The BIS estimates dollar reserves - including dollar reserves held offshore that do not show up in the US data - increased by about $500 billion in 2004. Roughly $400 billion of reserves were invested in US securities or deposited in "onshore" US banks and thus appear in the US data, and another $100 billion or so show up in "offshore" dollar deposits.
That may not be all "dollar" holdings either. In the middle of 2004, the sum of "recorded central bank assets in the US" and "offshore dollar deposits reported by BIS banks" was $90 billion less than the estimate of dollar reserves that BIS produces bases on what central banks tell the BIS and the IMF. If that "gap" grew, dollar reserve accumulation would be even bigger.
But that is a technical detail. Let's take the BIS estimate at face value.
The IMF's data shows a $700 billion increase in global reserves in 2004 (other sources put the total at $710 billion). Using end of 2003 estimates of the stock of euro and other non-dollar reserve assets, it is reasonable to assume that $60 billion of the increase is explained by the fact that the dollar value of central bank's existing euro (and pound and yen) assets increased during 2004. That leaves an underlying increase of $640 billion.
$500b is 78% of $640 billion. And 78% is higher than any estimate of the dollar/ non-dollar breakdown of central bank's existing dollar reserves. So no diversification. That doesn't mean that central banks delivered equal support to the dollar throughout 2004 -- the fraction of reserves going into dollars might have been higher than 80% in the first 1/2 of the 2004, when Japan was most active, and lower in the 1/2 of 2004, when most reserve accumulation came from the major emerging economies.
The fraction of new reserves invested in dollars in 2004 is not quite as a high as in 2003, when the BIS estimates that over 87% of all new central bank reserves were invested in dollar assets ($441 of $503, see here). But it is still a very high number.
The story the BIS is telling is a bit at odds with the story Goldman has been telling - namely that central banks started to diversify their reserves in 2003. Since Goldman's story doesn't really fit with either the existing BIS or IMF data, Goldman is betting either that the BIS data is wrong or that the BIS data (and the underlying IMF data) for 2003 will be revised. Remember, the IMF's data, in euro terms, shows only a 30 billion increase ($30 to $40b, depending on when the euros were bought) in the central bank holdings of euros in 2003. The dollar value of the central bank's euros shot up from about $500 b to about $640b over the course of 2003, but that mostly reflects the fact that the euro rose from around 1.05 to around 1.25. Both the 2003 IMF data and the 2003 BIS data suggests that as the dollar value of central banks euro's rose, central banks increased their dollar purchases at the margin to keep the euro's share of their portfolio from rising. The story that emerges is one of central banks resisting diversification out of the dollar.
Or, alternatively, the data on new purchases in 2003 was just dominated by Japan, which was intervening like mad back then. Japan is known to basically keep all its reserves in dollars.
Japan still played a big role in 2004, but China's relative importance increased dramatically (China's contribution to the global total in 2003 was kept down by the transfer $45 billion to two state banks).
The continued high dollar share of the world's reserves makes a certain amount of sense. The US has a current account deficit and thus an ongoing need for financing. Setting the UK aside, the other main reserve currencies are parts of economic areas that have balanced current accounts (the Eurozone) or current account surpluses (Japan) and thus don't need financing.
And even 22% of the 2004 reserve increase works out to $140 billion -- or euro 110 b. Not all of the implied increase in non-dollar reserves went into euros, but a substantial share did. And a large fraction of euro 110 b works out to a number that is quite large in relation to total inflows - particularly debt related inflows - into the eurozone.
As the BIS makes clear, not all "offshore" dollars are used to finance the US. They can be used to purchase the dollar denominated debt that emerging markets issue to finance themselves, that Italy issues to finance itself, or that companies located outside the US issue to finance themselves.
But the "offshore" dollars placed in the world's banking system often do find their way back to the US. Think of the following process:
- A central bank deposits its funds in an offshore dollar bank account.
- The bank in turn lends its dollars to hedge fund, whether a London based hedge fund or a US hedge that is domiciled in the Caribbean.
- The hedge fund uses the borrowed dollars to buy longer term US securities - be they Treasuries or various types of corporate bonds.
The central bank is left with the currency risk - if the dollar's value falls, the central bank takes the hit. The bank takes on the risk the hedge fund won't be able to repay it. And the hedge fund takes on the risks associated with borrowing short-term money to buy long-term debt, but not any currency risk
The hedge fund provides the demand for long-term US assets, and thus "finances" the US current account deficit. But it is willing to finance the US current account deficit in part because it has access to dollar financing at a reasonable rate. The hedge fund is willing to borrow in dollars to buy US dollar denominated debt, but not to borrow in euros to buy dollar debt ....
Narrowly speaking, if all offshore central bank dollars were used to finance the purchase of various "onshore" US assets, the increase in dollar reserves financed about 75% of the 2004 US current account deficit ($498 v $668). But the BIS notes - accurately I think - that the US needs to finance an "equity" outflow as well as a current account deficit, so the amount of "debt" the US needs to place abroad exceeds the amount needed to finance the US current account deficit. According to the BIS, the US needed to raise $958 billion abroad in 2004, and central banks provided a bit over ½ the total needed financing.
It will be interesting to see what more the IMF's data for 2004, which should be out soon, will tell us.
The IMF data though won't help answer the really big question. 2004 is interesting, but most people are by now more interested in what is happening in 2005. That is something I'll explore in more detail soon ...