Brad Setser

Follow the Money

Cross border flows, with a bit of macroeconomics

Print Print Cite Cite
Style: MLA APA Chicago Close


2007 all over again? The dollar, central bank reserves and US bonds

by Brad Setser
May 26, 2009

Lower interest rates in the US than in much of Europe and most emerging economies

Slower expected growth in the US than in the emerging world

Rising oil prices

Falling dollar.

That describes the past week.

But it also describes most of 2007 and the first part of 2008.

In the last WEO (Box 1.4), the IMF argued that the world’s imbalances weren’t at the heart of the recent crisis, as the trigger for the crisis wasn’t a withdrawal of foreign financing to the US. The credit crisis, in other words, wasn’t a dollar crisis.

That argument was a bit overstated. The Bretton Woods 2 system was central to the ability of the United States to sustain a large deficit in the household sector – just as the expansion of the US household deficit was central to the ability of many emerging economies to grow their exports. Absent central bank demand for dollars, the natural circuit breakers would have kicked in earlier, before so much risk accumulated in the financial sector.

Moreover, it ignores the fact that there was something of a dollar crisis from the end of 2006 to early 2008.

When the US slowed and the global economy (and the European economy) didn’t, private money moved from the slow growing US to the fast growing emerging world in a big way. The IMF’s data suggests that capital flows to the emerging world more than doubled in 2007 – and 2006 wasn’t a shabby year. Net private inflows to emerging economies went from around $200b in 2006* to $600b in 2007. Private investors wanted to finance deficits in the emerging world, not the US – especially when US rates were below rates globally. Normally, that would force the US to adjust – i.e. reduce its (large) current account deficit. That didn’t really happen. Why?

Simple: The money flooding the emerging world was recycled back into the US by emerging market central banks. European countries generally let their currencies float against the dollar. But many emerging economies didn’t let their currencies float freely. A rise in demand for their currency leads to a rise in reserves, not a rise in price. As a result, there has been a strong correlation between a rise in the euro (i.e. a fall in the dollar) and a rise in the reserves of the world’s emerging economies. Consider this chart – which plots emerging market dollar reserve growth from the IMF’s quarterly COFER data against the euro … **


If the rise in reserve growth in the emerging world is a sign of the amount of pressure on the dollar, then the dollar was under tremendous pressure from late 2006 on. It central banks had broke – and lost their willingness to add to their dollar holdings then – there likely would have been a dollar crisis. A fall in inflows would have forced the US to adjust well before September 2008.

The chart, incidentally, was prepared for a CFR contingency planning memo I did on the foreign policy implications of a potential dollar crisis. By crisis, though, I had something rather more dramatic than the slide of last week in mind. Especially as it seems that the dollar has stabilized a bit this week.

Why is all this relevant?

Last week felt a more like the fourth quarter of 2007 than the fourth quarter of 2008. For whatever reason — an end to deleveraging and a rise in the world’s appetite for emerging market risk or concern that the Fed’s desire to avoid deflation would, in the context of a large fiscal deficit, would lead to a rise in inflation and future dollar weakness – demand for US assets fell.

In some sense, the dollar’s fall shouldn’t be a surprise. Low interest rates typically help to stimulate an economy is by bringing the value of the currency down and thus helping exports. Moreover, as Broda, Ghezzi and Levy-Yeyati of Barcap argue, it is reasonable to think that financial deglobalization will generally make it harder for any country, the US included, to sustain large deficits. A weaker dollar helps to limit the spillover of the US stimulus into external demand – and pushes other countries to rely less on exports for growth. It thus pushes them to do more to stimulate domestic demand (as in say Japan). And by lowering the US trade deficit, it reduces the amount of financing the US needs to attract from the erst of the world.

A declining dollar, though, forces a host of emerging economies are to decide how much to intervene to try to limit their currencies’ appreciation. And in some cases, they have to decide whether or not it makes sense to (still) peg to the dollar.

Don’t forget, China’s currency depreciated alongside the dollar last week. China’s exporter are thrilled. But if China is serious about reducing its exposure to the dollar, it cannot allow dollar weakness to turn into RMB weakness. We know how that story plays out.

