The May TIC data

Sorry about the unimaginative headline.

I guess the TIC data was sort of blah. Net inflows were not as strong as in January or February, nor as weak as in March and April.

One thing I don’t understand. Why do so many people compare the TIC flows to the trade numbers, and say so long as the net inflow from the sale of long-term securities is bigger than the trade deficit, everything is fine?

OK -- the two data points are released close to each other, so there is a natural comparison. But conceptually, the right comparison is between net inflows and the current account deficit.

The current account deficit is a bit bigger than the trade deficit. I estimate the current account deficit will be around $100b bigger than the trade deficit this year, or about $8 b bigger a month.

And the sale of long-term securities is only one way to fund a current account deficit, so their are other potential sources of inflows. Current account deficits also can be financed by taking out a bank loan from abroad, with foreign direct investment, or by selling off your existing external assets.

However, recently there has been good reason to focus on the net financing raised by the sale of long-term US securities to foreign investors. Net banking lending has not been a major source of financing for the last year, and to lesser degree in q1, the net flow of FDI was out of the US, not into the US.

Bottom line: unless something changes from 2004, the amount of financing the US needs to raise from the sale of long-term securities needs to exceed the trade deficit by a decent margin.

Remember, in 2004, the US raised a net $804 billion in financing from the sale of long-term securities, far more than the $667 b US 2004 current account deficit.

This year looks a bit different.

Average net monthly inflows in 2005 from the sale of long-term securities have been around $64 billion, about their May level. That works out to $768 billion a year --or a bit less than in 2004. The 2005 current account deficit, however, is going to be far bigger than the 2004 deficit. I expect a deficit of $820-850 billion -- more than the US will get from the sale of long-term securities unless foreign demand for US assets picks up in the second half of the year.

On the other hand, China’s new interest in buying US assets may auger a broader resurgence of interest in buying US firms, and other forms of direct investment in the US. The resumption of (net) FDI inflows is one way to square this particular circle.

Another, of course, would be for US firms in Europe to be running down their euro-denominated bank accounts as they repatriated profits to take advantage of the US tax holiday, or selling euro-denominated securities. Running down your existing foreign assets is another way to finance a current account deficit.

Some additional (WARNING: VERY WONKY) points.

1) Reported inflows from foreign central banks are down substantially if you compare June 04 -May 05 v. June 03 v May 04 ($156 b v $252 b). I am a bit suspicious though. Japanese central bank purchases are clearly down, but it is not clear that overall central bank purchases have fallen by quite this much. I don’t doubt that they have fallen some, but I do wonder about the magnitude of the fall. The fall in recorded central bank purchases in the TIC data seems substantially larger than the fall off in central bank reserve accumulation. So some central bank purchases may not be showing up as central bank purchases in the US data.

2) Monthly net purchases of US assets by foreign central banks have averaged $8.7 b in the first five months of this year. If the June number matches the average, that works out to $52.2 billion in financing from the sale of US debt securities to foreign central banks the first half of 2005. Needless to say, that would be a drop off from 2004, when total annual sales were $236 billion. However, this number does not match up all that well with data on central bank reserve accumulation.

By my estimates, adjusting for valuation changes, the central banks of eleven major emerging economies added $235-240 billion to their reserves (the recorded increase, without any adjustment for valuation, was $180b), and the central banks of Middle Eastern oil exporters clearly also added to their reserves in the first half of the year. Adding in their reserve accumulation might push total reserve accumulation in emerging market economies in the first half up to $280 billion or more (if someone has the bloomberg number, do tell ... ). $280b would not be that far off the 2004 average total once adjustments are made for valuation changes, which added to the 2004 total.

Obviously, not all reserves are invested in dollar denominated securities issued by the US -- some are invested in euros, and some are deposited in the world’s banking system. But right now central bank net purchases of Treasuries recorded in the TIC data account for less than 20% of the (guestimated) global increase in reserves. That "feels" a bit too low.

3) Japanese investors remain on the sidelines, at least in the Treasury market. They bought a bit over $4.7 of Agencies in May and $2.7 b of corporate bonds, but sold $1.75 of longer-term Treasuries. For the year, the Japanese have sold about $5 billion in Treasuries; their holdings of short-term bills have fallen by $5.25 billion, and their net purchases of long-term notes only totals around $0.25 billion.

