As someone who rails against the “let’s assume we have a can opener” nature of conventional economics, I like to think that I’m relatively immune to making false assumptions (one of my favourite expressions is that “to ASSUME makes an ASS out of U and ME”). So it’s humble pie time: I made a false assumption about the way that the Federal Reserve Flow of Funds recorded debt, and it requires a significant revision of my estimates for the level of US private debt.
The false assumption was that the Federal Reserve Flow of Funds aggregate debt table—Table L.1—only recorded debt to banks. Since banks can lend not only to businesses and households but also to other financial institutions that are not banks, and bank lending is crucial to my macroeconomics, I added the rows 3, 4, 5 and 9 of Table L.1 together to produce an aggregate private debt level for the USA since 1952 (see page 9 of this file).z
Well, Table L.1 would work that way if I (or any other exponent of “endogenous money“) had designed it. But I didn’t. It’s a minor miracle that something as comprehensive as the US Flow Funds exists at all, but it wasn’t put together with any theory of the role of debt in the economy. Instead, it records all debt, regardless of where it originates—including not only lending by banks, but borrowing by them as well.
As a consequence, the Flow of Funds record of household debt includes debt issued non-banks (credit unions, S&Ls, non-bank mortgage originators, etc.); its record of corporate debt includes corporate bonds, and its record of financial sector debt includes borrowing by banks. It’s relatively easy to cope with this for both households and businesses, but separating bank and non-bank lending in the finance sector is about as easy (and as much fun) as unscrambling an egg. So though it necessarily omits some bank-created debt, the easiest way to proceed when calculating how much debt the private sector is carrying is to add together household and business debt and ignore financial sector debt.
I’ve found this out because I’m currently helping The Governor’s Woods Foundation, a US philanthropic body, to assemble a long run database on private debt. Moritz Schularick has also assembled such a data set for INET; the Governor’s Woods data set covers more countries, and also plans to provide disaggregated data.
The Governor’s Woods Foundation is one of the few private institutions that “gets” endogenous money, and understands the crucial role of changes in private debt in macroeconomics—primarily because its founder Richard Vague has an intimate knowledge of banking from his three-decade career as a banker, during which time he established two of America’s largest credit card companies.
He became concerned about the rocketing level of mortgage debt in the mid-2000s, mainly because of the impact this could have on the capacity of his customers to meet their credit card commitments. He was assured by the chief economist of the company that subsequently bought his business that the rising level of mortgage debt was absolutely not a problem, because his customers’ rising debts were more than offset by their rising assets. “That’s a trade I’ll take any day”, the economist remarked—at which point Richard decided to take the other side of the deal, and also began to suspect the soundness of economic theory.
The company was Barclays PLC, and Richard sold out to them in 2004.
Having exited banking, Richard decided to investigate whether the pattern he had seen in the USA was replicated elsewhere—and found out the hard way that formal, comparable statistics on private debt simply didn’t exist. So he started the Debt-Economics project to develop such a database, and a debt-aware approach to economics based upon it. I’m spending a month in Philadelphia, helping Richard and his team to develop it. Currently they have data on 21 countries, and the current projects are to acquire data on China today and the USA in the 19th century—and any assistance on either would be most appreciated.
We differed on one point: whether financial sector debt should be included or not. This hinged on whether financial sector debt was debt by non-bank financial institutions to banks—which is what I had assumed it was—or a grab-bag of all forms of debt issued by financial institutions including banks. So I dived into the minutiae of the Flow of Funds definitions, and found that Richard was right and I was wrong.
Figure 1: The Financial Sector component of the Flow of Funds and its components
This makes a pretty drastic difference to the level of private debt today since at its peak, financial sector debt (as recorded by the Flow of Funds) hit 120% of GDP. With all of this removed, the peak private debt level in the USA in 2009 was 180 per cent of GDP—well below not just the previous level I cited of 303%, but also below the peak that deflation drove private debt to back in the Great Depression (in fact not all of it should be removed—more on this below).
Figure 2: Old and corrected figures for the USA’s private debt to GDP ratio
So does that scotch my debt-based analysis of capitalism as well? Nope. The relationships between change in aggregate private debt and economic activity—relationships that today’s mainstream economists say shouldn’t exist—are as strong as with the previous somewhat erroneous data, because it’s the change in debt that matters for the level of economic activity, and both private non-financial and financial debt moved in the same direction most of the time. The correlation coefficient between change in private debt (completely excluding financial sector debt) and unemployment is -0.66 for 1980 till now, and -0.9 for 1990 till now.
Figure 3: Correlation of change in private debt and unemployment
What it does change is the relative scale of today’s debt crisis to the Great Depression. Taking 1930 and 2008 as the start of both crises, the aggregate debt level in 1930 was 180% of GDP, versus 175% in 2008—so the levels were broadly comparable before deflation kicked in during the Great Depression and set the ratio rocketing up to 240% of GDP in 1932 (though in reality today’s ratio is probably still slightly higher, because there is still some finance sector debt that should be included—the problem is isolating it from the rest).
The cause of this problem—apart from my own failure to check the definitions in the Flow of Funds properly—is that private debt isn’t regarded as an economic issue by mainstream economists. Even those who take it into account at the moment—such as most obviously Paul Krugman—argue that it matters economically now only because of a “Liquidity Trap” when interest rates are at the “Zero Lower Bound”. The perspective that private bank debt is economically significant at all times remains the province of just a handful of economic renegades—including Michael Hudson, Dirk Bezemer, and Richard Werner as well as me.
Since statisticians only collect the data that economists tell them matters, they don’t collect data on private debt in a systematic way. That’s why it’s a minor miracle that the Flow of Funds exists in America at all—it has no equivalent in any other country in the world. If the debt-focused approach that I take to macroeconomics was dominant, then the Flow of Funds would have plenty of cousins across the globe, and they would all clearly distinguish bank debt from non-bank debt.
This is not a new phenomenon: the same problem bedevilled economists interested in unemployment prior to the Great Depression. Up until then, unemployment was not seen as a macroeconomic issue by the majority of economists, but an individual and microeconomic one: since Say’s Law was assumed to apply, the only reason for unemployment was that some workers must be holding out for a higher wage than the market was willing to offer. So if you wanted statistics on unemployment, you had to check Trade Union records—which is what Bill Phillips largely did to assemble his database from which the Phillips Curve was derived.
In a future post I’ll publish another series that includes the elements of bank lending to non-banks that exist in the Flow of Funds data on private debt (Table L.1), and that will have a higher estimate for private debt since 1952 (when the Flow of Funds was first published), but for now Figure 4 is my preferred measure of America’s debt levels over the last century.
Figure 4: Long term US private and public debt levels
Hopefully at some point economists in general will realize the economic significance of private bank debt, and statisticians will then collect data on bank debt in a systematic way around the world. But at present they don’t, which is why I’ll have to waste some valuable time unscrambling the omelette of the US Flow of Funds records of financial sector debt, and why projects like the Governor’s Woods debt database are so essential in the meantime. You can’t diagnose a problem like the ongoing economic malaise of the world if you don’t even measures its major cause.
The Governor’s Woods Foundation database on private debt is available at:
Visit the site, check out the data, and if you can help us extend it—particularly with Chinese and 19th Century US data—please get in touch.