Reading Bloomberg on Friday — and the Wall Street Journal this morning — left little doubt that many central banks were once again intervening heavily in the foreign exchange market. Russia’s central bank bought over $5 billion last week. A surge in intervention would explain the strong rise in the Fed’s custodial accounts over the last two weeks. They rose by over $50 billion in the last two weeks of data – a $100b monthly/ $1200b annual pace.

That pace of reserve growth would allow the US to sustain it current account deficit – and add to their foreign portfolio. So long as Americans want to buy the financial assets of fast growing emerging economies, Americans actually need to borrow even more than is required to cover the trade deficit from the rest of the world.

The data then suggests not all that much has changed – despite all the talk about China’s desire to find an alternative to the dollar. China still buying dollars to keep its currency from appreciating. Words and actions haven’t matched.

At the same time, May 2009 isn’t quite May 2008 – or November 2007.

Three things have changed:

a) First, the US trade deficit is about half as big as it was in 2008 – or early 2008. The amount the US needs to borrow from the world has gone way down. If American households desired level of savings has gone up – and thus their desired level of consumption has gone down – the size of the dollar depreciation needed to get rid of the trade deficit and America’s need to be a net borrower from the world has also gone down. Expenditure reduction (spending less on all goods, including imports, no matter what the level of the dollar) has substituted for expenditure switching (shifting from foreign to domestic production as the price of imports rises)

b) Second, “strong” economic performance in the current context consists of shrinking less rapidly than the rest of the world, not growing more rapidly than the rest of the world. The US in that sense is doing better than Europe. And it isn’t clear that the gap between say US and Chinese growth is actually any larger now than it was a year ago.

c) Third, the rise in central bank reserves isn’t translating into a rise in demand for longer-term US bonds. Central banks are just buying short-term bills. That presumably is one of the reasons why long-term rates are rising now – while they remained (surprisingly) low back in 2006, 2007 and 2008. Central banks weren’t willing to buy long-term notes at 2% — or even at 3%. Maybe they just didn’t want to lock in low rates. Maybe they feared a mark-to-market capital loss if rates rose. Or maybe they fear that inflation will rise, eroding the real value of longer-term claims. In some sense, it doesn’t matter. The dynamics of the market changed …

One of the “stabilizers” that artificially dampened volatility in the pre-crisis world – the tendency for a fall in the dollar to be associated with a rise in intervention and a rise in central bank demand for longer-term US treasuries (and agencies) — seems to have gone away. Or at least it is operating a bit differently. Central banks are still intervening to prop the dollar up. But they aren’t currently willing to buy longer-term US bonds. And the US doesn’t really want to finance its fiscal deficit just by selling bills.

* Private inflows to the emerging world were a bit lower in 2006 than in 2005. This though is somewhat misleading. In 2006, there was a surge in private outflows from China – but that surge was entirely due to a policy decision to use swaps to shift the management of some of China’s reserves to the state banks and other state-run financial institutions. Those outflows weren’t really private. True inflows consequently were almost certainly around $300b in 2006.
**The dollar’s value against the euro is an input for the model estimating dollar reserve growth among those countries that do not report detailed data to the IMF. It isn’t the key input though – the total rise in reserves matters more. Note though the correlation also holds among the countries that do report detailed data to the IMF – and here the data series are truly independent.

Some additional supporting analytics illustrating how central bank demand for dollars has tended to go up when the dollar goes down (v the euro):

With the help of Paul Swartz, I used a bit of math (or magic) to turn the IMF’s quarterly COFER series into a monthly series. The COFER series has been adjusted to include Saudi and Chinese non-reserve foreign assets reported by their respective central banks. The correlation between estimated dollar reserve growth on a rolling 12m basis and the dollar is uncanny.


To be sure, I am using a portfolio balance model to estimate the increase in the dollar reserves of countries that do not report data to the IMF, and thus a slide in the dollar automatically increases dollar reserve growth for any given level of overall reserve growth. This though doesn’t really drive the results — and it strikes me as a reasonable assumption. The IMF data for countries that don’t report the currency composition of their reserves is dominated by China — and I have used the TIC data to get a decent sense of the dollar composition of China’s reserves. The Saudis are the second biggest component of the data, and I think it is safe to assume that they haven’t shifted away from the dollar in mass. Moreover, I am in effect assuming that the countries that do not report detailed data to the IMF are acting like the countries that do — a chart that just used the data for reporting countries would show the same trend.

