from Follow the Money

Where to invest 4.2 trillion? The latest IMF data on the currency composition of the world’s reserves

April 5, 2006

Blog Post
Blog posts represent the views of CFR fellows and staff and not those of CFR, which takes no institutional positions.

More on:

Monetary Policy

Via Menzie Chinn of Econbrowser comes word that the IMF has released its latest data on global reserves.   For reserve data and capital flows data geeks, this is big news.  I have been trying to use the latest Treasury Bulletin to figure out where oil exporters are parking all their funds.   Without much luck.  More on that later.

What does the IMF data tell us?  Not all that much in some sense.   As Menzie notes, the IMF simply doesn't have data on a lot of the world's reserves.   And that fraction is growing over time.  It was a bit under 25% a few years ago, it only a tiny bit below 33% now.  Blame China.

But the IMF still has data on the currency composition of about 67% of the world's reserves.  And the dollar's share of the reserves of countries who do report currency composition to the US has stayed remarkably constant.  It was roughly 65% in 2002.  It is now a bit over 66%.

I actually think there is an interesting story buried in those numbers.  There is some evidence that some countries -- at least those that report the currency composition of their reserves to the IMF -- stepped up their euro purchases in 2005.

But before I delve into it, let's cover the basics.  All data, obviously, comes from the IMF.

In 2004, the world's foreign exchange reserves, measured in dollars, increased by about $720b.  ($415b from countries who report currency composition to the IMF, $305b from countries that don't)

In 2005, the world's foreign exchange reserves, measured in dollars, increased by about $421b (170b from countries that report currency composition to the US, $251b from countries that don't)

As a result, by the end of 2005, central bank reserves reached $4170b - that is two times what central banks held at the end of 2001.

Those are slightly different numbers than Sangeetha Ramaswamy and I used in our reserve watch.  But only slightly.

The raw totals are deceiving. 

The $562b in euros that central banks held at the beginning of 2004 were worth $604 b at the end of 2004.  And the $660b in euros that central banks held by the end of 2004 (they bought some euros during the course of 2004) were only worth $577b at the end of 2005.   Valuation matters.

I had to make a few assumptions to get an updated estimate for the "real" increase in central bank reserves - that is the increase after stripping out valuation changes.

The key assumptions: the all non-dollar reserves moved in tandem with the euro (this clearly isn't true, but I haven't had time to set everything up for more than two currencies); and roughly 35% of the reserves of countries that don't report to the IMF are in euros/ euro-equivalents, and 65% are in dollars.  In other words, countries that don't report have basically the same portfolio as those that do.

That is rough, but it allows me to do a very crude estimate.   The global increase in reserves as around $643b in 2004, and $582b in 2005.   All the increase in 2005 came from emerging economies.

For a host of complicated reasons, the IMF's data also doesn't include the majority of the increase in the official assets of the Gulf states, so it underestimates the overall growth in official assets.  It also leaves out the $15b China transferred to a state bank.

Now for the fun bit.

If you look at the IMF's data in dollars, it seems like the world added a ton of euros to its portfolio and then stopped adding euros in 2005.   The (dollar) increase in euro reserves was $98b in 2004, and only $25b in 2005.   Again, this isn't the full global total; it just covers the 67% of the world's reserves where the IMF has data on currency composition. 

But if you look at that data in euros, the story looks different.

The world held 408.3b euros at the end of 2002, 446.4b euros at the end of 2003, 487.5b at the end of 2004 and 578.7b euros at the end of 2005.   In other words, the world's central banks added a lot more euros to their portfolio in 2005 (91b euros) than in 2004 (41b euros) or 2003 (38b euros).

Aha, diversification!

Not so fast.

Relative to the overall increase in reserves for countries with known currency composition  the purchases of euros in 2003 and 2004 were super low. 

To do the following calculations, I converted euro purchases into dollars at the end of the year rate (one source of error) and compared those purchases to total central bank purchases among countries that report their currency composition.  But I only adjusted for the euro/dollar, not for the yen/ dollar or pound dollar.

