This is Brad Setser once again. I am back from my spring break. I want to thank Dr. Frankel and Rachel Ziemba for taking my place last week.
No one in their right mind paid much attention to the TIC data release last Monday. Too many other things were going on. A major broker-dealer doesn’t come close to collapse every day.
I also would guess I am about the only person who finds the release of the Treasury’s survey data on foreign portfolio holdings interesting. It was released a bit earlier than I expected - at the end of February rather than the end of March. But I didn’t notice until I checked the TIC data -- and it doesn’t seem like many others have noticed either.
Both data releases tell the same fundamental story. The US now relies very heavily on foreign central banks for financing.
The January data also hints at another important but less obvious story, namely that central banks seem to be less willing to take credit risk than in the past.
So long as they are piling into safe US assets, central banks are contributing the "liquidity" to a market that doesn’t need any liquidity. They are helping to push Treasury rates down. And their activities, while rational from the point of view of conservative institutions seeking to avoid losses (beyond those associated with holding the dollar), also may be aggravating some of the difficulties in the credit markets. Private funds fleeing the risky US assets for the emerging world generally end up in central bank hands and currently seem to be recycled predominantly into safe US assets.
The January TIC data
In January, official investors - central banks and sovereign funds - provided the US with $75.5 billion in financing. Annualized, that is about $900b. That’s huge. It is also more than the US current account deficit. Central banks and sovereign funds are effectively financing the runoff of some private claims on the US. If the US were an emerging economy, that might be called "capital flight."
$53.4b of the $75.5b in overall official inflows came from the purchase of long-term US debt and equities. $22.1b came from a rise in short-term claims (the $15.2b increase in short-term Chinese claims likely explains most of the overall rise in short-term claims).
That $53.4b in long-term inflow was concentrated at the two poles of the risk distribution: Official investors purchased $36.1b in Treasuries, next to no agencies (*), sold corporate debt and bought $13.9b in US equity. This is what an anonymous (but well informed) commentator here called a barbell portfolio. Buy safe stuff or buy risky stuff but don’t buy much in between.
* The $20b in "private" purchases of Agencies in January is a bit suspicious. It might well reflect additional central bank purchases.
Where did the $13.9b in portfolio equity inflows come from. It isn’t hard to tell: $7.54b or so likely came from Singapore (Merrill? Or Citi) and $6.2b came from the Gulf (Citi?). Given how quickly the market value of Bear collapsed, some sovereign funds must now be more than a bit nervous ...
There likely will be more official equity purchases in the February data. I don’t think all the big announced transactions have been reflected in the data yet (Korea’s $2b investment in Merill for example) .
But if I had to guess, I would bet that current official purchases are overwhelmingly weighted toward super-safe assets. There are hints of that in the January data. China stopped buying Agencies, preferring Treasuries and short-term debt. Korea seems to have stopped shifting into Agencies. In total, nearly $60b of the $75.5b in total official inflows went toward Treasuries and short-term deposits and securities.
If I am right, then the official sector -- foreign central banks and sovereign funds alike -- isn’t coming to the rescue of the credit market. Indeed, by buying only safe assets at a time when private demand for risky assets has disappeared, the official sector is adding to the current market dislocations rather than reducing them.
For example, the absence of say Chinese demand for the assets Bear wanted to sell in order to cover its maturing obligations is one reason why Bear had to turn to the Fed for financing last Friday morning to avert bankruptcy - and ultimately had to sell at firesale prices to JP Morgan.
As large as the $75.5b in central bank and sovereign fund purchases in January was, it may well be an underestimate. Central banks and sovereign funds added - by my estimates -- at least $160b to their assets in January. China alone accounted for $55b after adjusting for valuation changes; the oil exporters likely chipped in heavily as well.
$75.5b is less than half the global total. If that is really all the dollars the world’s official investors bought, then they are diversifying away from the dollar and adding to market dislocations in another way. Keeping the dollar’s share of official portfolios constant would imply putting at least 2/3s of the overall increase into dollars - or a sum above $100b.
If that seems to big to be reasonable, good. The numbers for official asset growth right now are truly shockingly large. I have trouble believing them myself. But trust me, they check out. Just look at the latest data out of China.
Moreover, the survey data once again has revised the United States’ estimates for official demand up.
The June 2007 survey
The numbers speak for themselves:
Official purchases of Treasuries in the TIC data from July 2006 to June 2007: $68.9b.
The change in official holdings of Treasuries reported in the survey data: $238.9b
Official purchases of Agencies in the TIC data from July 2006 to June 2007: $132.0b.
The change in official holdings of Treasuries reported in the survey data: $277.7b
Official purchases of corporate bonds in the TIC data from July 2006 to June 2007: $31.5b.
The change in official holdings of corporate bonds reported in the survey data: $2.3b
The Survey data implies that the official sector increased its holdings of long-term US bonds by $518.8b between mid 06 and mid 07. That is a record. It easily tops the peak inflows during heavy Japanese and Asian intervention in late 2003 and early 2004. From June 2003 to June 2004 central banks only added $332b of long-term bonds to their portfolios (including around $270b of Treasuries). From June 2005 to June 2006 the increase in central bank’s holdings of long-term debt was $344 (including about $160b of Treasuries and $150b of Agencies).