I added the 12m change in the Fed’s custodial holdings as a stress test. The Fed’s custodial holdings generally moved together with overall dollar reserves until mid 2007. That isn’t hard to explain. Asian countries also tend to use the Fed’s custodial facilities more than the oil exporters; that helps explain the gap from mid 2007 to mid 2008. Moreover, we know that China was buying equities from q2 2007 on, and in general central banks were taking more risks with their portfolio. The result was less use of the Fed’s custodial facilities.

More recently, the Fed’s custodial holdigns have increased by more than dollar reserves, as countries shifted back into safe assets.

I also looked at a a plot showing the 3m change in the Fed’s custodial holdings and the dollar’s value v the euro. In most cases, a rise in the euro’s value v the dollar is tied to faster growth in the Fed’s custodial holdings. That is what one would expect if other central banks are resisting pressure for their currencues to rise along with the euro against the dollar.


There is an important exception though: q3 2007, just after the subprime crisis? Reserve growth was still pretty strong then, so the fall off in inflows to the Fed’s custodial accounts isn’t explained by a fall in reserve growth. My guess is that central banks shifted funds to the BIS, but I am not sure — it is a bit of a mystery.


  • Posted by FollowTheMoney

    Rising Oil=Falling dollar. The equation is that simple.

    $200 oil would imply a return to global growth, but what would that mean for the dollar?

    Back to the dollar, an interesting note from Market Ticker:

    “This week we will be treated to the US Government attempting to sell $100 billion in new Treasuries to finance its profligate spending habits.

    Bernanke, for his part, is on the cusp of losing control of the long end of the bond curve. If it gets away from him he will have only two choices: pull liquidity and allow the curve to spike higher, repeating almost exactly what happened in the 1930s, or ramp up his “monetization campaign” to meet the issuance, risking an immediate tender of the entire outstanding float, resulting in an even worse outcome – a choice between collapse of the government or an Argentina-style currency implosion followed by that same collapse.”

    Which will it be?

    Market Ticker

    “If Foreign Central Banks are selling into Ben’s bid then the game is literally weeks or even days away from being over.”

  • Posted by bsetser

    follow the money — except the relationship hasn’t always been that simple. back in the late 90s, when a US investment and consumption boom (IT driven) was driving global growth, the dollar and oil went up together. The current relationship exists in large part b/c other parts of the world have been growing faster than the us (05-mid 08) and now are expected to recover before the US.

  • Posted by Michael

    re; question on follow the money’s comment:

    ‘follow the money’ says:
    “$200 oil would imply a return to global growth, but what would that mean for the dollar?”

    OK; if supply-side issues are not to blame for $200 oil, I agree that $200 oil would imply a return to global growth, but why does the price of oil have to exceed $80- $90 { what some industry experts beleive to be the current marginal cost of new oil sources } by so much to signal a return to global growth?

    Thanks for the thought provoking post !

  • Posted by FollowTheMoney

    ” The current relationship exists in large part b/c other parts of the world have been growing faster than the us (05-mid 08) and now are expected to recover before the US.”

    Since EM may recover quicker than us, how about proposing oil to be traded in exchange for a basket of currencies? in theevent of $200 oil, the dollar would not be devalued as greatly…the way things are going $100 oil would signal 1.7/euro…if rising oil/falling dollar scenario remains in line…

    @ Michael-

    Love it or hate, peak oil is more fact than fiction. Credit has also been cut so it’ll be more challenging for oil drillers to get loans to find new exploration. Furthermore China is adding X number of cars on the roads each month, as is India and Brazil. Why else would China be making agreements with Iran and Brazil. Alot of whats happening in the world is based on humans dependence on energy.

    I think the world will have $200 oil within 5-10 years. Something we all have to live with unless we enter an awful over extended period of global depression or some new technology is introduced.

  • Posted by Michael

    Thanks for this uber-meaty international finance post Brad, it will take
    a few hours of homework for me to really go through it, but I have done enough homework to ask a question or two for clarification.

    When you say
    “A weaker dollar helps to limit the spillover of the US stimulus into external demand – and pushes other countries to rely on exports for growth, and thus pushes them to do more to stimulate domestic demand (as in say Japan). It thus helps to limit the amount of financing the US needs.”

