« Pictures of adverse selection in an insurance market, with and without death spirals | Main | Naive vs rational expectations is a (partly) false dichotomy »


Feed You can follow this conversation by subscribing to the comment feed for this post.

About #3, couldn't you just look at data for debt of the nonfinancial sector? Look at the Fed's Flow of Funds tables. If I recall correctly, the US ratio of nonfinancial sector debt to financial sector debt is now about 3, where it was for a couple of decades before skyrocketing during the financial crisis when financial sector debt actually contracted while nonfinancial sector debt continued to grow. Since then the ratio has returned to 3.

If rich people in poor countries have foreign bank accounts, shouldn't that lower the debt/GDP ratio in the poor country and raise it in the rich country? This would relate to trust.

If a simple economy uses gold and silver coin as money, then that portion of wealth is not matched by debt. The switch to bank money would raise the debt/GDP ratio.

I presume that fiat currency doesn't count as private debt. So, the development of banking would also show an increase in private debt. If fiat currency counts as public debt, then it would be a shift from public to private debt. If fiat currency is treated like paper gold, then the development of bank money would be an increase in total debt.

The development of trustworthy stock markets would be a factor that would go the other direction. A large portion of saving and investment starts as successful proprietorships reinvesting profit to expand their own operations. If the yield on further expansion is low, then there is a motivation to either enjoy the accumulated wealth by consuming more or else accumulate gold and silver in the treasure box. (Or U.S. currency.) No debt involved. Debt allows the owners of firms that made profits in the past to fund better yielding investments by other firms by purchasing the bonds they issue. When the heirs of John Rockefeller purchase stock issued by General Motors rather than bonds, then we go in the other direction--back to equity finance and less debt.

Jeff: I think you could.

There's the "Anglo Saxon" model of direct finance, where households buy corporate stocks and bonds, and the "Continental" model of finance, where households save in banks, and banks lend to corporations. As those names imply, some countries tend to go more one route, and other countries tend to go more the other route, even when equally rich. But I think it's also true to say, from a very long run perspective: that poor countries do neither; middling countries tend to go more Continental; then as they become rich they tend to go more Anglo Saxon.

It may be a question of trust. With low trust nobody borrows-and-lends. With middling trust, you only borrow-and-lend via a trusted middleman. With high trust, you borrow-and-lend with anyone, and don't need a middleman.

But this would tend to create a non-linear relationship between debt/GDP and GDP per capita, because there's more double-counting in the middling countries. But I don't really see that non-linearity in the data. (Though, to be honest, it's really too noisy to say either way.)

Bill: very good points. I missed those. In the World Bank data you see some small countries that have banks that cater to foreign clients having very high credit/GDP ratios.

I find that for a lot of the reasons above, the direction of causality is not clear.

Also, I don't know if there is such a thing, but I'd be curious to see a breakdown into debt that is used for investment versus debt that is used to consume more final goods now instead of later.

To me there is a huge difference between businesses financing tools, infrastructure they need to grow, people financing education or a basic car to get to work and consumers financing big screen TVs, luxury cars or overly nice houses.

Financial institutions that charge fees on every transaction and loans have incentives to lend as much as possible including getting people to gamble away their future for a bit more luxury now. I wonder how prevalent this "bad" type of lending is and if it gets more predominant in more prosperous countries.

Benoit: suppose I own a business, and a luxury car, and have a $10,000 debt. Did I borrow the $10,000 to finance the business or the car? We can't tell, just by looking. Even if we know that the loan is secured by the car, I might have bought the car anyway, and just invested less in the business, if I had been unable to borrow. It might be a business loan, but the car provides better security.

I am beginning my comment with a quote from Item 0.:

"Total private debt is also a measure of the amount of trade. It's a measure of intertemporal trade. It's a measure of the accumulated stock of past flows of borrowing-and-lending that has not yet been repaid."

Total-government-debt is also a measure of past trade, identical to private debt. There is however, a very important difference between the two measures of past trade: Private debt measures promises from INDIVIDUAL WORKERS to perform in the future. Government debt measures promises from GOVERNMENT to perform in the future.

When individuals promise to perform in the future, they either do perform or lose their property. The result is a measurable increase in employment when private loans increase above base rate.

When government promises to perform in the future, the promise is to replace-existing-debt-with-additional-debt or increase-taxes. As a result, the relation of increased debt to improved employment is much less noticeable. This is to be expected; introduction of a project such as a space program would employ people for a sustained people, increasing GDP. Raising wages for government workers would increment for inflation far more than it would increase employment.

It would be an interesting exercise to tease from the data the employment-effectiveness of increased private debt VS increased government debt. I have not done this exercise.

