Home > Uncategorized > The seven countries most vulnerable to a debt crisis (10 graphs)

The seven countries most vulnerable to a debt crisis (10 graphs)

from Steve Keen

For decades, some of the most important data about market economies was simply unavailable: the level of private debt. You could get government debt data easily, but (with the outstanding exception of the USA—and also Australia) it was hard to come by.

That has been remedied by the Bank of International Settlements, which now publishes a quarterly series on debt—government & private—for over 40 countries. This data lets me identify the seven countries that, on my analysis, are most likely to suffer a debt crisis in the next 1-3 years. They are, in order of likely severity: China, Australia, Sweden, Hong Kong (though it might deserve first billing), Korea, Canada, and Norway.

I’ve detailed the logic behind my argument too many times to count, and I won’t repeat it here (if you want to check it out, try this Forbes post on Krugman, this one on money, this one on the Fed, or this one on our dysfunctional monetary system). The bottom line is that private sector expenditure in an economy can be measured as the sum of GDP plus the change in credit, and crises occur when (a) the ratio of private debt to GDP is large; (b) growing quickly compared to GDP. When the growth of credit falls—as it eventually must, as growing debt servicing exhausts the funds available to finance it, new borrowers baulk at entry costs to house purchases, and numerous euphoric and Ponzi-based debt-financed schemes fail—then the change in credit falls, and can go negative, thus reducing demand rather than adding to it. 

This is what caused the Global Financial Crisis, and the simplest way to simply substantiate my argument—which virtually every other economist on the planet will advise you is crazy (except Michael Hudson, Dirk Bezemer and a few others)—is to show you this data for the USA. The crisis began as the rate of growth of credit began to fall, and the Great Recession was dated as starting in 2008 and ending in 2010. As you can see from Figure 1, the sum of GDP plus credit growth peaked in 2008, and fell till 2010—at which point the recovery began.

Figure 1: America’s crisis began when the rate of growth of credit began to fall


The BIS database lets me identify other countries—several of which managed to avoid a serious downturn during the GFC—which fill these two pre-requisites: a high level of private debt to GDP, and a rapid growth of that ratio in the last few years. The American ex-banker turned philanthropist and debt reformer Richard Vague, in his excellent empirical study of crises over the last 150 years, concluded that crises occur when (a) private debt exceeds 1.5 times GDP and (b) the level grows by about 20% (say from 140% to 160%) over a 5 year period.

America fitted those gloves in 2008, as did many other countries—all of which are either still in a crisis (especially in the Eurozone), or are suffering “inexplicably” low growth after an apparent recovery (as is the case in the USA, the UK, and so on). Using the BIS database, I can identify 21 countries that meet Richard’s first criteria, but to “go for broke” on this forecast, I restricted myself to the 16 countries that had a private debt to GDP ratio exceeding 175% of GDP. To simplify my analysis, I then limited the second criteria to countries where the increase in private debt last year exceeded 10% of GDP. That combination gave me my list.

Firstly, here is the private debt to GDP data for that set:

Figure 2: Countries with private debt/GDP > 175% & debt growth in 2015 > 10% of GDP, ranked by debt growth


Here is the rate of growth of debt data.


Clearly China and Hong Kong are the standouts here—and the fact that Hong Kong’s credit growth has been plunging since mid-2014 is why it may deserve top billing. But in all of these countries, credit growth is a very significant component of aggregate demand, and when it slows down, their economies will go into recession.

The next seven charts repeat the information shown in Figure 1 for each of these seven countries.


Timing precisely when these countries will have their recessions is not possible, because it depends on when the private sector’s willingness to borrow from the banks—and the banking sector’s willingness to lend—stops. This can be delayed by government policy—as it was in Australia in 2008, via a strong government stimulus, the restarting of the housing bubble by a government grant to first home buyers, and the boom in investment and exports set off by China’s own stimulus program. But the day when credit growth stops can’t be put off indefinitely. When it arrives, these countries—many of which appeared to avoid the worst of the crisis in 2008—will join the world’s long list of walking wounded economies.

  1. April 3, 2016 at 1:46 pm

    Hi Steve, what about the UK?
    Also what are the chances of a debt jubilee by September?
    is it worth putiing helicopter money to carers of chronically ill people instead of disability benefits?

