Is Brazil’s central bank betting on a US slowdown?

In August, Brazilians – the central bank and private investors – bought $11b of US Treasuries.   That is a lot.   Back in 2002, I would never have thought Brazil would provide (annualized) $120b in credit to the US Treasury in any month …   

Here is one way of thinking about the magnitudes.  Brazil’s 2002 IMF bailout was in the $30b range.  Suppose 1/3 of the IMF’s useable funds come from the US.  Then  Brazil provided as much financing to the US government in the month of August as the US provided Brazil’s government (through the IMF) back during Brazil’s entire crisis.   The times they are a changing.

Actually, Brazilian purchases of US Treasuries are more like a swap than outright financing of the US.  US and European investors looking for yield buy high-yielding Brazilian debt, and Brazil’s central bank, which is intervening to keep the real from appreciating, uses the funds to buy US debt.     That is a trade that – at least so far – has generated big profits (dark matter, in other words) for American and European investors. 

Back to August.   How did Brazil manage to afford $11b in US treasuries in a month?

First, Brazil’s reserves increased by a bit less than $5b in August.   Second, Brazilian central bank/ banks reduced their short-term claims on the US by $2.7b or so.   So even Brazil's central bank accounted for all the fall in short-term claims, it would have only been able to generate $7.5b or so of $11b in purchases ....  Yet even if Brazil's central bank didn’t do all the buying, it rather clearly invested a decent chunk of growing reserves in Treasuries and shifted a significant sum from short-term investments to longer-term investments. 

That is the sort of thing you do if you think US rates are about to fall.    In other words, it was the sort of bet that makes sense if you think Mr. Recession in 2007 is going to be at least ½ right, and not just about the third quarter.

I rather suspect Brazil was not the only central bank making this kind of bet in August and September.  There were big inflows to the US fixed income market in August, and given who has spare cash these days, that usually means big central bank and oil fund purchases.

More importantly, there is pretty good evidence that central banks and oil funds were building up their ammunition in the first half of the year.

The latest BIS data (Table 5c) shows a rather remarkable buildup in central bank deposits in the international banking system in the first ½ of the year.   Their dollar, euro, pound and yen claims increased by about $140b (after adjusting for valuation effects).    That compares to $150b or so in recorded inflows to the US ($40b of those were bank claims – so there may be some double counting) and $350b or so in total global reserve growth.     

Central bank deposits in the world’s banks were rising fast, while their purchase of securities (especially US Treasuries) tailed off a bit.   It seems like they were waiting for the Fed to end its rate hiking cycle before buying longer-term US debt.

No wonder the world was flush with liquidity.

The most extreme example of maturity shortening comes from a country that keeps most of its oil windfall in its government pension fund, not its central bank: Norway. 

From the TIC data, we know Norway cuts its long-term dollar portfolio by $18.4b, while adding $33.5b to its short-term claims in the first half of the year.   That one of the many tidbits that emerged out of RGE’s petrodollar watch (warning: this requires an RGE subscription).

And judging from the fall in Norwegian short-term claims since the end of June, they too have been lengthening the maturity of their portfolio in the second half of the year … 

Incidentally, for those who care about this sort of thing – if you add the increase in central bank dollar deposits in the BIS data to recorded central bank inflows to the US and assume that there is no double counting, central banks added at least $235b to their stock of dollars in the first half of the year. 

That is likely a minimum.   Neither the BIS bank data or the US TIC data would pick up central bank purchases of US securities made through London custodians.   And neither the BIS data nor the USdata is picking up net inflows from the Gulf on a scale that is comparable to the buildup of the Gulf's foreign assets implied by their current account surplus..  

Still, $235b works out to around 67% of my estimate for (valuation-adjusted) global reserve growth (a bit under $350b, counting all SAMA foreign assets but not most oil investment funds) in the first half of 2006.    That suggests that the world's central banks weren't shifting out of dollars.   Indeed, it fits reasonably well with my estimate of $273b in total dollar reserve growth in the first half.

True reserve geeks will note that $235b is higher than the increase in dollar reserves ($119b) reported in the COFER data.  That is for one simple reason – the COFER data is incomplete.  It doesn’t pick up the increase in all Saudi Monetary Agency foreign assets.   And a lot of big players – China included – don’t report the currency composition of their reserves to the IMF.