4) China’s central bank probably accounts for a large share of the overall increase in official holdings of US treasuries. (Recorded) Chinese purchases of long-term Treasuries this year total around $11.2 billion, and China has also increased its short-term Treasury holdings by $9.45 billion. Add that up, and the total is a bit over $20b (through May). Compare that with an overall increase in official holdings of Treasuries -- either the $8.5 billion overall increase in official holdings of both short and long-term Treasuries or the $24.7 billion increase in holdings of long-term Treasuries. Other central banks must have reduced their holdings of short-term treasuries, since overall central bank holdings fell even as the PBoC’s holdings increased (assuming that the PBoC accounts for most of the change in overall Chinese Treasury holdings).

Of course, if China’s reserves increased by $116b ($134 b on a valuation adjusted basis) in the first half, and by about $96 billion through May ($110 b or so on a valuation adjusted basis), a large fraction of the increase in China’s reserves continues NOT to show up in the US data. Recorded Chinese purchases of T-bills (data for T-bills is here), Treasury notes, Agencies and Corporate bonds through May totaled $38.35 billion -- about 35% of China’s underlying reserve increase. Chinese demand for US corporate debt, probably mortgage backed securities, continues to be strong -- $9.7 b YTD.

Note: my Chinese reserve accumulation numbers include reserves transferred to the state banks.

5) Norway is buying again. Its Treasury holdings now exceed their pre-March sell off levels. However, OPEC’s holdings of Treasuries (recorded holdings) are flat through May. And Russia’s net purchases of long-term Treasuries ($1.8 b ytd on a flow basis) are not that impressive; indeed, its combined purchases of Treasuries and Agencies ($4.3 billion ytd on a flow basis) lag well behind the reported increase in Russian reserves ($23b through May, $27 b through June -- roughly $28/$33 b on a valuation adjusted basis).

Don’t look to the US TIC data to find the savings surplus of the world’s oil exporters.

6) Through May, foreign holdings of long-term Treasuries increased by something like $153 billion. Central banks bought $24.6 b, and private investors something like $128.7 b. Geographically, Chinese and Japanese holdings of long-term Treasuries rose by only $11.5 billion (data here). The big increase in Treasury holding seems to have come from the Carribean ($54-55 billion increase in holdings of both short-term and long-term Treasuries -- I did not look at the long-term only flow numbers, in part because the Carribean data is combined with other Latin countries, only the combined stock data) and the UK ($30-31 billion increase in short-term and long-term Treasuries).

That hardly settles the question -- clearly the UK and the Carribean (read hedge funds) could only buy so many Treasuries because they themselves were getting financing from someone else.

While I doubt central banks have reduced their net new purchases of Treasuries by quite as much as the US TIC data implies, I don’t doubt for that central banks are buying fewer Treasuries than they did in 2004, and private investors are buying more.

All in all, the TIC data increasingly fails to match up well with the global current account surplus. We know Japan, the OPEC countries and Russia all have very large current account surpluses and thus are building up their foreign assets. But their growing foreign assets, by and large, are not showing up in the US TIC data -- that may reflect limits in the TIC data, or it may indicate that they are building up their bank assets or investing heavily outside the US. But strong Japanese and Middle Eastern demand for euros is hard to square with the euros fall unless europeans -- and Americans with euro holdings -- are fleeing the euro in droves ...

The last explanation -- American residents fleeing the euro and selling their existing euro denominated assets -- is consistent with US firms repatriating profits held in euros to take advantage of the tax break the US offered every bit as much as fallout from the refendum, and interest rate differentials that favor the dollar.

I don’t yet have a fully coherent story that matches the available data points. It is hard to draw a clean, straight line from countries with current account surpluses to the TIC data to the financing of the US current account deficit. Yet, given the size of the US current account deficit, we know that pretty much every region with a large current account deficit has to ultimately be contributing to the financing of the US, one way or another, directly or indirectly.

Final caveat -- Most of my numbers should be correct, but when you do calculations quickly, there are no guarantees.