I am back in the United States, and ready to get back to work.
Before turning to the TIC data though, let me first thank Michael Pettis for filling in here so ably once again. I cannot overstate how much I appreciate the ability to outsource (to China even) the responsibility for putting up new content every so often.
Foreign investors were net buyers of US long-term assets in September, unlike in August. And foreign purchases ( after adjusting for the repayment of principal on asset-backed securities) of US long-term assets were even slightly larger than net US purchases of long-term foreign assets.
But the main story in the September data is that capital flows to the US remain very weak. At least to my ming, the right headline for the September data is the continued absence of foreign demand for US assets, not the (relative) improvement from August.
- Net purchases of long-term US assets ($26.4b) remain well below the roughly $65b a month needed to finance the US current account deficit.
- Net (private) long-term inflows remained negative (-$2.1b), as private US investors took more money out of the US ($29b) than private foreign investors brought in ($26.8b).
- The positive overall number comes entirely from recorded official – i.e. central bank – inflows. And there is good reason to think that the US data understates real official inflows. Some of the $22.4b in Treasuries sold to the UK in September were bought by (private) banks and hedge funds looking to hoard liquidity, but some were also likely bought by central banks.
- If – as I think likely – the adjustment associated with the payment of principal on asset-backed securities can largely be assigned to the private rather than official side of the ledger, net private inflows were even lower – negative $17.5b. Official investors hold asset backed securities, but they often use private fund managers for this portion of their portfolio. That is one reason why the US data understates official inflows.
Once the adjustment for principal payment is made and short-term flows are added in, net flows were negative in September. Negative $14.7 to be precise. And once short-term flows are added in, net (private) flows were really negative in September. Negative $27.8b. That is better than the negative $129.6b net private flow in August, but it is still a rather substantial outflow of private funds from the US.
Bottom line: private demand for US financial assets has disappeared. In emerging market terms, the US has experienced a sudden stop.
Consider the change in (net) demand for US financial assets between q2 and q3. In q2, net inflows were, according to the TIC data, $237.4b, or about $950b annualized ($194.1b in recorded private inflows and $43.4b in recorded official inflows). In q3, net inflows were negative $82b, or negative $328b annualized (negative $112.2b in private flows and positive $30.2b in official flows).
The overall swing was rather large. Net inflows -- really net non-FDI flows -- fell by $320b between q2 and q3. Annualized that is a huge number, $1280b.
The main source of that swing is a huge fall in net private flows – and I caution that the private inflow data includes a significant amount of official purchases. Net private flows swung from a net inflow of $194b in q1 to net outflow of $112b in q3. That is a swing of over $300b in a quarter ($1200b annualized).
I don’t have quarterly data for the emerging markets back that far, but the IMF's annual swing in net private flows to the emerging world back in 1997/1998 was far smaller.
UK purchases of long-term debt other than Treasuries fell from $178.95b in q2 (and $136.4b in q1) to $17.7b. UK purchases of US corporate debt fell from $94.4b in q1 and $98.7b in q2 to $16.7b in q3.
Whether that represents a fall in demand from central banks diversifying out of Treasuries (unlikely), European pension funds looking for a bit of yield (more likely) or London based SIVs and conduits operated by US and European banks no longer able to access the US money market for funding (most likely), it is a big fall.
Given this huge swing in capital flows, it really isn’t a surprise that the dollar has collapsed. I was up late on Friday night (really Saturday morning – I had a classic case of jet lag) and I heard a 30 minute debate on the BBC (Jessup of Capital Economics v Redeker of BNP Paribas) over the dollar.
I guess is a signal that the dollar is now weak enough (against the euro) that it might be due for a rebound. If not now, then over time.
Goldman’s Jim O’Neill certainly thinks so – and he is a long-term dollar bear. I tend to agree. At $1.45 to $1.50, the US trade balance with Europe should adjust quickly -- if only because Europeans will fly to the US to buy European and Asian goods sold at a lower price in dollars than in euros by European and American MNCs. That at least helps the US services balance …
On the other hand, it does seem likely that the Gulf will eventually end its tight dollar peg – a move that is unlikely to be considered dollar positive.
The details of the TIC data – which show the purchases by country as well as the official/ private split -- raises more questions that it answers.
The official/ private split. Total official inflows in Q3 were only $30.15b, with most of the total coming from a $27.4b rise in short-term on the US. Foreign central banks and wealth funds only bought $2.8b of long-term US debt.
