The TIC data provides information about the most important way the US is financing its ongoing current account deficit: selling long-term securities to foreigners, or exporting debt. It is important to remember that there are other ways of financing a deficit: borrowing from foreign banks, for example, or attracting net FDI inflows. But since FDI flows have been negative or, at best, close to balanced, and bank flows tend to be small, most of the financing the US needs has come from the sale of long-term securities to foreigners.
What did the November data release tell us?
Above all, that the US is still are not getting net equity financing. Foreign purchases of US stocks surged in November, but remained smaller than US purchases of foreign stocks. There was a small (-1.6 billion) net outflow. Most of the net financing -- $82.6 billion of the $81.0 billion total -- came from the sale of debt. Sales of Treasuries totaled $32 billion; Agencies, $28 billion and Corporate debt $25.5 billion (US citizens only bought $2.5 billion of foreign debt).
Who was buying was all this debt? The TIC data suggests that private creditors bought roughly twice as much US debt as the world’s central banks -- $59 billion v $26.5 billion. One of my major themes is that US data often under reports foreign central bank support for the US debt market (for an explanation of why, see this Higgins and Klitgaard article). If $15 billion of the private purchases of US long-term debt was really foreign central banks in disguise, a rather different picture of November emerges. Private purchases would be roughly equal to central bank purchases: both would be in the $40-45 billion range.
I think that is case. "Real" central bank buying was almost certainly larger than $27 billion in November. The remainder of the post lays out my argument. But a warning is in order: the next section is weedy and data intensive.
1) A small discrepancy, but one that puzzles me: the "Agencies" held by the New York Fed on behalf of other central banks increased by $12.5 billion in November (Compare this release, with this one). The TIC data only registered a $3.5 billion increase. Conversely, the Fed’s custodial holdings of treasuries only grew by $11 billion, while the TIC data showed a $21 billion increase. I see how the TIC data could have a higher number than the change in the Fed’s custodial holdings, but have trouble understanding how it could have a lower one. (Help from readers is always appreciated). My gut tells me central banks bought more than $3.5 billion of the $28 billion in agencies sold to foreigners in November, in part because one very large Asian country whose reserves are increasing very fast is rumored to be quite keen on Agency bonds.
2) The other way to get at official support is to look at "creditor side data," i.e what happened to our central bank creditors reported reserves in November. The definitive statement of creditor side data comes from the BIS, but with a huge lag -- so working off creditor’s side data requires making a number of assumptions. We generally know how much central bank reserves increased in November, but we don’t know whether this increase came from the dollar’s fall v. the euro (which increases the $ value of central banks’ holdings of euros), the purchase of new dollar assets, or the purchase of new euro assets.
But let’s see what we can try to piece together. The reserves of 8 major emerging market central banks (the four Asian NICs, Russia, India, Malaysia and Thailand) increased by $46 billion in November. Some of that came from roughly 5% increase in the dollar value of their reserve holdings, but not all. Malaysia’s reserves went up by $3.5 billion in November alone; for the whole year, currency moves only increased Malaysia’s reserves by $2 billion, and not all of that gain came in November.
We don’t know what happened to China’s reserves in November for sure: the data is still not out on its web page. But we do know that it started November with $543 billion (roughly) and ended December with $610 billion. It is reasonable to think that a decent chunk of that $67 billion increase came in November -- indeed, tomorrow’s FT reports that the November increase was $31.5 billion.
That gives us $77.5-78 billion in reserve accumulation in Asia and Russia in November (looking only at the big economies). If China held $150 billion in euros (on the high side of most estimates, but who knows) going into November, valuation gains would have increased its reserves by about $7.5 billion. But at least $24 billion of its reserve increase still came from market intervention.
Let’s assume that the other Asian economies and Russia had a slightly higher stock of euro reserves, and thus enjoyed a slightly bigger valuation gains. Say $9 billion. That leaves $34 billion from market intervention. Put differently, even taking into account the potential for large valuation gains, nine central banks probably added $58 billion to their reserves in November -- a lot more than the $28 billion in reported central bank inflows to the US.
What explains the difference? One possible explanation is that the central banks put some of the dollars they were buying in the market into the world’s banking system, and private banks, in turn, used higher deposits from central banks to finance additional purchases of US long-term debt securities (The FT speculates about this possibility). One possible explanation is that central banks bought more US debt than the US data system recorded (likely). And one possible explanation is that central banks were buying euros, not dollars in November ...
I seriously doubt central banks used their growing reserves to buy $30 billion in new euro denominated securities and only $28 billion to buy new dollar denominated securities in November. A 50/50 split is a bit implausible, particularly since Asian countries were trying to keep their currencies from appreciating v. the dollar. A 75/25% certainly split is plausible, but that would imply $43.5 billion in net dollar purchases -- lots more than the US recorded.
All this only from looking at data from 9 of the world’s central banks, and only one big oil-exporter (Russia). Obviously, other oil exporters were adding to their reserves too ... as were some Latin countries.
My bottom line: either the world’s central banks are buying a lot more euros than anyone thinks, or the US TIC data underreported "real" central bank support for the US in November.
The amount of support for the dollar that potentially came from Asia in q4 should not be underestimated. China’s reserves increased by an increasible $95 billion inq4. $10 billion of that probably was from valuation gains, maybe more. But that still leaves $80-85 billion in reserve financing for the world from China alone -- over $25 billion a month, and $300 billion on an annualized basis. Few questions matter more for the global flow of funds that what what the PBOC does with its reserves.
One final note: I don’t mean to imply that ALL private purchases of long-term debt securities are really coming from central banks in drag, or from private banks playing around with a surge in central bank deposits. The reported Treasury holdings of Germany, Ireland, Belgium, Spain, Australia and Canada all went up in November -- clearly there are some private purchases (There was also a surprisingly large fall in bonds held in the Caribbean and a surge in bonds held in the UK -- something was going on with the data. Maybe some Caribbean holdings were reassigned to Europe.)
This data suggests that the US got funding in November, in some sense, from two sources: from private Europeans and Canadians who thought their currencies had already moved so far against the dollar that the next move in the dollar would be up, and from the central banks of countries who are still working hard to keep their currencies from rising against the dollar in the first place. And if -- as I think is plausible -- the US received $45 billion from foreign central banks, both directly and indirectly, then foreign central banks are financing about 3/4s of the (large) US trade deficit. Talk about vendor financing.