Safety Net, http://www.demos.org/pub654.cfm which finds: "According to our survey, the average credit card debt of a low- and middle-income indebted household in America was $8,650; the median was $5,000. One-third of households had credit card debt over $10,000, while another third reported credit card debt lower than $2,500."
Ed Mierzwinski
USPIRG

[Angie Littwin]

This is such a comprehensive post. I know I will find it useful in the future. Regarding your first bullet point, I'm not sure which question you used to determine who was a "transactor" and who was a "revolver," but the SCF does have a question asking what the respondent's balance was *after* she paid her most recent bill. Using the data from that question would allow you to avoid the transactors-in-revolvers'-clothing problem.

[FIRE Finance]

This is a nice comprehensive post about the discrepancies that arise through various methodologies used in polls. Also its a nice round up of Ms. Weston's MSN Article. However we are still perplexed about the $531 billion discrepancy? Still not able to figure the cause behind such a huge mismatch. Thanks for the great post.

[Bill]

Let's not forget that in the three years since that survey has taken place that revolving debt has continued to grow.

Revolving debt at mid-point of 2004: $780 billion
Revolving debt at mid-point of 2007: $904 billion

This is the same period that many moved their revolving debt to non-revolving through refinance and home equity loans. With the housing bubble and credit crunch most recognize it's time to get rid of that anchor holding them down and start saving. Since our economy is so dependent on consumption this will start a negative spiral down. The desperate will take on new debt in attempt to stay afloat and many will be unable to continue servicing the debt they have. Unfortunately I think this is a large enough percentage of our people that it will be some time before we see a minus sign in the G.19 reports.

[Quitaque1]

Dateline August 23, 2007.
I'm NOT an economist, so maybe I shouldn't be posting here, but I'd sure like somebody to check my figures, which are estimates culled from several different articles.
Average annual income for a person employed full time in America: $ 40,000
Total number of people employed full time in America: 123,000,000
Total amount of all personal debt owed by Americans, including mortgages, car notes, student loans, credit cards, etc.: 12.1 trillion (12,100,000,000,000 if I've counted the zeroes right)
Average amount of debt for each employed person: $ 98,000
Conclusion: The average American owes 2.5 times his annual salary to somebody for something. Granted, most of the debt is offset by assets (houses, cars) which the debt was used to purchase. But if this hypothetical average American is paying 7% interest (on average) to service this debt, then the average American is paying around 17% of his annual income as debt service. And when you consider that I've used gross annual income without making any allowance for taxes, the situation looks worse. I'm thinking the average American is paying 20% of his after tax income as INTEREST. If these figures are right, then the current Wall Street problems are just the tip of the iceberg. So someone tell me I've miscalculated.

[Kevin]

Your G.19 data is misleading you. A huge chunk (>50%) of that "revolving credit" is in something called "securitized pools". See: http://www.federalreserve.gov/releases/g19/hist/cc_hist_r.txt