  2. April 3, 2016 at 1:47 pm

    Credit growth is exponential via the interest factor but not a smooth function because of the varying periods and interest rates of loans. Eventually an exponential function will “reach for the sky’, its slope (rate of change per unit time) being equal to its value. When this point is reached the system fails. Loan repayments coming due exceed money in the economy and there must be foreclosures on loans. That is when retrenchment happens and it is caused by our dysfunctional monetary system as you described it in your Forbes article. It is called by many names, recession, depression, liquidation phase, etc. And it happens over and over again. When will we ever learn….????

  3. April 3, 2016 at 5:34 pm

    Steve, I understand your position to be “demand = GDP plus growth in borrowing”. But you must surely give some credence to the concept of deferred consumption. So should your position not be “demand = GDP plus growth in borrowing minus growth in saving”?

  4. April 3, 2016 at 9:38 pm

    I don’t think Steve replies to comments on this blog…

    My take on this is that people who are in debt are not the ones who have substantial savings. It is the ability of debtors to service their debts that seems to be the primary risk factor.

    • April 3, 2016 at 9:45 pm

      Thanks Michael. My point is that Keen’s central idea, that demand is equal to GDP plus change in debt seems to ignore savings, which by definition constitute demand deferred. If you accept that, the next step is to point out that, for growth periods, growth in borrowing is equal to growth in saving, so Keen’s dictum collapses to “demand = GDP”. On your other point: of course, those with debt are a different strata of society than those with savings.

      • April 4, 2016 at 7:12 am

        If, macroeconomically speaking, change in debt is always equal to change in savings, then in your formulation Expenditure=GDP+D-S you are counting debt twice, once as Debt (D) and once as Savings (S), which is negative debt (D=-S).

        In other words, total expenditure includes spending on GDP related items plus money ‘spent’ on non-GDP related items, including savings and equity. If my understanding is correct, Steve is right on the money.

        One could plausibly focus on changes in savings to GDP ratio and come up with very similar predictions to Steve’s.

      • April 4, 2016 at 10:37 am

        Interesting position Michael! If I leave £10k in my current account it would be a rather twisted interpretation to say that when I “purchased” this financial asset I have increased demand to that value. On the contrary I have stopped consuming to that extent. Or, if I buy shares to that value I have again deferred my own consumption, which the debt-issuer undertakes in my place. Putting it another way: I can spend £10k on a new car, thus employing metal-bashers, or the owner of my financial asset can spend it on the same, thus employing those people. In the transfer of purchasing power embodied in the purchase of that financial asset there is no increase in employment let alone a doubling of it.

        One might concede this however: “demand = GDP plus financial services profit” where that extra demand employs bankers rather than metal-bashers.

      • April 4, 2016 at 11:52 am

        If total savings in the economy are £10k, held in current accounts, the total debt in the same economy must be approximately £10k also. You can count one or the other but not both towards the aggregate expenditure, because when I spend money to you it is still the same transaction, the saver is just the recipient of someone else’s expenditure, which has originated as debt to the banking system. All new savings are ultimately new debt, and all new debt must become someone’s savings (possibly, but not exclusively, on account of financial services profit).

        By ‘purchasing’ and holding the financial asset (called ‘savings’) the savers have increased the demand for this financial asset while leaving the demand for goods and services unaffected. If savings are invested rather than held in current account they are still someone’s savings of the same amount but may also contribute to GDP-related expenditure, many time over, hence both are counted.

  5. April 4, 2016 at 8:16 am

    Borrowing has no necessary relationship to saving. A substantial fraction of loans are created by ‘keyboard money’ out of nothing. They do not require any infusion of exogenous money or any deferral of consumption.

    • April 4, 2016 at 10:22 am

      Avner, I understand your Endogenous Money position on this, which exposes a vital part of the money creation mechanism. However Endogenous Money will always be only half a theory if it does not get empirical and realise that Fed and BoE statistics show this identity: “borrowing = savings minus reserves” or “loan-engendered spending is always smaller than deferred consumption”.