TYPO: Not " sustained people". Please substitute "sustained period."

Roger: this post is about the long run question. Whether or not private or government debt increases aggregate demand and employment is a short run macro question. The whole theme of this post is that everyone keeps talking about short run questions and ignores the long run question.

The OECD has data for 2011 that I pulled down into Excel. It should be in your mailbox. Limited data set but it is upward sloping.

Thanks very much Kathleen! It does show an upward slope, roughly linear (though with a lot of variation around that slope, of course).

I have made 4 attempts to upload it into the post, and have failed miserably. Typepad seems to like JPEG images.


What is *your* debt-income ratio? And how does it compare to poorer people?

The data you refer to is gross debt, not net debt. Your theory seems to be about net debt.

David: my gross debt/GDP ratio is around 1% (must remember to pay my credit card), but that's just because I'm an old guy. It peaked around 200% when I was in my late 20's/early 30's (mortgage).

I wanted my theory to be about gross debt, and thought it was. Net debt is always zero, in a closed economy.

"which means that if per capita GDP doubles, per capita private debt quadruples."

Non sequitur. It might be the other way around. I. e., if per capita private debt doubles, per capita GDP increases by 41%. :) Perhaps as people, whether consumers or business owners, are more credit worthy, they make each other more wealthy, with the help of loans.


You are well aware that the long run is the sum of the short runs.

You might add item 7. Government Policies.

Government policies, ranging from private loan guarantees to stable and predictable taxation, may contribute to conditions favoring private loans. Government taxation of interest received from loans and government induced inflation-expectations are both examples of government discouraging additional private lending. Richer countries support private lending and reward future performance, to the long term betterment of the entire economy.

Min: Yep. I was trying to find a way of saying it that was neutral with respect to causality. Either could cause the other, or some third thing might cause both. I meant to imply only correlation.

Roger: "You are well aware that the long run is the sum of the short runs."

No I'm not aware of that. The level of water in the lake is not determined by the determinants of the waves in the lake. Rainfall determines the one; wind determines the other.

This graph shows RGDP , real consumer debt, real net worth and (consumer debt)/gdp. I've used a log scale for the first 2. The interesting thing is the behavior of the ratio. Note the flat period from 1960-1980 and the slope increase from 2000.

I have some ideas about this, but I'd like to hear what other people think. I haven't shown it here, but over the same period the ratio of liabilities to assets has gone from around .05 to .2.

Two additional reasons for varying debt/GDP ratios, both due to intricacies of national accounting:

1. Owner-occupiers put their money into their houses. Tenants lend it to banks who, in turn, lend it to landlords. This may explain the high value for Switzerland, the proverbial land of happy tenants.

2. Sole proprietors put their money into their firms. When the legal status is changed into that of a corporation, financial assets and liabilities emerge (whereas net worth remains unchanged). The fraction of corporations is higher in anglo-saxon as compared to continental countries.

Peter: That's US data, right?

Yep, that time scale is about long enough to see the long run upward trend in debt/GDP.

I can't think of any obvious reasons for the flat period in the 1960's and 1970's. Baby boom demographics???

(Your being able to get the data, and post the graph, in comments, makes me feel even more of a failure at computer stuff! Sniff.)

Herbert: Good points. I wonder why there are those differences in tenants/owner-occupiers and sole proprieters/corporations? Different laws and tax treatments?

(Off-topic: thank you very much for those Turgot and Homburg tips a few days ago in comments on my retirement post. Homburg is especially good and important, if a little too mathy for my brain. I have a very vague memory that I might have skimmed through the Homburg paper decades ago. I've got a post half-written on it.)

Nick Rowe: "Either could cause the other, or some third thing might cause both. I meant to imply only correlation."

¡Bueno, Nick! :)

GDP growth rates are an important driver for things like this. Debt ratios tend to get higher as nominal GDP growth slows..

In the developed world experience, you had a mass default in the 1930s, and then highly cautious lenders. This dropped private debt ratios near zero. Going into the Depression, US private debt was quite high. Then as time passed, lending standards have dropped in each cycle. There also is the effect that more levered firms will gain market share at the expense of their more cautious competitors. Hyman Minsky described this process a couple of deacdes ago.

Assuming your not kidding, you make yourself a free account at


You specify your graph, choose the link option, copy the resulting URL and save your graph.

then you take the url and use the construct below, delete the $ (which I put in to disable it so you could read it), and substitute your URL for mine.

<$img src="http://research.stlouisfed.org/fredgraph.png?g=o98">

In a first order, big picture view there should be no difference between debt financing and equity financing, so there is no reason to look at private sector debt alone, you should look at the enterprise value of the private sector, and just assume that the private sector will allocate some of that as debt and some of that as equity however it wants.

