The IMF just released new quarterly data on global reserve accumulation (Hat tip, Menzie Chinn). Just don't look to it to find out if the world's central banks are adding to their dollar or euro reserves. Of the $294 billion increase in global reserves in the first three quarters of this year, $209 billion of it has come from countries that do not provide data on the currency composition of their reserves to the IMF.
Consider the third quarter.
- The increase in global reserves: $95.3 billion.
- The increase in dollar reserves among reporting countries, according to the IMF data: $21.1 billion.
So over $74 billion of the increase in global reserves in the third quarter came from countries that don't report the currency composition of their reserves to the IMF. Yep, China doesn't report.
We actually know a bit more than can be divided from the IMF data alone. Official inflows to the US in the third quarter, according to the US data, were $38.4 billion. But even if we use the US data for the total increase in dollar reserves, there is still a big gap -- $56.9 billion - between the recorded official inflows to the US and known increase in global reserves. Like Martin Feldstein and a host of others, I suspect that the US data signficantly understates central bank flows.
And there is plenty more detective work than can be done using the IMF's data, and a few other sources ...
While in dollar terms, the pace of reserve accumulation in 2005 ($294 billion in the first three quarters, $392 billion annualized) is far lower than the pace of reserve accumulation in 2004 ($720 billion), a bit of caution is in order. Valuation changes increased the global total in 2004, and are reducing it in 2005.
The impact of these changes is pretty easy to see if you look carefully at the euro-denominated reserves line of the IMF report. In dollar terms, euro denominated reserves soared by 69 billion in the fourth quarter of 2004 (when the euro went from $1.242 to $1.354); in euro terms, the increase was a far more modest euro 12.1 billion (roughly $16.4 billion). That is why it is easy to say that at least $50 billion of the rise in the world's total reserves in q4 2004 stemmed from valuation changes.
The same type of analysis can be done for 2005. In dollar terms, the world's euro reserves have stagnated in 2005 - rising only $4.4 billion. In euro terms, though, euro-denominated reserves increased by euro 62.7 billion. In other words, the world's central banks have added roughly $75 billion to their holdings of euro reserves during the course of 2005, offsetting a roughly $70 billion fall in dollar value of their existing holdings of euro reserves.
That is why it is important to adjust the global reserve accumulation totals to reflect valuation changes. In dollar terms, reserves increased by about $295 billion in the first three quarters, or about $100 billion a total.
But we need to add $70 billion to that total to reflect known valuation losses on the world's euro reserves.
If even 20% of the world's $1136 billion of "unallocated" reserves at the end of 2004 -were in euros, we need to add another $25 billion to the global reserve increase in the first three quarters of 2005.
So let's add $95 billion to the $295 billion to reflect valuation losses. That brings total reserve accumulation up to $370 billion in the first three quarters of the year.
One more adjustment is also needed. The Saudi's official foreign currency reserves fell by $3.3 billion in the first three quarters of 2005. That is a bit misleading though. The total assets (mostly foreign assets) of the Saudi monetary authority rose by $41.4 billion over the same time ($17.9 billion of the increase came in the fourth quarter). Data here, in riyals. The exchange rate is 3.75 riyal/ dollar.
Incidentally, the other big oil exporter - Russia - reserves increased by only $8 billion in the third quarter, largely because Russia was paying down its foreign debt in big way.
No matter. Add in the overall increase in Saudi assets, and total global reserve accumulation was around $410 billion in the first three quarters of this year. Annualized, that works out to around $550 billion.
At least $80 billion of the 2004 increase comes from valuation gains - actually the total is probably a bit higher. So $550 billion should be compared to say $640 billion in 2004.
That is a small fall off, but not much of a fall off.
And far too small a fall off to explain the fall in recorded central bank inflows to the US. Recorded inflows to the US in 2004: $395 billion. Inflows to the US in 2005 if the first three quarters provide a reliable quite to the year: $195.
Recorded inflows to US were 62% of the global increase in 2004.
For 2005, recorded inflows into the US are more like 35% of the global total (including the increase in the official assets of the Saudi monetary authority in 2005 as part of the global total).
Something is going on.
I know all the adjustments I made to turn the $295 billion increase in stock of global reserves in the IMF data in the first three quarters (estimated $395 billion for the year) into a $440 billion "real" increase over the first three quarters (estimated $550 billion for the year) can appear like a bunch of hocus pocus. But I tried to be very transparent with my calculations. The only assumption I made was that 20% of the unallocated reserves were in euro. Everything else follows directly from the data. Trust me: I am not making this stuff up -
Indeed, if I worked a bit on the data for pound and yen reserves, I suspect my estimate for the pace of the underlying increase in global reserves in 2005 would end up being a bit higher.