Saturday, 25 October 2014

Financial hurricanes

I recently wrote a post on Pieria about the (sizeable) role of the Eurodollar market in the 2007-8 financial crisis, or rather in the credit boom that led to the crisis. In order to explain the vast increase in transatlantic two-way flows (round-trips) from 1997 to 2007, I used this diagram from Hyung Song Shin's 2011 paper on the "global banking glut":

But there's a problem. This diagram appears to show that all money lent by both US and European banks came from US households. Where did they get their money? It wasn't wages. We know that US wages have been stagnating for two decades. No doubt some will suggest it came from Asian savers - the "global saving glut" that is apparently caused by thrifty Chinese households. But how could thrifty Chinese households provide US households with money? Most Chinese households save in yuan. It is Chinese government and corporations that lend to the US, and it is not households they lend to. Mainly, it is the US government. No, the "Asian savers" argument won't wash either. Where US households obtained the extraordinary amounts of money that crossed the Atlantic in the decade before the financial crisis is a mystery. It grew on trees, perhaps?

No it didn't. It came from banks. To show this, I've produced a diagram of my own:

Shin's diagram is a "loanable funds" model, whereas mine is endogenous money. Note the two-way flows. Property (and derived securities) flow clockwise: money flows anti-clockwise. The red lines are the two regulatory boundaries: US (Basel I) and Europe (Basel II). The gap between them contained not only the Atlantic, but most of the shadow banking sector.

In my diagram the flows can start anywhere. All actors in this diagram should properly be regarded as part of the "pump" that maintained the two-way flow. In my earlier post I suggested that the principal driver of the pump was regulatory arbitrage by banks between the Basel I and Basel II regimes, with the shadow banking sector and money market funds facilitating. But the desire of borrowers for loans and households & corporations (particularly construction companies) for property sales are important too. Without these, the US banks would have lent much less and the European banks would have had much less funding.

However, although this is a circular flow, the quantities of money and property/securities circulating are not constant. As the money flows around this model, it expands: and correspondingly, property and securities also expand.

Money was created by US banks when they lent. The "Loans" arrow at the left hand side is always new money. Similarly, money was created by European banks when they purchased securities. The "Payment" arrow at the top right hand side is also new money.

Lending against property created not only new money, but also new securities. The loans were sold by US banks to securitizers, who pooled them, tranched them and created new securities, which they sold into the capital markets. In the "shadow banking" box is not only the securitization process itself, but also rehypothecation and leveraging of the ensuing RMBS products. The value of the securities that came out of the other side of that box was far higher than the value of the loans that went into it. The "Payment" arrow was enlarged both by the ability of European banks to create new money AND the ability of the shadow banking sector to leverage up property-backed securities.*

So the Eurodollar transatlantic flows grew enormously in the decade before the financial crisis, both money and property-backed securities. US lending against property expanded enormously, and there was as a result a construction bubble - the "Property" arrow at the bottom left-hand side itself expanded. But it expanded less than the other arrows. Most of the expansion of money was absorbed in price rises.

Normally, we would expect such a massive expansion of credit to have caused consumer price inflation to rise significantly. But it didn't. This period is known as the "Great Moderation", when consumer price inflation remained low and stable. For me, this is where the "Asian savers" theory gains traction. But the "savers" were not thrifty Chinese households. They were Asian governments, scarred by the Asian financial crisis of 1997 and anxious to protect themselves from damaging flows of capital by building up foreign exchange reserves. Instead of borrowing from Western countries, as they have done in the past - and suffered for it - they exported to them. I have observed before that the "Great Moderation" in the West is paralleled by the "Great Expansion" in the East. I would go further: I think it was caused by it. The massive expansion of low-cost imports to the West depressed consumer price inflation even though spending was increasing as a result of the credit bubble. I would venture to suggest that had there not been a credit bubble, this would actually have been a period of deflation. The stagnation of wages in many Western countries supports this line of argument.

Having said that, it is also clear that spending didn't increase as much as might have been expected. Spending would be a "leak" from the two-way flow in the diagram, reducing the flow and dampening the leverage. But most of the money seems to have stayed within the flow, driving up house prices to an unsustainable level.

Toxic cross-border feedback loops come in many kinds. This was a property-based one. The yen carry trade was a currency-based one, and once again, when it abruptly reversed the consequences were disastrous - although I don't think it was the sole cause of the financial crisis, it was certainly a significant contributor. Capital flows into Eurozone periphery countries prior to 2012 were a similar leveraging cross-border flow system that failed disastrously. So were the capital flows into Asian economies that reversed abruptly in 1997, causing the Asian crisis: I think QE-driven capital flows into emerging markets exhibit similar characteristics, and policy makers should be on their guard. Nor are these flow systems a new phenomenon: going back further, the Latin American debt crisis of the 1980s was the disastrous consequence of the failure of an enormous oil-based leveraging cross-border flow system.