    Did you mean
    “and pushes other countries to rely [ LESS ] on exports for growth” ?

    When you said
    “It thus helps to limit the amount of financing the US needs.”
    did you mean ( my attempt to paraphrase ….. )
    ‘it thus limits the funding available from abroad to finance the US needs”

    The point of clarification here is that non-US countries focusing more on their own domestic demand will not of itself “limit the amount of financing the US needs”, that is saying that these actions of non-US countries will lead to reduced funding needs in the US is not what you were implying.

    Addressing these questions will hopefully make reading this post more efficient for others and/or resolve my own misconceptions.

    Good day.

  • Posted by Peter

    Brad, every once in a while you produce a major post, one that illuminates the big picture connection between global imbalances and financial fragility. This is one of them. We have “enjoyed” a temporary respite from the force of imbalances due to the collapse of trade, but (1) surely recovery of the world’s major economies has to entail the recovery of trade, and (2) speculative private capital flows remain volatile, complicating rebalancing. We hear much about financial regulation as a basis for returning to long run growth, but nothing about the structural reforms needed to avoid chronic, massive imbalances. Yours is one of the few voices that calls attention to this.


  • Posted by bsetser

    michael — yes, i meant rely “less” on exports for growth. that means more demand for us goods and less demand for us bonds — but it also means the us has less need to sell bonds to the world. I need to edit that sentence.

  • Posted by Michael Carroll

    ” May 2008 isn’t quite May 2007 – or November 2007. ”

    You mean May 2009…

    Nice post.

  • Posted by Cedric Regula


    I still think it’s a little premature to tally up our new found savings rate and apply it to the Federal deficit.

    The average household credit card debt is $8400. If the average income household of $60k saves 5%, that is 3000/year. So we’ll have to check back with the average Joe&Jane in about 3 years to see if we can get them interested in whatever USG debt deal is available at that time.

    Tho it may be a better deal then. Today that bad boy 30 year bond has been taking some love taps and I’m up nearly half a handbag on my short so far today.

    USD decided to go sideways today.

    Also, the 10 year budget (if one believes in such things) says someone has to buy $9 Trillion in USG debt.

  • Posted by bsetser

    michael — i still haven’t mentally accepted that we are in 09 rather than 08 …

  • Posted by Glen M

    Dean Baker has a great post today…….

    There is a bizarre theory circulating in high Washington circles, expressed today by Sebastian Mallaby in the Post, that China is concerned that the huge dollar reserves it holds will lose value. The reason this is bizarre is that the dollar already plunged in value over the years 2002-2008 and China just kept buying more dollars.

    The euro went from being worth just over 80 cents at the dollar peak in 2002 to over $1.60 at its trough early last year. Through this whole slide, China just kept buying up more dollars. Does anyone think that China’s leaders did not notice the plunge in the value of the dollar? This is not exactly secret information.

    Obviously, China’s central bank was fully aware that the dollar was losing value but was willing to buy dollars anyhow in order to preserve its export market in the United States. That is the reason that it continues to buy dollars even though its leaders know that they will lose money on the deal. The economy can easily afford the loss, contrary to the bizarre calculations Mallaby uses in his column.

    If it seems strange that elite Washington types can push economic views that are far removed from reality, remember, these people could not see an $8 trillion housing bubble.

    PS. Brad it would be interesting to have a look at the collateral damage to other trading relationships as a result of the currency manipulations.

  • Posted by Dennis Redmond

    Brad Setser wrote:

    c) Third, the rise in central bank reserves isn’t translating into a rise in demand for longer-term US bonds. Central banks are just buying short-term bills.

    The structural issue here is that the US, for various reasons (regulatory capture, neoliberalism in high places, etc.) is basically replacing busted private debt with public debt — trying to monetize the problem, instead of wiping out the debt altogether (Roubini has said this a thousand times better than I can).

    Rising rates mean the BRICs, the EU, and core East Asia have decided they aren’t going to fall for this con game — the US will have to pay an “American premium” to hire foreign money. In short — more stagnation for the US.

  • Posted by FollowTheMoney

    this is not a pump and dump forum.

