Why didn’t the current account deficit start to adjust in 2003? Will the adjustment start in 2006? And where are all the world’s reserves?

I think I can now welcome Brad DeLong to the "gloom and doom caucus, trade deficit division."  He joins some pretty good company:  Paul Volcker, Robert Rubin, Larry Summers, Warren Buffet, Martin Feldstein (bipartisianship!), Stephen Roach and Bill Gross (at least some of the time).   But not Ben Bernanke.  Bernanke anchors the "don't worry" camp.

DeLong highlights why we should worry.   At some point, the deficit will have to correct.   And that correction implies huge changes in the US economy - and in the composition of employment in the US.  And equally huge changes in the Chinese economy - and the composition of Chinese employment.   China here is a metaphor for all the countries that have grown fat and happy exporting to the US.  

The question that the doom and gloom caucus cannot answer is when will the correction start.  Tis pretty clear that the trade deficit cannot continue to widen for ever.  But it has widened a lot recently, in the face of a lot of hand-wringing.  DeLong's Socratic dialogue:

"Agathon: And when do your models predict the dollar will fall?

Kapelikos: 2003.

Agathon: Three years ago?

Kapelikos: Yep.

Agathon: But as long as people believe the argument on the other side, the dollar doesn't decline, and the argument looks correct?

Kapelikos: Yep--for one more year.

Agathon: How many more yars are you going to be saying, "Wait just one more year"?

Kapelikos: Until I can say, "I told you so." (End of DeLong insert)

Ironically, the models that suggested that the deficit should have started to correct in 2003 were not totally off.

Why?  Because the deficit would have corrected in 2003 but for a change in official policy - one which led to surge in reserve growth and a surge in central bank financing of the US. 

Continues, with charts on global reserve growth and central bank financing of the US!

In the following graph, I plotted net private capital flows to the US against net official flows (using the US data) and the US current account deficit.   I was lazy, so I did all the numbers in nominal dollars, but the chart is still revealing.

Net private capital flows to the US fell significantly - from $400 b to $200b - between 2001 and 2004.   That normally would have led the current account deficit to fall.   But it didn't - instead it rose from around $400 to a bit under $700b.   Why was this possible?

Because central bank financing of the US soared!  It rose from under $50b in 2001 to $400b in 2004.  And the Bank of International Settlements (BIS) thinks - accurately in my view - that the US data understates central bank dollar accumulation, and thus central bank financing of the US.  The BIS thinks total dollar reserve growth was more like $500b in 2004.

You will note that this changed in 2005.  Private flows soared - they rose above their .com peak.   This wasn't because foreigners fell in love with US equities.  It was because foreigners fell in love with US debt.    Interest rate differentials and all.

And central bank financing of the US - at least according to the US data - fell sharply. So all is good.   The US is once again financing large deficits because it is the best place in the world for private capital ...

No so fast.

The fall off in recorded official inflows to the US has not been matched by a fall off in global reserve accumulation.

The IMF's data shows that the world's stock of reserves increased by $427b in 2005, well below the $695b or so increase in 2004.    But that is misleading.   The 2004 number was inflated by the rising dollar value of the world's holdings of euro and yen, the 2005 number was deflated by the falling dollar value of the world's holdings of euro and yen. 

Adjusting for these valuation effects, I think global reserves grew by about $615b in 2004, and about $590b in 2005.  

And that total leave out the $15b China transferred to a state bank in 2005, the $6b (maybe more) that the Chinese swapped with their banking system, the $27b the Saudis added to their non-reserve foreign assets in 2004 and $64 the Saudis added in 2005.   Indeed, if you include the transfer of reserves to China's state bank and all Saudi central bank foreign assets, global reserves actually grew by more in 2005 - roughly $670b - than in 2004 (roughly $640b). 

What happened?  How can the fall off in recorded inflows to the US be reconciled with the absence of a fall in global reserve accumulation?

That is a question that Sangeetha Ramaswamy and I explore in some depth in our global reserve watch.  It is available, alas, only for RGE subscribers.

But I can provide a few of the key points.

The fall off in recorded inflows is, in our view, largely explained by the change in the set of countries adding to their reserves.

The US data picks up virtually all dollar assets purchased by Japan's central bank, but only picks up only a fraction of the dollar purchases of China and the oil exporters.  Both are probably buying US bonds through London custodial accounts - and those sales show up in the US data as the purchase of US bonds by private investors (the custodian) in London.

Consequently, as reserve accumulation shifted from Japan and China (2004) to China and the Middle East and Russia (2005), recorded inflows to the US fell.  Consider the following two charts.

The first plots the rolling four quarter sum of central bank purchases of Treasuries (Red) against the rolling four quarter sum of (valuation adjusted) reserve increase (Blue).   There clearly has been a sharp fall in central bank purchases of treasuries (at least recorded purchases) since the third quarter of 2004.

The second chart plots the rolling four quarter sum of central bank purchases of treasuries lagged by a quarter.  The lag reflects Japan's practice of first building up its dollar bank accounts, and then buying Treasuries.    And it plots that sum (the white bar) against the rolling four quarter total for Japanese reserve growth (in blue), and rolling sum for non-Japanese reserve growth (red).

I think it is pretty clear that the surge in recorded central bank purchases of Treasuries matches up pretty well with Japanese reserve growth - and not-so-well with the growth in the rest of the world's reserves.

Obviously, the increase in global reserves that doesn't show up in the US data could be going to Europe.  Or even into emerging economies. 

I don't think there is good evidence suggesting those flows increased from $200b in 2004 to something like $400 b in 2005, if you base your estimate for dollar reserve growth off the US data.   I like to include the reserve growth that show up in the BIS data on central banks offshore dollar bank accounts.   But that still implies that the flow of "missing reserves" increased from around $100b in 2004 to a bit under $300 in 2005.  And I don't there is any good evidence suggesting that kind of pick-up in central bank purchases of non-dollar assets.

My conclusion:  Martin Feldstein is right.  

The 2005 US data significantly understates central bank financing of the US.   Central banks may be buying agencies and mortgage -backed securities and other riskier dollar-denominated bonds rather than Treasuries, but they are still holding their reserves in dollars and buying US bonds.  And still financing a big chunk of the US current account deficit.