It is difficult to see how conventional monetary policy could have dampened this leveraging cross-border flow without causing outright deflation in domestic consumer prices. And I would add that attempting to interrupt the flow once the system was fully developed - for example, by imposing capital controls - would have been dangerous. All the players in this game depended on the flow continuing. When it was suddenly interrupted, first in 2007 with the freezing of the ABCP market (top RH corner "Funding" and "Securities (collateral) arrows) and secondly in 2008 with the fall of Lehman (collapse of "shadow banking" box) followed by Reserve Primary breaking the buck (failure of "money market funds" box), the results were disastrous. All the players suffered, and many of them failed.

Policy makers need to watch for the formation of cross-border leveraging flow systems and act EARLY to dampen them. This requires international cooperation, of course: it is depressing that so far the only central banker who seems to have recognised the importance of global monetary policy coordination is India's Raghuram Rajan. It is also unfortunate that there is still no internationally-agreed approach to regulation of the financial system. But worst of all, fiscal authorities don't seem to see any need at all for coordination of national objectives. These flow systems are driven as much by competitive and nationalistic fiscal policy as they are by uncoordinated monetary policy and regulatory arbitrage.

Cross-border leveraging flow systems of the kind I have illustrated above are the storms and hurricanes of the financial system. We ignore them at our peril.

Related reading:

When the Nile floods fail
European banks and the global banking glut - Pieria
Currency wars and the fall of empires - Pieria

* I should emphasise that this diagram ONLY shows the Eurodollar flow. European banks were not the only purchasers of US RMBS and derivative products.


  1. I feel that leverage is now fully transportable on a global basis and will remain so without huge protectionism, which would threaten trade efficiencies too. Money will always leak across borders and with it any leverage. Control the normal transmission methods and new ones will spring up to replace them and take the available margin. So in general the complaint must be about leverage in general where I fully agree that leverage is the cause of all crashes. So then the old debate reignites about who decides how to allocate leverage in society and the associated arguments about bank lending, whether they should be socially responsible and lend more, or socially responsible and not lend so much they blow up and lose the depositors money. No answers I'm afraid Francis, but thanks for pointing out the transmissions of old, but sticking a finger in the observed breach in the dyke won't prevent others!

    Oh and re the point of deflation being a probable side effect if the credit bubble hadn't happened, due to wage growth staying low. Perhaps that wage growth was kept low by the very growth in cheaper Far East supply that the US leverage drove. Perhaps no credit bubble, no importation of deflation, No lower wage pressure. I dunno.

    1. Indeed, the offshoring jobs/wage stagnation/cheap imports/credit bubble is a feedback loop all of its own whose effects we don't understand and which in my view is still running.

  2. And Sorry.. Was thinking to much to add.. 'and thanks again for another great post!'

  3. Some points.

    "Money was created by US banks when they lent."

    And destroyed when the sold the loans to the shadow banks. As to whether purchase by the European banks creates money, I don't think that eurodollar claims are counted in current US money supply statistics. Shares in retail MMMFs would be.

    "The value of the securities that came out of the other side of that box was far higher than the value of the loans that went into it."

    Not sure what you mean, here. If the shadow bank box buys $100 of loans, it can't issue $200 of securities to the European banks box. It can issue $100 of securities to another shadow bank, which then issues $100 to the Eurpoean banks box, making $200 in total, but that's not leveraging.

    "Spending would be a "leak" from the two-way flow in the diagram..."

    The spending isn't really a leakage. The dollars just end up somewhere else. Like in the big cash surpluses that many corporates have been accumulating. Which in itself is part of the explanation of what's going on, because whilst households may just deposit their money with US banks, large corporate investors are more likely to want secured money-market instruments.

    1. Nick,

      1) Selling loans destroys money within the boundary and creates it outside the boundary, since for every sale there is a purchase. The effect is neutral. Therefore the INITIAL creation of money when the loan is granted remains. However, the money will disappear from domestic monetary aggregates.

      One of the biggest and most dangerous fallacies of our time is the belief that because something doesn't appear in an official statistic, therefore it doesn't exist. European banks create US dollars when they lend US dollars. That these dollars may not appear in monetary aggregates at the time is irrelevant. The US recognises the money created by European banks as soon as it crosses into the US as a funding flow to US banks. Central bank officials then scratch their heads and wonder where all this money is coming from.

      2) I deliberately have not in those post gone into the process by which property-backed securities were leveraged up in the shadow banking system, as it is complex and this post was already long enough. However, it has been well documented by others elsewhere.

      3) Spending is a leak from THIS PARTICULAR flow system. See my footnote. However, some of it would undoubtedly come back into it via large corporate deposits in MMMFs. Bizarrely, Shin did not include corporations at all in his model - I added them for precisely this reason.

    2. I would certainly agree that monetary aggregates are unsuitable measures for looking at what's going on here. Much of this process relates to the creation of deposit substitutes: repo or ABCP in the place of unsecured bank deposits. Should we treat these as money? In my view that question doesn't add much to the understanding of these developments. But if we want to deploy the "banks create money" meme, we probably need some idea of what we mean by money.

      There has indeed been stuff written on the leveraging of ABS and I have no problem with the idea that it can be leveraged. My concern here is an idea that I sometimes hear that rehypothecation of collateral allows a greater amount of end-user funding. It doesn't. You can certainly get a greater gross asset volume. But it doesn't change the amount of the end investment and the end borrowing. It works like a chain. You can add as many links to a chain as you want, but it will still only have two ends. I'm not saying that is what you meant, but I've heard this view from people who really should know better and it's quite wrong.

