This morning’s TIC data release tell us two things:
One, the dollar can rally without any obvious supporting inflows. Net TIC flows in August (line 30) were essentially zero (-$0.4b). Foreigners (apparently led by official investors, but the data here is deceptive) were net sellers of US long-term assets, selling about $9b. Americans sold $23b of their foreign assets -- providing the US with a bit of financing. But things like amortization payments on the outstanding stock of Agency bonds (line 20) cut into the inflow from the sale of US assets abroad. Net long-term flows were essentially zero. And so were net short-term flows.
Don’t ask me to explain how the US has sustained a $120b trade deficit in July and August on the back of a $34b net outflow of funds over those two months in the TIC data. Something doesn’t quite compute.
Two, the argument that official investors are a stabilizing presence in the market needs to be marked to market -- or at least marked to the observed flow. And the observed flow suggests an enormous flight by official investors out of bonds with any hint of credit risk and into Treasuries. That hasn’t been stabilizing. It has added to the stress in the credit and inter-bank markets -- and ultimately contributed to the broad financial environment that forced the Treasury to step in and guarantee the Agencies and now most bank debt.
The evidence here is overwhelming.
The TIC data release indicates that official institutions sold $13b of long-term Agencies, reduced their holdings of short-term Agencies by about $9b and sold another $2b of corporate bonds and equities while buying $5b of long-term Treasuries and $12b of short-term Treasuries.
That though likely understates the swing, as most "private" purchases of long-term Treasuries and Agencies have recently been from the official sector (see the pattern of revisions that followed the June 2007 survey -- which found that official buyers accounted for nearly all foreign purchases of Treasuries and Agencies). Overall, foreign investors -- including official investors -- bought $35b of long-term Treasuries (and $55b of short-term Treasuries) while reducing their holdings of long-term Agencies by $30b and their total holdings of Agencies by close to $40b (the fall in short-term Agencies is on line 39 of this data release). Stabilizing that was not.
The country detail tells a similar story.
China bought $18.3b of long-term Treasuries and $4b of short-term Treasuries while reducing its holdings of long-term Agencies by $7.1b and its holdings of short-term Agencies by $3.3b. Think $22.3b to Treasuries (and another $5.7b into bank deposits) and at least $10b out of Agencies. China also sold "corporate bonds" in August. In the past it has been a significant buyer. Watch this going forward. I suspect (but don’t know) that Chinese state banks and SAFE may have significant holdings of the paper issued by large US financial institutions. China certainly has been buying some kind of US corporate debt recently -- and given that the state banks have been running down their foreign portfolio investments, by best guess is that it has come from SAFE.
The "Asian oil exporters" (think the Gulf) acted similarly. They added $1.4b to their long-term Treasury holdings, while cutting their holdings of long-term Agencies by $3b. Above all though, they piled into cash -- their total short-term claims increased by $11.7b (call it $12b). This data support Landon Thomas’ excellent reporting on the buildup of the cash reserves of various Gulf sovereign funds.
Russia was selling everything as its reserves fell, but it clearly was selling its Agency holdings faster than its Treasury holdings. It cut its holdings of short-term Agency paper by something like $8.6b (using other short-term custodial holdings as a proxy for Agencies)
The most recent TIC press release is interesting for another reason. The data on total flows in the 12 months through August 2008 and be compared to total flows in the 12 months through August 2007 (or even better, July 2007) to provide a fairly clean pre-crisis and post-crisis comparison.
Here as well two stories emerge:
a) Total demand foreign demand for US assets fell after the August crisis
Net flows in the TIC data went from $879b in the 12 months to July 2007 to $314b in the 12 months to August 2008 (line 30). Private flows (a number that includes a lot of official flows, notable "private" purchases of Treasuries and Agencies by private banks that are then sold to central banks during London trading) tell from from $665b to $18b. Official flows -- which the TIC data understates -- rose from $214b to $296b. The swing in long-term flows is a bit less pronounced but still substantial -- total long-term flows fell from $1252b in the 12 months to July 2007 to $798b in the 12ms to August 2008.
b) Demand shifted away from Agencies and Corporate bonds toward Treasuries.
This is where the long-term flows tell a dramatic story. Let’s focus on total foreign purchases rather than the private/ official split, which is known to be off. Just remember that almost all Treasury and Agency purchases have come from the official sector. In the 12 months to July 2007, foreign investors bought $217b of long-term treasuries (and sold $10b of short-term Treasuries) while buying $284b of Agency bonds, $540b of corporate bonds and $210b of US equities. In the 12 months to August 2008, foreign investors bought $389b of long-term treasuries and another $134b of short-term Treasuries, bringing their total purchases of Treasuries to $523b (that is how the fiscal deficit was financed ... ). Purchases of Agencies fell to $121 billion, purchases of corporate bonds fell to $193b (with almost $60b coming from official buyers, mostly in Asia) and purchases of equities -- counting the $35b or so of bank equity bought by sovereign funds -- fell to $97 billion.
The story is simple, at least in my view. Most foreign demand for US assets over this period came from central banks. And they increasingly only wanted to buy Treasuries.
UPDATE: Here is a graph of foreign holdings of long-term Agencies. Notice the recent fall.
The TIC data can be found here. This report draws on both the short-term stock data and the long-term flow data.