Or at least that is all the US data shows. The total is low relative to an (estimated) $200b or so in global reserve growth in q3. China alone added $80-85b to its reserves (after adjusting for valuation effects) and probably another $13b to its sovereign wealth fund. And a host of other countries – led by India and the Saudis – also increased their reserves/ central bank assets.
Global reserve growth slowed in q3 as private inflows to the emerging world fell (they picked up again in October, big time). But the fall in the pace of global reserve growth is far smaller than the fall in recorded inflows to the US.
There also was a very clear shift away from long-term US debt. The $2.8b in central bank purchase of long-term US debt in q3 was well down from $58.4b in recorded purchases in q2 and $39.9b in q1.
BRIC inflows in q3. The total for the BRICs – at least the recorded total -- is instructive. The BRICs increased their short-term claims on the US by $61.65b in q3 while adding only $16.2b to their long-term claims. And $15.6b of the $16.2b in long-term purchases came from Brazil.
I suspect that buy and hold type investors decided that they didn’t want to look in low US long-term rates. They would rather hold short-term deposits/ securities than buy longer-term US assets that yield less.
China. Much will be made of the fact that China’s net purchases of long-term US treasuries was negative 17.2b in q3 -- and its total holdings fell by $9.6b, to $396.7b, as the increase in China’s holdings of short-term t-bills didn’t offset the fall in its holding of long-term notes.
Indeed, China only bought $2.1b of US long-term debt in q3, as its $13.2 b in (recorded) Agency purchases and $6.2b in recorded corporate debt purchases only just offset the fall in (net) purchases of US Treasuries.
But three notes of caution are in order.
- First, China’s short-term holdings increased by $33.65b. China clearly built up a lot of cash. Looking just at the long-term data misses the story.
- Second, China now buys a lot of US debt through London. Some of the UK’s $72.0b in q3 Treasury purchases likely were then sold on to SAFE. I do not doubt that China is reducing the share of Treasuries in its portfolio (as are the Bank of Japan and the Bank of Korea). But the scale of the shift may be somewhat smaller than flow data suggests.
- Finally, the US data almost certainly isn’t capturing all China’s holdings. The $35.8b in recorded inflows in q3 is small relative to the increase in China’s reserves ($80-85b after adjusting for valuation), the combined increase in China’s reserves and the CIC ($95-100b) let alone the $180b the Financial Times Richard McGregor thinks China added to its total foreign assets in q3. China’s banks and state companies are supposedly sitting on a lot of dollars right now (there is an allusion to the pressure on Chinese banks to hold more dollars in this WSJ article).
Russia took a similar approach to China. Its short-term holdings rose by $19.6b in q3 ($11.2b in September), while its (recorded) long-term holdings fell by $0.7b.
Brazil’s short-term holdings rose by $7.6b and its long-term holdings by $15.6b (with almost all of the increase from Treasuries). The total $23.2b increase matches the overall increase in its reserves.
India’s rapid reserve growth though wasn’t matched by any increase in its recorded US holdings. Short-term claims rose by $0.8b, long-term claims fell by $0.8b.
And there isn’t much sign of the Gulf’s money in the US data either. Long-term purchases from the Gulf were only $0.8b (with a fall in Treasury holdings). Short-term holdings rose by $3.1b, but the $3.9-4.0b total increase is small relative to the region’s current account surplus and soaring assets..
Indeed, total Asian purchases of US assets in q3 are a bit too low to be believed. In aggregate, Asian was a net seller of US long-term assets to the tune of 6.7b in q3, with $67.4b in net Treasury sales.
I don’t buy it. Not really.
Asia may have been selling Treasuries to American and Europeans banks scrambling to increase their liquidity. Asian central banks have clearly been shifting toward Agencies. But the global current account doesn’t balance without a significant net flow from Asia to the US. The world’s biggest surplus region has to finance the world’s biggest deficit region. Asia after all includes most of the world’s current account surplus, whether from the goods producers in East Asia or the oil exporters in West Asia.
Some of the gap is explained by the rise in Asian short-term claims. But not all.
The TIC data is the base for any attempt to understand the pattern of global capital flows.
But in my view, it also to be assessed against other variables – current accounts surplus, reserve growth and the like -- to understand what it captures and what it doesn’t capture.