      • April 4, 2016 at 10:52 am

        Steve’s point is my point. The Fed and BoE statistics are not ’empirical’. They are arbitrary and do not provide empirical info on private sector debt. That is why we need the BIS data that Steve has been using. By the way, Drehman and Juselius have already made this point very well in a 2012 article ‘Do debt service costs affect macroeconomic and
        financial stability?’ BIS Quarterly Review, September 2012. And Central Banks have embraced endogenous money theory. The latest concession is McLeay et al. ‘Money creation in the modern
        economy’ in Bank of England Quarterly Bulletin Q1, 2014.

      • April 4, 2016 at 11:56 am

        Avner, I am not sure why you say that the Fed and BoE “do not provide empirical info on private sector debt.” For example the Fed do in great detail at http://www.federalreserve.gov/releases/h8/current/default.htm, to the extent that you can even find out how much debt is secured on farms for example.

        Also the BIS state this “Like the LBS, the CBS are reported to the BIS at an aggregate (banking system) level rather than individual bank level. A central bank or another national authority collects data from internationally active banks in its jurisdiction, compiles national aggregates and then sends them to the BIS to calculate global aggregates.” (http://www.bis.org/publ/qtrpdf/r_qt1509e.htm) Also this: “Central bank cooperation: The BIS would like to offer a special thanks to member central banks for their consultations in this endeavour. The BIS has made every reasonable effort to ensure that the long series on credit are accurate, but no guarantees are made.” (http://www.bis.org/statistics/totcredit.htm)

        I am inclined to the view that the BIS gets its statistics from central banks. Do you have evidence to the contrary?

      • April 4, 2016 at 3:07 pm

        You are right about the Fed. Steve also says so. I am not aware that similar data were publicly available in the UK — I have tried to find them. The BoE may have data that it doesn’t share, and the BIS may have prompted central banks to make a special effort to find private debt data.

      • April 4, 2016 at 4:50 pm

        Avner, the BoE stats are a labyrinth, but everything is there and the staff are happy to help, though they can sometimes be a bit slow in answering queries. All the data you could ever want on UK lending is available here: http://www.bankofengland.co.uk/statistics/pages/bankstats/current/default.aspx. There is incredible detail should you want it for example in Table C1.2 detailing lending to all sectors of industry. Or, if you want lending to individuals try Tables A5.2, A5.3, A5.4, A5.6 etc.

        So, if you want to declare that either Fed or BoE statistics are “arbitrary and do not provide empirical info on private sector debt” I suggest that you have a bit more work to do!

      • April 4, 2016 at 7:02 pm

        I defer to your knowledge. Thank you and very good to know. My information was based on the discovery a few years ago in the course of research on flow of funds that the Office of National Statistics had lost about two decades of historical data in the course of transition to a new standard in 1997 (see my piece on

        Click to access FlowOfFundsReport20130522.pdf

        Partly in response to that, the B of E has posted a good deal of new statistical material which I haven’t looked at yet. The Office of National Statistics is also on to the job, and there is a good prospect of complete reconstruction. However those data provided only very broad sectoral data. I hope the data you refer to have long-term historical information.

      • April 4, 2016 at 8:35 pm

        Thanks Avner. It is quite true that the BoE has patchy data, for example on asset purchase and on other things, though it does offer some data not found at the Fed, for example gross lending figures. This means that for good empirical work one probably needs to work with half a dozen central banks and take an average picture.

        Your work on flow of funds looks interesting.

        Do have a look at my paper here where effectively I attempt to bridge Endogenous Money with Loanable Funds: http://www.jnani.org/money/papers/King%20-%20Why%20Positive%20Money%20is%20Wrong.pdf

      • April 6, 2016 at 8:07 am

        Mike — I look forward keenly to reading your paper. Before I read it, I wonder whether it takes into account differences in maturity, i.e. liquidity. I alluded to that in one of my posts. Perhaps go off-site to continue? You can find my email on the web quite easily. Avner

    • April 4, 2016 at 10:29 am

      That is true, but it does not negate the claim that deposits (savings) approximately equal liabilities (debt). I grant that the term ‘borrowing’ or ‘loan’ are, in the context of fractional reserve banking a misnomers, but they are used nonetheless.

      If the means of payment are created as credit, that is ‘keyboard money’, as most means of payment are, it becomes someone’s debt to the bank (say, a ‘home loan’) and a deposit (saving) for someone who is subsequently paid with this funds (say, the home seller). Debt (of home buyer) equal savings (of home seller), even if there is no perfect transfer of purchasing power from the original ‘lender’ to the ‘borrower’.