In which case it makes sense that
1) The ratio of EV/GDP is constant
2) as interest rates fall, the ratio EV/GDP goes up non-linearly (as 1/r)

Then, I would adjust for things like

3) as credit constraints loosen and capital markets becomes more established, EV/GDP goes up a lot as many more projects become funded that would not be funded because of credit constraints, and land prices rise as households are able to borrow to buy land and are no longer required to pay for land with cash up front.

Yes that's US data from the public economic statistics database at the St Louis Fed. They have many thousands of data series there.

BTW, you'll notice in the graph I posted that the ratio of debt to GDP is increasing at the rate of 1.8% a year. Another similar 60 years would reduce US private net worth to $0, so the future can't be like the past in this respect.

You said:

"And let's not forget that someone's debit is someone else's credit. The amount of borrowing-and-lending is one thing, not two things."

This is a bit misleading. There are problems

1) Assets are stocks that are valued by their flow price. This value is not realizable if a significant part of the asset holders want to sell at the same time. Debt OTOH is valued at par. Mark to market accounting makes this particularly painful.

2) Debt is subject to default. Over $1 trillion of mortgage debt has defaulted to date in the current depression. Default cancels both the asset and the liability, but there is usually no economic benefit from the cancellation of the liability. Moreover foreclosure destroys value in a number of ways.

3) There are many debtors who are not creditors.

Bottom 40 percent

Mean household net worth -$10,600
Mean household financial (non-home) wealth -$14,800

We have here very different wealth effects and propensities to consume.

From a fascinating paper:

Fisher Dynamics in Household Debt: The Case of the United States, 1929-2011
J. W. Mason and Arjun Jayadev


"The importance of gross liabilities becomes obvious in periods of financial distress, when a signicant fraction of units face diculties in servicing their debt. The focus on net wealth implicitly assumes that assets are liquid, and can be mobilized (either through sale or as collateral) to meet debt obligations. But assets cannot always be reliably converted to means of payment, either because their market value uctuates, because they are inherently illiquid, or because they become so in a crisis. Thus leverage, as opposed to net wealth, matters mainly in the context of liquidity constraints. If units' assets are not reliable sources for either funding or market liquidity, then the capacity to service debt out of current income becomes paramount. (Tirole, 2011) These are the conditions in which leverage matters."

and this explains at least part of the graph:


During this period, households switched to primary surpluses, but as a result of financial deregulation and higher interest rates following the Volcker shocks, household debt ratios rose at about half the rate of the postwar years (1.4 percent annually compared with 2.6 percent). This increase took place despite primary surpluses averaging 1.4 percent of household income. With growth rates essentially unchanged from the previous period, the growth of leverage was entirely due to higher real interest rates, with higher nominal interest rates contributing two thirds of the increase and lower in ation the other third. It is striking to realize that over this period, accounting for about half of the post-1980 increase in leverage, saw the lowest levels of household spending relative to income of the whole postwar period. Leverage rose only because of the effect of higher real effective interest rates on households' existing stock of debt."

The true consumerist society substitutes services for goods plus debt, think mobile phones and cable boxes, which would tend to lower debt ratios as these have high discount rates relative to company borrowing.

The 1st variables that come to mind are: size of finance industry, housing and rental prices, consumer ease of access to credit from finance industry (bankruptcy laws, state of venture cap), laws pertaining to housing values (nonrecourse or recourse mortages, whetehr 20 or 40yr mortgages and such), recent change in housing/rental prices, recent economic growth or recessions obviously.

@Nick ("Herbert: Good points. I wonder why there are those differences in tenants/owner-occupiers and sole proprieters/corporations? Different laws and tax treatments?")

Tax treatment can be an important explanation. Switzerland taxes imputed rent expenditures of owner-occupiers, thus closing what is the most important tax loophole in most other countries. In case of sole proprierters I am not so sure whether liability laws are a more important influence.

...I guess healthcare must be considered for our special needs friends to the south.


I think you are wrong. The debt you are talking about does not just constitute financial claims against other agents (this must indeed sum to zero). Much of the debt you are talking about is backed by capital (e.g., mortgage debt). You are missing the asset side of the balance sheet.

David: I know I'm missing the whole asset side of the balance sheet. These data miss the whole (real) asset side of the balance sheet. But these data are still something I would like to explain.

For example, if mortgages are easy to get, I might borrow $100,000 from you to buy a house. Alternatively, you buy the house, and I rent it from you. Same real assets in both cases, but in the first case I have a $100,000 debit and you have a $100,000 credit. What determines whether we see one vs the other?

