The flight from risky US assets

It is hard to focus on data from over a month ago when a large emerging economy’s stock market is down double digits and the Fed is debating whether or not to extend a lifeline to the largest US insurance company. But the TIC data is stunning in its own right.

It tells a simple story: demand for risky US assets disappeared in the month of July. That continues a long-standing trend. But that trend intensified significantly. And I suspect its intensity increased even more in August.

Among other things, the TIC data challenges the common argument that sovereign investors have been a stabilizing presence in the market. Best I can tell, sovereign investors joined private investors in retreating from all risky US assets in July, and thus added to the underlying distress in the market. I don’t fault sovereigns for limiting their risk. It has proved to be a sound financial choice. But I also find it hard to square their (inferred) actions in the market with many claims about their behavior.

The TIC for July pains a very clear picture: Treasuries were the only US asset foreign investors were willing to buy. Foreigners bought $34.3b of long-term Treasuries, while selling $57.7b of Agencies, $4.2b of corporate bonds and $5.2b of equities. On net, foreigners sold about $25b of long-term US assets.*

That would normally make it hard to sustain a large current account deficit. The US still needs roughly $60b a month in net inflows to cover its external deficit. Net sales of foreign assets of $32b provided some financing -- but not nearly enough to cover the outflow of short-term funds. $75b in net outflows isn’t exactly a good sign, even if the dollar’s rebound suggest more flows (perhaps from large US sales of foreign assets) in August.

The same basic trend is apparent in the data for the 12ms through July 2008, which can easily be compared to the 12ms through July 2007 -- think pre-crisis and post-crisis.

After the crisis, foreigners have bought roughly:

$350b of long-term US treasury bonds, and another $125b of short-term bills -- for a total of $475b of US Treasuries. That explains how the US has financed its fiscal growing deficit. Foreigners also bought around $150b in Agency bonds, $210b of corporate bonds and $55b of US equity ($20b excluding the SWF capital injections into the banks and broker dealers).

Before the crisis, foreigners bought roughly:

$205b of long-term Treasury bonds (and reduced their holdings of bills by $10b), $285b of long-term Agencies, $540b of long-term corporate bonds and $210b of US equity.

Notice a trend?

Sovereign wealth fund flows into equities clearly have been trumped by a broader retreat from risk, including a retreat by sovereign investors.

There are a couple of other important swings in the data.

US investors dramatically reduced their net purchases of foreign assets. US net purchases fell from $280b to $95b. That helped. US capital outflows have to be financed by capital inflows.

And US banks dollar liabilities fell by $400b -- producing a huge net outflow that offset much of the inflow.

As a result, net "TIC" flows for the last twelve months are only $210b -- well below the current account deficit. Some of that gap could be covered by net FDI inflows, but in general FDI inflows and outflows match up pretty well, so it is hard to see that large an inflow from FDI. The remaining gap is an error term -- we simply don’t know what all is going on.

That net flow can be broken into a $67.7b net private outflow (counting short-term flows) and a $277.8b net official inflow.

Two points here:

The US data consistently understates official inflows. We know $280b is too low for a host of reasons -- not the least the fact that the Fed’s custodial holdings increased by more. The past "survey" revisions have consistently revised official flows up and private flows down. There is no reason to think that this has changed. Indeed, the past revisions would suggest the entire $260b in "private" purchases of Treasuries over the past 12ms in the TIC data are likely official inflows, and that the $90-95b in private purchases of Agencies are too. If those flows are allocated, the net official inflow is over $600b -- and the net private outflow is correspondingly larger.

The US data doesn’t seem to be capturing all the flows associated with the unwinding of the shadow banking system. That at least is my perception. I don’t have any other good explanation for the gap between identified flows and the US deficit.

Now to the country data.

I found one country with large reserves that added to its long-term Agency holdings in July: Hong Kong. Everyone else was a net seller. China, Russia, the Gulf ("the Asian oil exporters"), Brazil, Korea, Singapore, and Japan. So was the UK -- which likely indicates a further fall in official demand. The TIC data indicates a huge reallocation by official investors from Agencies to Treasuries -- and a far larger reallocation than showed up in the FRBNY accounts.

Let’s look at five specific countries where central banks and sovereign funds typically generate a sizable share of the flow.

China added $20.4b to its short-term holdings and Treasury portfolio, while cutting its holdings of long-term agencies by $3.4b. That strikes me as a flight from risk.

China was still buying US corporate debt in July, but in smaller amounts than in previous months. This isn’t a flow I understand well. Chinese purchases of corporate debt have increased recently -- and that presumably has driven the rise in overall "official" purchases of corporate debt. But there were also substantial Chinese purchases from mid-2006 to mid-2007. The mystery: these flows weren’t match by a rise in stocks in the survey data. Count me confused.

The Asian oil exporters (i.e. the Gulf) added $7.5b to its short-term holdings, and another $0.8b to its long-term Treasuries -- for a "flight to safety" flow of $8.3b.

China and the Gulf account for a large share of the global surplus, so their flows matter.

Russia cut its long-term agency holdings by $0.4b while adding $3.1b to its long-term Treasury portfolio. It also added $5.7b to its short-term Treasury holdings while cutting its holdings of other short-term securities (think Agencies) by $4.4b. One great irony in the data is that in the month before the conflict with Georgia broke out, Russia provided $8.8b in net financing to the US Treasury. Recent pressure on Russia’s reserves at least has eliminated that bizarre flow.

Korea -- which many claimed might be running out of liquid assets -- seems to have reduced its Agency holdings far more rapidly than its Treasury holdings: its long-term Treasury portfolio fell by $1.1b, its long-term Agency portfolio fell by $3.6b and its holdings of corporate debt fell by $1.4b. That strikes me as sound liquidity management; it didn’t sell off its most liquid asset first.

Finally Singapore. Singapore cut its long-term Agency holdings by a modest $0.2b while adding $1.1b to its long-term Treasury holdings. But the big story in the data is that it sold $5.2b of US equity. There is no way to know for sure that these sales came from the GIC or Temasek, but it is certainly possible that they did.

All in all, I saw a lot of evidence of a sovereign flight from risk at a time when the market for risk assets was under stress. At the global level. And at the national level.

I’ll try to flesh this out with a few charts later.

* One data point. The detailed data for Agency purchases here doesn’t seem to match the summary data. I used the detailed data. Agency data should be adjusted for ABS repayment, but that adjustment is complex and it doesn’t affect the basic trend, so I reported the unadjusted number.