    3. For the purposes of this model I would regard highly liquid dollar-denominated tradeables created by the shadow banking system and by foreign banks as "money". They are to all intents and purposes indistinguishable from bank deposits apart from their lack of FDIC insurance.

      A rehypothecation chain increases the leverage of all intermediaries involved in the chain. The net output is the same as the input, but the intermediary balance sheets gross up. You can regard this as an illusory effect - in many ways it is - but when a chain unwinds, the losses for intermediaries can be considerable.

    4. Let's work on the capital requirement.

      I hope we agree that with a 100% capital requirement commercial banks can’t increase “money”.

      If the capital requirement is 100% for the central bank, doesn’t the same idea apply?

      I don’t believe redeemable and owed have to do with a 100% capital requirement.

      Assume the capital requirement is 10% for the commercial banks and no deposit insurance.

      Assets = 100. Liabilities (demand deposits) = 90. Equity = 10.

      Assets drop to 81. What most likely happens?

      Assume the central bank does all of the loans and there is only currency. Its capital requirement is 10%.

      Assets = 100. Liabilities (currency) = 90. Equity = 10.

      Assets drop to 81. What most likely happens?

    5. I don't believe you can equate private commercial banks with central banks (which are an arm of government) in this way.

      Also, this comment is off topic. Please confine your comments to the subject of the post. If you wish to discuss something else, email me.

    6. "Assume the capital requirement is 10% for the commercial banks and no deposit insurance."

      Why isn't that on topic?

      Or, assume the capital requirement is 10% for the shadow banks and no deposit insurance.

    7. Not on topic for this post, sorry.

  4. Excellent and interesting analysis. I am pretty sure governments monitor financial transaction systems - how well is anyone's guess. But the problem here would be identifying systematic loops in a welter of seemingly random transactions. When analysing a system one does not know much about some hypotheses might be necessary - does this set of transactions look fishy - is it likely to be some kind of loop or part of a loop - are some dubious types involved? All sounds a lot like the sort of thing the spies do, but the spies spend the thick end of £2Bn/y, the FSA around £500M/y.

  5. It seems to me like a different dynamic may be at play, different from what you describe here. Please let me explain.

    I will use the example of China who holds much American debt

    How did China get that debt? China sold to America. China sold goods of all kinds and qualities. China sold much more to America than it bought in return. These American sales of products resulted in a large holding of American dollars.

    Now, American dollars do not trade well in China. In fact, all the goods sent to America (but made in China) were made with Chinese labor paid in yen. It took a lot of yen to build/make all the goods that were ultimately sold in America.

    Now here is the question. Here is the problem. How does a Chinese manufacturer pay yen when sales are in dollars? How can this happen at the same time that China is building a large wealth of American bonds?

    The only answer that I can see is that China is accepting American debt (in dollars) as collateral for China debt (in yen). China seems to be comfortable in exporting Chinese labor and materials and receiving-in-return American debt. And it must be with the full acceptance of Chinese banks and the Chinese government.

    This description makes even more appealing if we think of Chinese workers who build products to be sold in America as not being Chinese workers. Instead, they are American workers who spend their productive hours working for America and have the good fortune (or not) of living in China.

    Thank you for allowing me to explain my concept of International Trade. You can see that it involves a different perspective from the one you outlined.

    1. Actually it is very similar to the one I outlined. China invested in manufacturing and assembly facilities to produce goods that American people wanted to buy. It paid its workers far less than equivalent workers in America, which encouraged companies to outsource goods production to China. The resulting inflows of cheap goods to America put downwards pressure on inflation - this is why I say the Great Moderation was caused by the Great Expansion in China and its satellites. The inflows of dollars to China enabled it to build up reserves (in the form of USTs, which are a guaranteed claim on dollars) which it uses to anchor its currency.

      This story does not in any way invalidate what I describe in this post. American workers whose wages were stagnating were encouraged by the US Government to take on debt, particularly mortgage debt, in the expectation that their increased spending as a result would restore growth and hence improve their future incomes. Borrowing to spend more became almost a patriotic duty after 9/11. The FOMC minutes reveal that the FOMC was pleased with the combination of wage moderation and mortgage growth, believing that it would result in strong growth without inflation. They appear to have been totally blind to the risks.

      Chinese currency is the yuan or renminbi. Yen are Japanese.

  6. European banks borrowed large amounts of US dollars through the money markets and invested them in US asset-backed securities via the US's shadow banking system.

  7. Your article is a good hard read. So is part of the problem not really people borrowing in foreign currencies like the US dollar in a non-fixed rate exchange world ?!
    Maybe we need to move to using the SDR ?

  8. Hi Frances,

    You say that it was surprising that CPI did not go up with this huge expansion of money supply, but isn't it true that we had huge inflation in property prices? That's where the inflation happened because that's where the money went. And governments and CBs are now desperate to try to avoid deflation in housing as the money supply (debt) shrinks.

    And CPI, of course, does not include housing.