Brad Setser

Follow the Money

Cross border flows, with a bit of macroeconomics

Print Print Cite Cite
Style: MLA APA Chicago Close


One graph to rule them all …

by Brad Setser
July 1, 2009

If I had to pick a single graph to explain the evolution of the United States’ balance of payments – and thus, indirectly, the entire story of the world’s macroeconomic “imbalances” – this would be it.


All data is in dollar billions, and is presented as a rolling four quarter sum.*

The red line is the United States current account deficit.

The black line is the United States financing need – defined as the sum of the current account deficit plus US outward FDI and US purchases of foreign long-term securities.** The dip in the total US financing need from mid 2005 to mid 2006 isn’t real. It reflects the impact of the Homeland Investment Act, a holiday on the repatriation of the foreign profits of US multinationals that produced a sharp fall in outward FDI.*** The rise in the United States financing need over the course of 2007 by contrast is real; American investors bought the decoupling story and wanted to invest more abroad.

The shaded area represents official demand for US assets. The inflows from central banks that report data to the IMF and Norway are known. The inflows from central banks that don’t report and other sovereign funds are my own estimates. The key countries that do not report reserves are – in my judgment – China, Saudi Arabia and the other countries in the GCC. I have assumed that the dollar share of their reserves is closer to 70% than 60% (supporting evidence). I by contrast have assumed that the GCC’s sovereign funds have a diverse portfolio.

What does the graph tell us?

In my view, three things:

First, the rise in the US current account deficit from 2002 to 2006 is associated with a rise in official demand for US assets. The quarterly IMF data doesn’t extend back to the late 90s – or to the early 1980s. But trust me, that is a change from past periods when the US current account deficit expanded. To be sure, private investors abroad were also buying US assets. But the rise in the overall US financing need associated with the rise in the current account deficit wasn’t financed by a comparable rise in private demand for US assets.

Second, Official demand for US assets soared from the end of 2005 to the end of 2007 – even by the standards of this decade. That surge was hidden, as the majority came from countries that don’t transparently report the currency composition of their reserves (let alone their sovereign funds). But it happened. At the peak of official asset accumulation in late 2007 and early 2008, official demand for US assets (best I can tell, and if my estimates are off – do tell, and explain, with data) exceeded the total US financing need. Central banks and sovereign funds were financing both the US current account deficit and the “diversification” of private US portfolios. Absent that uptick in official demand, the US would have experienced a dollar crisis before it experienced a banking crisis.

Third, official demand for US assets has fallen quite sharply over the last four quarters. Paul Krugman is right. The US depended on China’s central bank – and other official investors – far less over the last four quarters than it did in late 2007 and early 2008. The collapse in gross private capital flows during the crisis rebounded in the United States’ favor, as Americans scaled back their investment abroad faster than foreigners scaled back their investment in the US. The fiscal deficit may be up, but US “dependence” on central banks for financing fell sharply.

At least through the first quarter. The second quarter of 2009 will be to be different. Reserve growth seems to have resumed.

But there is little doubt that reserve growth slowed sharply from mid 2008 to mid 2009. The countries that report detailed data on their reserves to the IMF reduced their dollar reserves from the end of q1 2008 to the end of q1 2009 by about $130 billion. That is a fact not a guess. Guessing the overall total requires guessing what China (and Saudi Arabia did). But there is little doubt that China’s overall reserve growth slowed – so unless it was diversifying into the dollar, its growth in its dollar reserves must also have slowed.

Wait. Doesn’t the US data (including the Fed custodial accounts) tell a somewhat different story? And doesn’t the US data show record inflows into Treasuries?

All true.

My estimates of dollar reserve growth (and the available data from the IMF) are at odds with the US data showing a pick up in official demand for Treasuries.


What gives?

My explanation: central banks reserve managers pulled funds from banks, the US agencies, and private fund managers – producing an uptick in demand for Treasuries.

We know that this happened inside the US. From the end of q1 08 to the end of q1 09 central banks reduced their dollar deposits in US banks by about $200 billion – providing a lot of the funds that flowed into Treasuries. I would bet the same thing happened globally.

From early 2007 to mid 2008, central bank demand – in even the revised US data – lagged my estimated of global dollar reserve growth. From mid 2008 on, central bank demand in the US data has exceeded my estimate for dolllar reserve growth.


I don’t think central banks shifted out of the dollar as rapidly as the US data implies in late 2007 and early 2008 (the IMF data rules out a shift by those countries that report detailed data to the IMF, so such a shift would have to have come from China and the Gulf). And I don’t think central banks shifted back into dollars on the scale the US data now implies. Rather central banks shifted first into riskier dollar assets (and made greater use of private managers) and then shifted back to traditional reserve assets like Treasuries.