The scale of the gap between the monthly TIC data and the survey suggests that studies based on the monthly TIC data are likely producing a false impression.*
Official inflows have not tailed off over the past few years. The TIC data is just picking up fewer of the flows. From mid 2006 to mid-2007, the TIC data picked up $232.4 of the $518b increase in the survey, or less than 1/2 of all official purchases.
*The BEA’s quarterly data on official inflows is a bit better - largely because the q3 and q4 data points for 2006 were revised up to reflect the results of the 2006 survey (the BEA smoothes out large data changes). Total long-term official inflows in the BEA data series were $419b ($151.2 Treasuries and $232.4 Agencies) - or about $100b less than in the survey.
The gap between the TIC data and the survey data is around $285b. Some early work has led me to estimate that the survey would revise official holdings up by $340b relative to the TIC data. I wasn’t too far off. My estimate assumed that the ratio between my estimate of total dollar reserves and official holdings in the survey would stay constant at its June 2006 level.
Official holdings of short-term debt securities fell between June 2006 and June 2007 -- but the US data suggests that "other" short-term official claims rose. Total short-term official claims on the US were flat. Offshore dollar deposits -- as reported by the BIS -- increased by about $117b from end June 2006 to end June 2007.
Finally, the survey showed a $51.6b increase in official equity holdings -- up a bit from the roughly $40b increase in the last few surveys. But some of the rise likely reflects valuation gains. Investors in the Gulf accounted for $28.8b of the overall increase.
Sum it up and total official dollar holdings -- at least those than can be identified -- rose by about $685b from end June 2006 to end June 2007. That is a large sum. But it isn’t all that large relative to the roughly $1150b increase in central bank reserves over this period (that total would be more like $1250-1300b counting the increase in sovereign wealth funds). And it is a big lower than the $825-850b or so I estimate the world’s central banks needed to buy to keep the dollar share of their reserves constant.
Call it evidence of diversification.
Or call it evidence that more central banks are making use of outside fund managers.
The survey has led the Treasury to revise China’s estimated holdings of Treasuries at the end of June 2007 up from $405.5 to $477.5b. That is over $70b in additional Chinese purchases over four quarters, or roughly an additional $5b a month.
The UK’s holdings by contrast were revised down from $192.9b to $49.8b. Gulf and Caribbean holdings were both revised up.
Throw in China’s huge purchases of Treasuries in January ($15b, mostly long-term) and it is clear that China hasn’t stopped lending to the US.
This shouldn’t be a surprise - at least to readers of this blog. There has been a very consistent pattern of revisions to the Treasury data series
That said, the overall revisions to China’s holdings were a bit smaller than I expected.
According to the survey, China’s recorded holdings of US long-term assets increased by $217b between end-June 2006 and June 2007. Total flows in the TIC data were about $127b - for an overall increase of $90b
Treasury holdings were increased by $71b, Agency holdings by $60b and equity holdings by $24b. It seems quite likely that SAFE now has a small US equity portfolio.
But the survey data also led me to reduce my estimate of China’s corporate bond by about $63b. The flow data suggested an increase of about $32b, while China’s recorded stock of US corporate bonds fell by $31b (from $58.5b to $27.6b). Corporate bonds are tricky because they amortize, but the size of the downward adjustment here was still something of a surprise.
The $217 billion increase in China’s holdings of US long-term US assets (and Xb increase in short-term claims ... ) should be compared with the $395b increase in China’s reserves. Actually it should be compared to the $373b increase in China’s reserves once valuation gains are stripped out. That works out to just under 60% of the increase in China’s reserves -- down from previous years.
But looking just at China’s reported reserves is deceiving. My analysis of the foreign currency balance sheet China’s banks suggests that they added $41b to their foreign assets from end-June 2006 to end June 2007 -- and the same data shows a $30.5 increase in the state banks’ holdings of portfolio investment abroad over the same time period.
If the increase in China’s known US assets (short-term claims were essentially flat) is compared to my estimate that -- including the state banks -- China’s official asset growth from mid-06 to mid -07 was around 413b with valuation adjustments, the identified increase is only 50% to 55% of the total
That could mean China is diversifying away from dollars.
That could mean China is making greater use of third party fund managers.
Or that could mean that the US data doesn’t pick up the purchases of China’s state banks, purchases that could well be channeled through Hong Kong. Moreover, the state banks could have been buying tranches of complicated offshore structures -- which really wouldn’t register in the data.
I would bet that China’s banks are part of the explanation. But there is also a hint of diversification in the data. That though is a subject for another post, one filled with illustrations ...
One last point: the new survey data -- together with the TIC data -- imply that Chinese holdings of US assets topped $1 trillion in December of 2007. Chinese holdings of US debt topped a trillion in January 2007. Congrats, I guess.