  • Posted by ReformerRay

    Brad –

    You do a wonderful job tracking flow of funds between governments and the private sector all over the world. This flow is very complicated.

    I am disturbed by one simplification you use. You say “First, the US trade deficit is about half as big as it was in 2008 – or early 2008. The amount the US needs to borrow from the world has gone way down.”

    I object to the conclusion that the size of the trade deficit controls the amount the U.S. needs to borrow from the rest of the world. I think the amount the U.S. does borrow from the rest of the world is controlled solely by the amount of Treasury bills sold and the share of them purchased by U.S. citizens. I insist that the Net Worth in the U.S. is large enough to absorb all the Treasury bills. Thus the U.S. does not have to borrow anything from the rest of the world. The fact that foreigners purchase a large share of U.S. treasury bills is their choice. The fact that they have a large number of dollars with which to purchase Treasury bills is due to the U.S. trade deficit AND to swaps they make between their currency and U.S. dollars. The impact of this flow on interest paid for Treasury bills and the value of the U.S. dollar is a very important issue which you do a good job explaining. But the amount of Treasury bills sold depends upon U.S. needs, not the trade deficit.

    The trade deficit does deposit dollars in the hands of foreigners. Those dollars are a claim on U.S. resources. But those dollars can be used to buy U.S. corporations, U.S. farm land, U.S. skyscrapers, U.S. private homes. etc. Once those dollars are transferred overseas, the U.S. loses control of what U.S. asset they will be exchanged for.

    Perhaps I am nit-picking. The trade deficit provides foreigners with dollars which are a claim on U.S. assets. My only point is that those dollars do not have to be converted into Treasury bills. If they are used to purchase something else of value in the U.S., the obligation of the U.S. to exchange those dollars for U.S. assets is ended. If the U.S. did not run a fiscal deficit, no net new U.S. Treasury bills would need to be issued each year (we would still have to issue new debt to pay for old debt that matures).

    Instead of saying that the amount the U.S. needs to borrow from the rest of the world, I would prefer the statement that “net payment to foreigners each year has gone way down”.

    The U.S. pays for the trade deficit only once – when the U.S. purchaser of imports sends money back to the producer of the goods imported..

  • Posted by don

    The move by foreign central banks to short-term debt is disturbing. The only silver lining is that CB actions are often a bit stupid (the timing of their sales of gold, movement of Asian CB’s out of the dollar towards the yen when the yen hit its record of about 80 yen to the dollar), so they may well be wrong about dollar inflation.
    I would not characterize foreign official inflows as “needed” to finance the U.S. trade deficit. Instead, I would say they are forcing the deficit and preventing it from shrinking. The recent large inflows are particularly disturbing and imply a step-up in competitive devaluation strategies to hold up demand abroad.
    Europe will have a very difficult time dealing with these strategies along with the the decline in U.S. consumer demand.

  • Posted by Cedric Regula

    At the close today those bad boy 30 year treasuries were again taken out to the tool shed and given a voracious spanking. Cries of pain were heard thruout Oz, and bond munchkins screamed for Wizard Ben to save the poor treasuries. But alas, too many wicked Treasuries in Oz and Wizard Ben could not save them.

    Ended up one and a half handbags today.

  • Posted by Rien Huizer


    A treat. Talking facts. Ps can we have more of this?

    No reason to worry about the T yield curve: it finally shows what the real markets (not politically motivated CBs) are prepared to pay. And obviously the market tells us that there is a significant risk that present policy will bring forth future inflation. Pretty normal at last.

    The USD interveners have no incentive to keep US interest rates down (by itself that observation merits clarification and further analysis though) perhaps they never had or in the past thwy were happy to play the yield curve along with doing their national duty. It may be worth our while to find out why the interveners bought linger term paper in the past, because it really does not make sense (assuming Twofish was right that there are a lot of people with US finance degrees now in policy making circles in Beijing)

    Of course if the market starts to believe that USD FX is too low and bond yields are too high, we should (may) get either a reaction in the USD or in long term yields or both. For China that would make a difference: a higher USD is problem, lower USTP yields?higher bond prices are merely a missed opportunity. So, in fact, not buying bonds may mean contributing to a US rise when normal international investors start liking USD bonds, which usually is accompanied by an upward movement in the currency. Confusing?