That does though mean that there has been an enormous gap between the rhetoric coming from some important holders of reserves — rhetoric that suggests a loss of confidence in the long-term value of US Treasuries — and their actual actions. For all their faults, central bank reserve managers still seem to consider US Treasuries safer than other dollar-denominated assets.

*The underlying data comes from the IMF’s COFER data, the US BEA, the PBoC, SAMA and the national balance of payments of countries with sovereign wealth funds. But I also had to make a few assumptions to flesh out the data, notably assumptions about the dollar share of countries that do not report detailed data about their reserves to the IMF
** I have excluded outward short-term flows because they are highly correlated with short-term inflows; the banking sector doesn’t seem to be a consistent source of net financing for the US.
*** The homeland investment act allowed American firms to repatriate profits earned abroad without paying US corporate income taxes. It led to a $200 billion or so fall in US outward FDI, as American firms (temporarily) stopped reinvesting their foreign profits.


  • Posted by But What do I Know?

    Great heading–redolent of doom. . .

    If I can sum this up–the demand for dollars around the world has dropped off dramatically in the past six to nine months. Why hasn’t this crushed the dollar? I’ll never understand FX. . .

  • Posted by bsetser

    central bank demand for dollars fell b/c private demand for dollars in the crisis soared, as a host of private players all over the world who borrowed in dollars to finance a range of projects (or borrowed short to fund a long-term portfolio) had to scramble to find dollars to repay their debts. i.e. the contraction of private flows (see my posts on financial deglobalization) was dollar positive.

  • Posted by bsetser

    p.s. all the data is lagged; the latest COFER data (which inspired the post)/ latest BoP data are from end of q1. central bank dollar demand reappeared in q2, especially in may at the height of the risk/ green shoots rally.

  • Posted by JKH

    “Central banks and sovereign funds were financing both the US current account deficit and the “diversification” of private US portfolios.”

    Alternatively, the US current and capital accounts were both sources of funds for foreign central bank and sovereign fund asset accumulation and liquidity management. The US has been the proactive risk taker in this story – in terms of running a current account deficit as well as investing in higher risk assets abroad. Foreign central banks have been the reactive risk absorbers – in terms of recycling dollars from both of these sources into low risk assets. Among other things, this allowed them to manage the risk of US liquidity withdrawal that their countries have actually experienced recently. Even though their reserves are excessive, their strategy still served its purpose in this case. They could more handily meet the liquidity pressure that resulted in the repatriation of US foreign investment during the credit crisis. This all suggests to me an overall causality whereby a proactive risk taker (the US) puts a reactive risk manager (the CBs) in defensive mode, in that they have responded to and deployed original US flows (current and capital) in low risk portfolios.

  • Posted by bsetser

    JKH — I would argue that EM central banks that are intervening to keep their currencies from rising despite large current account surplus are active players, even active risk takers — as their resistance to currency appreciate contributes to the US CAD.

    I take your point with respect to capital outflows from the US to EMs that are managing their currencies; absent the inflows, they wouldn’t build up reserves (india is a good example, brazil too). But in such cases the global equilibrium that existed required their ongoing willingness to build up reserves in response to us outflows. OTherwise there would have been pressure on the us to run a current account surplus to finance America’s desire for foreign assets — and pressure for countries with balanced current account that were attracting large private inflows to start running current account deficits.

  • Posted by jonathan

    Why did the US financing line start to dive in 2007? Is that appreciation?

  • Posted by Indian Investor

    Dr. Setser, could you please expand on your formula for calculating the Us financing need, i.e. financing need = current account deficit + outward FDI + US purchases of long term foreign securities. I’m not clear why these items create a financing need, and why certain other items, such as fiscal deficit, aren’t added in here. When Fallow’s hourney of the dollar is complete, shouldn’t the $30 paid to Chinese exporters end up back with the US Treasury?

  • Posted by Indian Investor

    Sorry I meant that in Fallow’s journey of the dollar, the $3 that was the actual amount paid to Chinese exporters for the $30 electric toothbrush from CVS – should end up back with the US Treasury, completing his example. Yet you say that the current account deficit creates a ‘financing need’ for the US. How is that possible?
    Despite getting the $3 back, if the US Treasury incurs a fiscal deficit …

  • Posted by q

    typo, i think

    > The countries that report detailed data on their reserves to the IMF reduced their dollar reserves from the end of q1 2008 to the end of q1 2008 by about $130 billion.

    second one should be 2009?