On that basis, $48b of $289b went into euros in 03 (16.5%) and $55b of 373b in 2004(14.7%).  Central banks were buying far more dollars that they should in those years, and far fewer euros than one would expect. That changed in 2005. But still only 43% of my crude estimate of the overall flow of new reserves ($108 of $253b) was going to euros. 

In reality, I am underestimating valuation shifts by not including yens and pounds and the like, and since I am underestimating valuation shifts, I am underestimating the flow for 2005 and overstating them in 2003 and 2004.  That implies the euros share of new reserve purchases would be a bit bigger in 03 and 04 and a bit smaller in 05.  But there clearly was a shift toward euros in 2005.

At least among central banks who report the currency composition of their reserves to the IMF.

The overall data could be totally different, since we don't know what China and others were doing and thus what was happening with the $251b (more after any reasonable valuation adjustment) that they added to their reserves.

Still, there was a shift toward euro among some central banks.

Possible explanations:

  • Central banks are targeting a constant euro/ dollar share in their portfolio, so when the euro rises in dollar terms, they have to sell euros (or buy less) to keep the euro's share from rising, and when the dollar rises, they have to buy more euros to keep the euro's share up.  
  • Japan buys a  lot of dollars.  Japan's reserve increase drove the data for 2003 and 2004, but 2005.  If Russia reports its currency composition, it would be the largest accumulator in 2005.  And it clearly buys more euros than Japan. 
  • Central banks really were diversifying just a bit.  And it had very little impact on the market ...

Let me conclude by noting that I think the evidence supports one argument that Dooley has made in the past.  The Eurozone got over $100b in inflows from central banks in 2005.   Probably more - since the countries that don't report also bought at least some euros.  The Eurozone didn't run a comparable current account deficit.  That implies that there were offsetting private outflows from the eurozone, as Europeans looked for higher returns abroad.

Yet more evidence that stagnant Europe can't make use of manna from heaven - foreign inflows into low yielding European debt -- to finance investment in Europe?  Perhaps.

But also evidence that Europe may be the real shrewd capitalists.  Europe take central bank inflows and use those inflows to finance the FDI of European companies.    Sort of like the US did during the initial Bretton Woods system.   Emerging economies get low-yielding reserves.  European companies get high-yielding investments in emerging economies (And elsewhere).  They just may channel all profits from those investments through Ireland for tax reasons - so there is no guarantee that these profits will actually show up in the data ...

What of the US?  Doesn't it do the same.  Not really.  It uses central bank inflows mostly to finance its current account deficit.  Remember, in 2005, the US didn't increase its foreign direct investment.   The homeland investment act.  But more broadly, even in years when the US does invest abroad, the US has to finance both its external investment and its current account deficit abroad.  And right now, the deficit is big.    That doesn't seem all that shrewd to me.

And even if the world's central banks bought $200b in euros, pounds and yen in 2005 -- about a third of the total (flow or valuation-adjusted) increase in the world's reserves during 2005, they still provided about $400b in financing to the US.   And that leaves out the Gulf states, who prefer to invest through investment authorities rather than central banks.

And lest anyone think I have completely embraced Dooley (really Dooley, Garber and Folkersts-Landau), let me just state this:

Their argument that the US financial system serves as an offshore intermediary for Chinese savings just doesn't work.   

The US has roughly $60b of FDI in China according to the latest data, out of a total of around $621b.  That may slightly understate US investment, but there is no doubt that most investment in China comes from Japan, Taiwan, Korea and similar high savings Asian economies.   Compare that $60b claim with the $540b in US dollars that China's central bank (and three of the four state banks) would hold if it had only 60% of its $900b or so in reserves (again, counting reserves transferred to the state banks) in dollars.    If the US is acting as an offshore intermediary, all $540b (or more) would need to be channeled back into China.    That simply is not the case.

What does seem to be happening? the central bank of China is acting as an intermediary between Chinese savings (in RMB) and Asian FDI (also flowing into China) and the US, by issuing RMB debt and buying dollars.  Call it onshore intermediation of China's savings surplus by the central bank.  Or vendor financing.

More on:

Monetary Policy