Martin Feldstein is right

Felstein has argued that more “competitive” dollar would contribute to reducing the US trade deficit, and at least help to slow the rise in the US current account deficit.  I agree – though I would think the case for Asian and Gulf appreciation is stronger than the case for Euro appreciation.   Dollar depreciation alone isn't enough to bring the US trade deficit down (and slow the increase in the US current account deficit, which is set to rise because of growing net interest payments even if the trade deficit stabilizes), but it is a part of the process.

But that is not what this post is about.  Feldstein also has argued that the US data understate central bank financing of the United States.   Throw in financing from the oil states' investment authorities as well.   And the more I look at the data, the stronger Feldstein’s argument that the US data understate official inflows looks. 

For one, recorded inflows of around $220b seem low relative to global reserve accumulation (after adjusting for valuation changes, and including all Saudi foreign assets) of around $660-670b.     There is no doubt that central banks bought more euros and pounds and yen in 2005 than they did in 2004.   Recorded purchases of euros, yen, pounds and the like in the IMF’s COFER data totalled $150 billion, and since the IMF does not have data on the current composition of China and others who added, by my estimates, $300b to their reserves (on a valuation adjusted basis), that total is no doubt low.    Add in another $80b in non-dollar purchases from that set of countries.   That makes $230b, out of around $580b in (valuation-adjusted) reserve  growth -- leaving $350b for the dollar.   I won't go through all the details for how I calculated this, but it is based on a fairly detailed analysis -- and includes an implicit guess about changes in the portfolio of those countries who don't report their currency composition of their reserves to the IMF.

However, the IMF COFER data exlude reserves China shifted to its state banks (and a smaller currency swap), Taiwan's reserves and the increase in the Saudi-Monetary authorities non-reserve foreign assets.  Add those in and global reserves grew by about $660-70b by my estimates.  So even if the world’s central banks bought $250b of euros, pounds and yen in 2005 – that still leaves a bit over $400b for the dollar.   

$220b shows up in the US data, maybe $90b will show up in the BIS data showing central bank dollar deposits …  and so on.   But there is still a gap of around $100b.    Maybe more.  

And the $660-70b total leaves out the Norwegian oil funds, and the oil money parked in the Kuwait investment authority, Abu Dhabi’s investment authority and so on.    Add in at least another $50b there.    A good chuck of that went into dollars.    So I strongly suspect the US data understates official financing by at least $100b, maybe more.    And that sets aside the question of what the world’s banks are doing with all the dollars that central banks have placed on deposit – some no doubt are lent out to folks who lend to the US (though some are lent out to folks who buy Brazil’s dollar bonds as well). 

Another way of getting at the same result:  There is a huge gap between the Gulf oil exporters current account surplus and recorded inflows into the US – which likely implies that a lot of purchases went through London.   The IMF's WEO data shows that Middle Eastern reserves (i think this data includes all Saudi assets) increased by $107b in 2005, and there were official outflows (Think Kuwait investment authority and the like) of anoter $92b.  Call it $200b.  Recorded US inflows from the Gulf were only a bit over $10b.   That's a big, big gap.   Not all the $200b went to the US, but more than $10b did.   The US data may understate official inflows from the Gulf by as much as $100b. 

If attempts to show that the gap between known central bank flows of dollars (and known Gulf purchases of dollars) and their overall reserve increase (the Gulf’s current account surplus) are not convincing, just look at the following graph.

I plotted the increase in global reserves against (net) debt inflows to the US – that is foreign purchases of US debt net of US purchases of foreign debt – and recorded official inflows to the US.    It turns out that official inflows tracked global reserve growth until 2004 – which makes sense, because all Japanese flows show up in the official US data – but not in 2005.     There is a much closer match between overall reserve growth and net foreign purchases of US debt.

Maybe that is not determinative, but it is certainly suggestive.  It certainly seems like the US data misses some official flows.

In the part of the Treasury's Foreign Exchange report that no one reads, the Treasury argued -- based on the official data -- that private demand for US assets picked up in 2005.  They are right.  But they insist a bit too much.    Private flows provided, in my view, maybe $400b of the total net flows the US needed in 2005 – far more than the $250b or so they provided according to the US data in 2004.  And I suspect that the US data somewhat understated central bank flows in 2004 as well.   Total dollar reserve accumulation, according to the BIS, was closer to $500b -- and most of that came back to the US, one way or another.

 
$400b in net private flows from foreign investors to the US in 2005 is more than the $250b (if you believe the official data) or $150b (if you believe me/ the BIS and think offshore dollar deposits from central banks indirectly helped finance the US) in private financing for the US in 2004.  But $400b is not enough to cover a $800b deficit.    The US still needed $400b from its friends in the governments of China, Russia and the Middle East.

And all the available data suggests it will need far more in 2006.