  • Posted by JKH

    “EM central banks that are intervening to keep their currencies from rising despite large current account surplus are active players, even active risk takers”

    Yes, although it’s an unusual type of risk taking. They’re doing it as a likely cost in subsidizing their exports. They’re not doing it in an attempt to make money from a US dollar long position. I would characterize it as passive risk absorption in that sense. They (over) compensate for this with active risk management through their choice of less risky credits such as treasuries.

  • Posted by bsetser

    q — thanks for picking up the type-o. I’ll fix it.

    jkh — well, actively managing your currency by building up unneeded reserves to support exports strikes me as an active policy choice. and for a while central banks did try to offset currency losses by seeking higher returns on their “forced” (b/c of the exchange rate) USD longs. they just got burned by buying risk assets in 07 and early 08 …

    jonathan — us l-term outflows fell after the q3 07 market blowup. and in q4 08 they turned negative, as the us sold its foreign portfolio. that enters into the rolling 4q sum over time.

    indian investor — a fiscal deficit doesn’t necessarily give rise to an external deficit (See japan or germany), so the US need for financing from the rest of the world is the trade deficit + net US demand for foreign assets. I left out s-term outflows b/c they are typically offset by s-term inflows.

    adding the fiscal deficit — an internal concept — the current account deficit would be double counting in a sense. The external financing need is a BoP concept, so the relevant concept is the trade/ current account deficit and us demand for foreign assets.

  • Posted by Indian Investor

    Dr. Setser, thanks for your earlier reply but I’d like you to reflect more deeply on my question. Answering it can easily make you eligible for a Nobel Prize in economics. Your equation basically is:
    capital inflows = imports – exports + capital outflows.
    Have I understood your equation correctly ?
    This equation would definitely hold in a gold standard world. But since August 15, 1971; the balance of payments identity doesn’t hold anymore.
    What the modern world holds as capital today is legal tender currencies that can’t be enforced for payment of all public and private debts outside the issuing sovereign’s territories. In a gold standard world capital outflows can be matched against flows of goods in the same equation, because the flow of capital was nothing other than the flow of gold,conceptually represented by a hard currency.
    In a legal tender world capital flows and goods flows can’t be considered together. Theoretically the exchange rate is the only mechanism available to balance the capital account for mismatched current account flows.
    The US can only consume more than it produces if there is a net inflow of capital to it. That’s only true in a gold standard world. In the modern world the US consumes more than it produces due to the overvalued exchange rate of the USD against major world currencies, not because of a physical inflow of capital.
    A dollar pegged to gold at 35 dollars to an ounce can be held indefinitely by foreign central banks against future US exports. The modern dollar can’t. Instead, the only possible adjustment is a major devaluation of the USD exchange rate.
    Reflecting on this more deeply will also tell you more about why the USD exchange rate is so high.Finally you will appreciate how the current international monetary system is set up.

  • Posted by don

    “Absent that uptick in official demand, the US would have experienced a dollar crisis before it experienced a banking crisis.”
    Absent that uptick in official demand, foreign currencies would have been higher, U.S. import demand would have been lower, demand for U.S. exports would have been higher, and the ‘need’ for foreign financing would have been lower. I’m not sure any ‘crisis’ would have ensued, at least for the U.S. If foreign official demand for dollars crumpled today, the U.S. current account would show dramatic improvement, much to the benefit of anemic domestic AD.

  • Posted by FollowTheMoney

    anyone if insight if China purchasing TIPS recently or last 12 months?

  • Posted by RodgerRafter

    “…one graph to bring them all, and in the darkness bind them…”

    I love the analogy and the graph. Personally, I like the first one you posted Sunday a little better. I think that one has even more powerful extensions into everything macro-economic these days.

    Thanks for a series of especially good posts this week.

  • Posted by jonathan

    I was misreading the graph. Actually missed the word “long-term” in the text and went bad from there. Thanks for the clarification.

  • Posted by yoda

    dollar and treasury in big trouble of crashing down. world of pain for dollar and treasury holders.

  • Posted by yoda

    use recent dollar & treasury rebounce to sell. FED’s targeting 0% fed fund rate & quantitative easy will last for a long long long long time. all bullish to equity stock, real estate, commodity, and gold, cheers!!! 🙂

  • Posted by Rien Huizer


    What would make the black line move away from the red line again?

    I would presume that this is the sum of FDI and portfolio investment. FDI may have undergone structural change along with structural changes in the way US firms do business overseas and the contraction of the financial services industry, and portfolio investment may have been driven by apparently better opportunities in the US (??), or, more likely, portfolio liquidation associated with the crash.