      • April 4, 2016 at 11:36 am

        Thanks. You are right. Every loan creates a deposit and vice versa in accounting terms. I think it is useful to distinguish the term ‘deposit’, which is a banking concept, from ‘saving’, which is an economic one. That is why we have two terms. A deposit is a liability for the bank. Saving is potential consumption, other things being equal. What might not be equal is liquidity, on either side. When the bank grants a loan, it creates new potential to consume, but no one is deprived of anything until consumption (or investment) takes place, and until debt service begins.

      • April 4, 2016 at 1:01 pm

        Good point. The liquidity discrepancy between various deposits certainly adds complexity to the system. But, no matter how the original liquidity of a ‘loan’ is spent or invested there always seems to be an equal amount of the same liquidity, traceable to the originating credit transaction, somewhere in the system. What I am getting at is that the original amount of credit/debt is always present in the system as someone’s savings (in the proper sense of deferred spending of the same liquidity).

        Between the originating credit transaction and the deferred spending of the same amount there may be many GDP-related transactions and secondary financial assets of reduced liquidity. For example, B takes a ‘home loan’ and buys a house from C; C invests the proceeds in shares of a public company D; D invests the same amount into property investment fund E; E pays a builder F who is constructing a new commercial development for E; F pays a junior carpenter G for his work on the development; G saves the money in cash for a surfing holiday in Indonesia. Since money is fungible, C may be able to sell the shares on short notice, irrespective of how and when G decides to spend his savings, in which case C will be able to defer the purchasing power obtained from the buyer of the shares.

        B has debt
        C has a non-savings financial asset
        D has liability to C and a non-savings financial asset
        E has liability to D and spends the money
        F spends the money
        G has savings.

    • April 4, 2016 at 8:47 pm

      As I thought. The data are all quite recent, so I was not wrong to think that they were not available. Their presence appears to be a recent innovation.

      • April 4, 2016 at 9:44 pm

        In the USA M2 is typically 4 times M1. M2 is TOTAL PRINCIPAL DEBT to BANKS. M1 is the only “money” available with which to make principal and interest payments. Thus the entire country is in the position of paying down a dollar of principal debt with only 25¢ available. The crisis has NOTHING to do with GDP and everything to do with the ratio of M2 to M1, a factor everyone including economists, and Keen in particular, seem absolutely determined to ignore.

        “Keen: “when I have some spare time I’ll take on your misconceptions in a genuine mathematical critique of your silly arithmetic… the only question for me is whether I can be bothered wasting the time to prove you wrong. At the moment, the answer to that question is no. When it’s yes, I’ll write a blog post on the topic. Until then I can’t be bothered reading any more emails from you.”

        It has been 7 years since I first issued my challenge to professor Keen and countless others to refute my facts, logic or arithmetic. So far no no one has. They just ignore the irrefutable facts and arithmetic and continue to ignore what they can’t explain away.

      • April 5, 2016 at 4:59 pm

        Avner, you are too hasty! I hate to say this, but as someone originally trained in physics I do find economists tend to prioritise hypothesis over research.

        The fact is that BoE statistics go way back, you just have to look at other parts of their statistical database. For example series LPQVQKQ, “Quarterly amounts outstanding of M4 lending (historical measure) (monetary financial institutions’ sterling net lending to private sector) (in sterling millions) not seasonally adjusted (*1)” This particular series goes back to 1963. Further back to that the BoE keeps scans of its paper records and you can extract what you need from those, though it would be long hard work, admittedly.

        LPQVQKQ is here: http://www.bankofengland.co.uk/boeapps/iadb/fromshowcolumns.asp?Travel=NIxSCx&ShadowPage=1&SearchText=LPQVQKQ&SearchExclude=&SearchTextFields=TC&Thes=&SearchType=&Cats=&ActualResNumPerPage=&TotalNumResults=1&XNotes=Y&C=1MC&ShowData.x=55&ShowData.y=8&XNotes2=Y

      • April 5, 2016 at 6:21 pm

        In the spirit of friendly banter, I concede the point about economists and hypotheses. Indeed, I am about to publish a book (in the autumn) devoted to making similar points. But your evidence does not quite settle the matter — they may well have had the data for a long time, but we don’t know when they put them up. To the extent that these data were part of flow of funds, the electronic versions between 1963 and 1987 were lost for at least ten years, and have been partly recovered recently. As I indicated, I have been involved somewhat with the Bank over retrieving and restoring these data, and they were making quite a song and dance about their new historical statistics website.

