Disappearing into London … The latest TIC data

I got my hopes up after the January TIC data. For the first time in a long time, the TIC data matched what I though I knew about the global flow of funds. Central bank reserve growth was very strong in January. Recorded official inflows were strong. The data matched.

It couldn’t last.

For every January, there is a February. If you believe the US TIC data, the world’s central banks stopped buying US assets. Just stopped. Official purchases (net) went from positive $78.3b in January to negative $9 billion in February. That is kind of like how foreign investors stopped buying “private” US mortgage backed securities and CDOs last summer.

A $90 billion swing in monthly capital flows is huge. It is hard to square with the notion that central banks are a stabilizing force in the market.

Of course, it is also hard to square with a lot of other data. If you believe that central banks were net sellers of US assets in February, well, you probably shouldn’t be reading this blog.

Official flows likely did fall off a bit in February -- at least relative to their torrid January pace. Global reserve growth seems to have slowed a bit. But global reserve growth -- and I suspect official purchases of US assets -- didn’t fall by anything like the TIC data indicates.

The TIC data for example shows that official investors reduced their Treasury holdings by $6.4b (short and long term) and their agency holdings by $4.4b. However, the Treasuries the New York Fed held in custody for other central banks rose by $13.34b between January 30 and February 27. The Agencies held by the New York Rose by $15.49b. Net sales of $10.8b or net purchases of $28.8b. Take your pick.

Anyone trying to write a “central banks and China stopped buying Treasuries story” also might want to look at the rise in central bank holdings at the New York fed between February 27 and April 3 ($67.27b) and the rise in central bank Treasury holdings over the same period ($29.49b). Central banks bought a lot of US debt in March.

The TIC data shows that China didn’t increase its US holdings at all in February. Net long-term purchases of $10.73b were offset by net short-term sales of $10.64b. Total Treasury holdings fell by $5.7b, and Agency holdings only rose by $1 billion.

That makes no sense. China’s reserves rose by something like $47b in February after adjusting for valuation gains. Keeping the dollar share of China’s reserves at around 70% would have required China to buy about $40b of dollar-denominated debt. Even if you think China is diversifying at the margin, it bought some dollar-denominated assets and some US debt. (more detail follows)

The Chinese flow that doesn’t show up in the US data may be going through the UK ($23.2b of Treasury purchases and $$19.4b of Agency purchase), or perhaps through Hong Kong (total purchase of $16.9b – mostly from a rise in short-term holdings and equity purchases).

The same broad story holds for Russia. It reduced its recorded short-term holdings by $10.1b (mostly by reducing its short-term agency holdings by $9.6b). Its $4.65b in long-term purchases didn’t offset the fall in its short-term claims.

Of course, it likely bought more long-term bonds than show up in the US data, as – like China – it often buys through London.

The rise in the Gulf’s claims – up around $5b, counting the $3.3b rise in its short-term claims – is rather small for a region pulling in $25-30b a month more than it needs to cover its import bill from $100 oil. $5b would be small even in relation to the $12b rise in the foreign assets of the Saudi Monetary Agency.

Only Brazil, which bought $9.4b of US assets, makes sense given all else that we know.

Korea’s $2 billion purchase of equities (KIC/ Merrill) showed up cleanly in the US data. Kuwait’s purchase of Merrill didn’t show up in the data for the Asian oil exporters. Go figure. Singapore incidentally offset the $3.6b it spent on US equities (including bank stocks) by cutting its Treasury holdings by $5 billion. Its net US holdings – as measured in the US data – fell by between $0.8 and 0.9 billion.

(all the detailed data I used can be found here, though it helps to know where to look)

I think the basic story in the data is simple: a host of big players (Norway as well as China and Russia) cut their short-term holdings – which had increased dramatically over the past few months – and reallocated their funds into somewhat longer-term securities. The fall in their short-term holdings registered in the US data. However, the US data didn’t pick up all of their long-term purchases.

Global reserve growth was weaker in February than in January, but it was still quite strong.

Finally, remember that the rise in the UK’s holdings of Treasuries and Agencies is deceptive. Every survey has revised the UK’s holdings of Treasuries down. The last one was no exception.

The TIC’s division between official and private simply isn’t to be trusted.

It always understates official purchases.

It also consistently understates Chinese purchases.