    The phenomenon that if you imagine the US as a unitary actor (‘George’), George during the pst 7 or so years became a hedge fund financed marginally by China and a few others, investing (marginlly again) in risky assets (meanwhile repatriating safe asstes!) . We seem now to have reached a stage where George’s outstandings are no longer growing (see also your earlier articles on the NIIP). What will George do next? What will George’s absence from the world economy casino mean for the gambling business? Yet another sign that a worldwide recovery will be very difficult. Someone has to put a few chips on the table (if you have a Schumpeterian view of economic history) and the others are to scared, poor or bureaucratic to do this.

    It looks like we missed a great opportunity with the SWF craze. A couple of trillions in a trigger-happy SWF is just what we need.

  • Posted by Rien Huizer

    Perhaps I will get my SWF, since it looks like the IMF is finally going into the hedge fund business: recycling EM surpluses to fund the former Comecon.. But not with the right mindset and at the wrong price. What a waste.

  • Posted by Rien Huizer
  • Posted by bsetser

    Rien — cynical, but not totally inaccurate, take on the sdr denominated bond issue …

    incidentally, my guess is that the black and red lines diverged in q2 09, as there are a host of indications that the US rediscovred its appetite for EM risk – -and that EM risks resumed reserve growth.

  • Posted by Indian Investor

    Dr. Setser, though I’m back to the same point, let me try to illustrate this complex issue with an approximate analogy.
    Suppose a new rule is enacted in the US that all rent payments, and all payments for home loan EMIs need to be paid exclusively in physical 20-dollar bills. What would this rule do to the demand for $20 bills? Since rent and mortgage EMIs are major expense items for ordinary Americans,savers would have a preference for stashes of $20 bills over all other denominations, bank accounts, etc. Imagine that this rule stays in force for 37 years. Due to the exorbitant demand for $20 bills, they would be increasingly exchanged for higher denominations. Also, many transactions other than rentals could also be required to be done in $20 bills, due to the interplay of market forces. If somebody started at the beginning of the rule enforcement with a large existing stash of $20 bills, or if they had an exclusive right to issue $20 bills, they would get an exorbitant previlege. Perhaps people will be willing to exchange the $20 bills for $200 in other denominations over a period of time. Most would forget the original rule, and that the bill’s denomination is actually only $20. People who have access to $20 bills would be tempted to consume a lot, and incur huge debts, based on the $20 bill being worth $200 in other denominations.
    What happens if the rule suddenly changes, or if it begins to change, e.g. 10 states decide to accept rent payments in denominations other than $20? How can this situation be explained through the balance of payments identity above?

  • Posted by Indian Investor

    Dr. Setser,
    In a short period of time delta t, total (units of local currency sold for foreign currency X)/(units of foreign currency sold in exchange for local currency) = Exchange rate (local currency/foreign currency X).
    As an example, in a particular minute, if 1000 USD were sold for RMB, and on the other side if 10,000 RMB were sold for USD, RMB/USD = 10.
    Over a long period of time T, the ratio:
    (units of local currency sold in exchange for units of foreign currency X)/(units of local currency bought in exchange for units of foreign currency X) = Exchange Rate(local currency/foreign currency X).
    This would be an accurate re statement of the balance of payments identity.
    The BOP identity arose out of the application of double entry book keeping to the national income accounts. the identity contains an implicit assumption based on the gold standard – that the exchange rate is an EXOGENOUS, or INDEPENDENT variable. Assuming the exchange rate to be an independent variable, you can arrive at a simple linear equation combining capital and current account flows, and the attendant reasoning for the balance of payments.
    Once you allow for the exchange rate to be determined as an ENDOGENOUS, or DEPENDENT variable inside your balance of payments model, you can more easily accomodate for real world implications of capital and current account flows.
    What is the relevance of this re statement to your essay above?

  • Posted by Indian Investor

    What you have calculated as an ‘external financing need’ for the US is simply the requirement of external financing that will keep the USD exchange rates steady against major world currencies. However your dominant thesis has been that misaligned exchange rates contributed to the development of unsustainable trade imbalances. The inherent dichotomy of your balance of payments equation:
    capital inflows required = imports – exports + capital outflows
    is that all the equation does is ensure that the exchange rate remains constant.
    Yet you have frequently issued calls for China to allow appreciation of the RMB against the USD, and work together with the US to address trade imbalances.
    As things stand in the real world, there are marked differences between the US and other countries in terms of external financing needs implied by current account flows.

  • Posted by Judy Yeo

    For all their faults, central bank reserve managers still seem to consider US Treasuries safer than other dollar-denominated assets.

    could it be a matter of a lack of choice? After all a switch away from the de facto currency cannot be done overnight nor a switch away from the de facto “safe” asset.