      • April 5, 2016 at 6:59 pm

        I appreciate the spirit of friendly banter and concede that you are party to information that I am not. I only started working with BoE statistics in 2012, so I had no idea that for some time prior to that great chunks were missing. Thanks for this!

  6. April 4, 2016 at 8:05 pm

    Great analysis. In Canada the prevailing wisdom is that that public debt to GDP ratios are what count not the private debt to GDP ratio. This one does not appear in our budgets.

  7. April 4, 2016 at 9:32 pm

    Regarding Graziani’s framework on money as a third party promise, given that money is created as debt isn’t interest a form of seignorage for the bank? It goes like this:

    – Customer takes a mortgage, say $100,000. This money is freshly created with the loan.
    – Customer pays off $130,000 after 15 years, of which $100,000 gets destroyed to extinguish the loan and $30,000 is transferred from the economy to the bank.

    Where did the $30,000 come from? It can only have been created as loans to other people, and the bank is the first agent who gets to use it for free. If the bank creates the money and uses it first for free, isn’t that seignorage?

    • April 4, 2016 at 9:51 pm

      The bank spends the interest in each payment cycle. The original $100,000 is sufficient to pay off both the $100,000 principal and the $30,000 interest. In fact a loan of ONE dollar with $99 of interest, TOTAL $100 can be paid off with the single created dollar provided that each payment is only one dollar and the lender spends it 99 of the 100 times it is paid.

    • April 5, 2016 at 7:01 pm

      Unfortunately the data on interest charged by banks on loans and interest paid by banks on deposits supports both Loanable Funds and Endogenous Money.

      • April 5, 2016 at 7:45 pm

        The analysis I have seen and understand labels the interest collected by commercial banks as “….interest on the seignorage…” Pavlos is correct. Banks must loan more each unit of time to provide the interest they will collect. The math is simple. It is an exponential function and at some point in time, determined by the real interest rate, it “reaches for the sky” and the boom results in a crash when the banks cannot loan enough to pay off existing loans. Then we have foreclosures for the very obvious reason that insufficient money is available to pay all loans. That is the “bust” and it is built into the system.

  8. April 5, 2016 at 3:31 am

    April 4, 2016 at 9:32 pm Reply
    Regarding Graziani’s framework on money as a third party promise, given that money is created as debt isn’t interest a form of seignorage for the bank?
    Not only that it is a seignorage on “counterfeit” money.
    IT is not new wealh, it is someone elses wealth being printed out of thin air.
    This is the reason why every boom must bust.
    Take your numbers:
    For a 20 year period of say housing growth

    …$1trillion to $100trillion….Apx. $150 trillion in housing
    80% finance at 6% with avg. 10% down= after 12 years the balance owed to the banks would be more than double $112 trillion =over $224 trillion still owed.
    HOW DO YOU PAY FOR the final year of the bubble for the $32 trillion in NEW construction?

    The timing factor should be when the debt payment required exceeds the availability of “real money” needed to make payment on the bubble. When the “credit money” has driven the “good money” out of circulation-BUST. (Grahams Law).
    BTW-What do you think would happen if all money were digital?
    Justaluckyfool would start to save coins.

    RE; comment by Steve Keen, ” I restricted myself to the 16 countries that had a private debt to GDP ratio exceeding 175% of GDP.”
    IMHO, as per SODDY that alone is not the factor, the factor should be “the interest rate of the private debt” versus “the labor rate of the GDP”.

  9. April 5, 2016 at 9:07 pm

    Neither Pavlos nor Charles 3000 comprehend the difference between stock and flow. Yes the “bust” is built into the system but interest is NOT the culprit. The culprit is MULTIPLE PRINCIPAL DEBTS of the SAME MONEY created by the design of banking itself.


    • April 6, 2016 at 2:02 pm

      I tried your link but it did not work. Is there math that shows the conclusion you state? I can show, mathematically that interest does that. Is “MULTIPLE PRINCIPAL DEBTS of the SAME MONEY” a reference to fractional reserve banking?

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.