@ Herbert

very interesting point about a Swiss detail with the tax code, which kind of reverses the general correlation

@ all, Nick

General: please see OECD Annex 58, subtract the mortgage part of the household liabilities, the remainder is very high in Japan and Canada (60% vs some 35ish in the rest), divide the mortgage part by wiki/List_of_countries_by_home_ownership_rate
(nudge Germany 3 % points up for old data shortly after reunification, numbers beyond the median of 1.43 warrant some attention for possible pecularity)

When I fumble the Japan mortgage number up to 87%, because they might get loans to finance housing without the backing with the house, just speculating, than Japan looks with a 1.45 vs 1.43 above, and 38% other household debt pretty similar to the others, and it makes Canada the sole odd man out`. It would be interesting to know how Japanese banks deal with their right to go after the whole property of the debtor, and how that works for the swaps mentioned for Canada. Could somebody explain this here? That would be very nice. I am really curious.

For the reasons of high/low homeownership, both settlement densification (e.g. NYC vs Kansas) and the tax treatment are important (wiki/Home_mortgage_interest_deduction) AND wiki/Property_tax, What would be the median number for Canada for that?

I am trying to post that in a table, please bear with me

country liability mortgages ownership mort/owner residual deviation
Canada 163 103 68 1.51 59 high residual
France 92 78 55 1.42 14 low residual
Germany 95 65 45 1.44 30 low homeownership
Italy 89 55 82 0.68 34 low mortgage ratio
Japan 126 90 60 1.50 36 high residual, if true
UK 173 134 69 1.95 38 high mortgages
US 116 85 65 1.30 31

all numbers are percentages, of "disposable income" according to the OECD, missing 2011 numbers for France and UK replaced by older FR 2008, cough, and UK 2009

liabilities are total liabilities

residual is total minus mortgage

mort/ owner is average mortage debt / home ownership fraction

and from this it would look like Canada has a solid residual debt problem

" Alternatively, you buy the house, and I rent it from you. Same real assets in both cases, but in the first case I have a $100,000 debit and you have a $100,000 credit. What determines whether we see one vs the other?"

All sorts of stuff. In the rental case I'll probably take out a mortgage and use your rent to pay it off. If this couldn't work, then I wouldn't buy, because rental property was too expensive. As for your debit, what is the form of the obligation - mortgage (recourse or non-recourse?), other secured loan, unsecured loan, fixed term loan, demand loan... This is all about liquidity and counter-party risk and these things matter.

Banks like to write mortgages, because they can securitize them. Go to the bank and try to get a bridge loan. Banks are more careful when they retain the risk. Thus mortgages are the flavor of the week.

genauer, *Canadian consumers* have a solid residual debt problem. I'm not too worried about that. I'm more concerned we are AGW-ing future inventors. And that electing a federal gvmt from a province with 0.5% PhD Graduation rate vs the 8.5% national avg, will keep total cdn R+D low. When I envision a sensor network to police future WMDs, it is little cameras and insects made of metals and plastics. About to read a nano-ink paper; coating copper nanowire with oxidation-resistant cupro. I'm wondering if the $100 3d printer innovation in SK can pattern PE, PC, or maybe PS, instead of acrylic as the former are all anti-microbial and could give people flooring and flat surfaces, covers, in a pandemic.
But that is a question of whether to export just raw materials or become the new early 20th century USA. They aren't that bright; lucky to be 2020s China in their two party future. If India can keep their software gorwing and port to medicine (but not immunology and neurology to excess), they might become the 2000s Canada, USA and UK, in 2050.

....I guess what your long run private debt/GDP should be, depends on the education/experiences of your present and past MPs, the present education/experiences of Crowns CEOs, and the present and past education of your major private employers. The past is relevant in that it affects debt levels. Too much debt and a downturn threatens sovereignty or at least quality-of-life. Feds, Crowns and corporate CEOs are all potential employers in a downturn. I would perhaps define cdn Gini rises since 1990s as a downturn; certainly in the USA since Reagan they have been in intellectual Great Depression. Lots of youtubing and texting to disguise all their best do is fight their dumb rich. Here it doesn't really matter: we've chosen low R+D and to make commdities for those that create future products and services.

Based on the slope of the debt/gdp ratio we can divide the period 1945-2011 into 5 periods.

1) 1945-1964 the postwar boom
2) 1965-1982 the era of inflation and financial repression
3) 1983-1999 deregulation and end of inflation
4) 2000-2007 housing bubble
5) 2008- the great recession

I've added the CPI and the prime rate.

It kind of looks like all that can stop debt increase is inflation or a major depression.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad