Brad Setser

Follow the Money

Cross border flows, with a bit of macroeconomics

Print Print Cite Cite
Style: MLA APA Chicago Close


Stephen Jen: only 60% of China’s reserves are in dollars.

by Brad Setser
November 27, 2006

Stephen Jen sometimes says the darnedest things.  Not about currencies.   But about reserves. 

His estimates for oil reserve growth have been consistently on the low side.   Norway, the UAE and Kuwait and a few other sheikdom’s all have sovereign wealth funds that manage their oil windfall, but a lot of other countries have managed the oil windfall the old fashioned way: at the central bank.     Russia and Saudi Arabia are the most prominent cases (and they are big).  But Nigeria, Libya and Algeria also have bulging central bank balance sheets.

Jen also thinks China only has $600b in USD assets: 

“China has US$1 trillion in reserves and possibly US$600 billion or so in USD assets.  (Incidentally, I think the market’s guess that 70% of China’s reserves are in dollars is possibly a bit too high.  While China may have had 70% of their reserves in dollars back in 2002, this ratio is closer to 60% now, I think.)  In any case, US$600 billion is a massive amount of dollar assets” 

I understand Jen views on the dollar, even if I don’t agree with them.  But I cannot quite figure out the basis for his current argument that only 60% of China’s reserves are in dollars. 

I have a bit of skin in this game.  I think I am the main source for now ubiquitous claim that "70% of Chinese reserves are in dollars."   Currency analysts are paid to get their currency calls close to right.  I, in some sense, am paid to get reserve calls close to right – or at least as close to right as is possible given the available data.  

I certainly don't know the exact composition of China's reserves.  But based on the available data, I am quite confident that China — counting the state banks along with the PBoC — now holds more than $600b in dollar denominated securities.   And I think it is quite likely that the PBoC itself holds more than $600b in dollar debt.  Caution – a massive data dump follows.  

We know that China had $524b in long and short-term US securities in June 2005, back when it had around $710b in reserves.     We know that China has added about $278b to its reserves between the end of q2 2005 and the end of q3 2006 (and a bit more since then). We know that it has bought about $132.5b (net) in long-term US debt since then ($48.5b in the second half of 2005, $84.1b in the first three quarters of 2006).    

That works out to Chinese holdings of $653b of US debt.    Minimum.

We also know that the TIC data has tended to understate Chinese purchases.    The increase in Chinese holdings reported in the survey from mid-2004 to mid-2005 was about $90b more than the reported flows.   

So I would say that we know that China held at least $653b of US debt securities at the end of q3, and quite possibly substantially more.   $750b is not out of the question.

Now the US data doesn’t distinguish between PBoC and “private” Chinese purchases … so not all of the $653b to $750b in Chinese holdings of US debt will be held by the PBoC.    If US TIC flow data is capturing all Chinese purchases and there are large private Chinese holdings, Jen could be right about the PBoC’s holdings. 

Some US debt certainly will be held by the banks as part of the $60b in reserves that have been transferred to three state banks as part of their recapitalization, and there is a lot of data suggesting that Chinese banks have been buying US debt with the proceeds from their IPOs.   

But Jen’s estimate still seems to be rather on the low side.  After all, the US data on securities holdings leaves out two additional sources of Chinese dollar exposure – its holdings of dollar-denominated emerging market debt (and bonds issued by the World bank and similar institutions) and its dollar denominated bank deposits. 

We know from the BIS data that Chinese banks (including the PBoC) have had between $95-100b on deposit in the international banking system (China has $120-125b in total deposits, and $25b or so of that comes from outside the financial sector).   And we know from the US data that about $20b of that is in dollars in US banks.   What we don’t know is what fraction of the other $80b is in dollars.   We also don’t know what fraction of the $100b or so China has on deposit with international banks comes from its state banks and what fraction comes from the central bank.

Combine the US data on securities holdings ($524b) with the US data on short-term holdings ($59b), and net out short-term securities ($40b), and we know, more or less, that China held $543b in US debt and dollar deposits in the US at the end of June 2005.   That works out to around 70% of China’s $770b in known reserve/ state bank holdings ($710b in reserves and the $60b that the PBoC had shifted to the banks at that time). 

That total assumes all Chinese holdings are official holdings, which is one source of error.   It also leaves out Chinese holdings of dollar-denominated securities issued by emerging economies/ the World Bank and Chinese offshore dollar deposits.

What about now?  Through the end of q3, China’s reserves increased by $278b – rising to $988.  That is the easy bit.  Estimating the growth in the offshore assets of Chinese banks takes a bit of work.  China, according to the IMF, shifted $14b or so to the banks using swap contracts in 2005 – and some of that likely was used to buy US debt in the tail end of 2005.  Chinese banks have been big buyers of foreign debt (maybe $45b in h1 and presumably more in q3 — $20b would be a conservative estimate), whether using domestic dollar deposits or IPO money.     Add it up, and Chinese banks probably had an additional $140b to play with on top of the government’s reserves — $60b from the recap and $80b from other sources.   That brings total Chinese reserves and external assets of the state banks up to $1128b.  Call it $1125b. 

China’s short-term deposits have neither increased or decreased since mid-2005.  It still has around $20b on deposit in the US (q3) and $100b on deposit internationally (as of the end of q2).   China’s holdings of US securities (recorded) rose by $133b.   So we know that China now has around $670b of US debt/ deposits in the US.    

That works out to about 60% of my rough estimate of the total external assets of China’s banks and the PBoC.  

But we know that the US TIC data has tended to understate Chinese purchases, so China likely has more securities than implied by the TIC data.  $133b of purchases of US debt out of an estimated $357b increase in China’s external assets (from both the state banks and the PBoC) is a bit on the low side, particularly when international bank deposits are not increasing. 

China’s holdings of dollar securities may be about $100b higher than one would estimate from simply summing up the flows, based on the gap between the TIC flows and the change in stocks.  And China also holds some dollar debt issued by emerging markets and presumably some dollar denominated bank accounts outside the US 

So Jen’s estimate consequently seems low to me.   So for that matter does the Deutsche Bank estimate in their monthly (also around 60%).

I think Andy Rothman of CLSA has it about right.   In the FT.

Andy Rothman … said Beijing had changed the composition of its reserves in recent years, not by selling US dollars but by buying less when making fresh investments.

“While a few years ago, maybe $75-$80 of every new $100 in reserves would go into dollar assets, now possibly only $65-$70 do,” he said.

He argued that China used to keep about $80-85b of every $100b in new reserves in dollars, and now that is more like $65-70b of every $100b of new reserves.   

Putting $65-70 out of every $100b in reserves will slowly lower the dollar share of China’s reserves over time.  But it still works out, I suspect, to a current dollar share of above 60%.    

One interesting question? Say China put $35 of every $100 in reserves in currencies other than the dollar over the summer, and only put $65 in dollars.   Is it still able to do so now, with the dollar under a bit more pressure?  Or are 80-90 of every 100 that China bought last week still in dollars? 

Another interesting question highlighted by Richard McGregor in the FT.:  Will the rising cost of sterilizing this reserve growth push China to change policies?    Or is it just another cost that the State Council is prepared to eat to keep the status quo going for a bit longer?


  • Posted by Guest

    Do you have something personal against Stephen Jen?
    Just wondering…

  • Posted by bsetser

    Nothing personal. Jen has been right on currency direction more often than he has been wrong, and that cannot be said for many others (including me). He has been off on the euro this year, but he is hardly alone.

    But I do think his analysis of reserve data has been sloppy — and given that he has the resources of a major investment bank to support his work, he could and should do better. He supposedly is one of the major fx strategists that SAFE listens to, so his views have some credibility.

    It also doesn’t take a lot of work to figure out that the Saudi Monetary agency’s foreign assets should be considered a form of reserves (try reading the IMF’s regional outlook; Jen worked for the IMF, he knows about this resource). And if you are gonna claim that China’s dollar holdings are smaller than the sum of the last stock data point in the latest Survey and the sum of subsequent flows, well, to my mind that does require something of an explanation!

    If he wants to outsource his reserves numbers work to RGE, I am all ears ..

  • Posted by HZ

    Have you considered the Chinese Fx holding on a net basis? If I were a Chinese airline, I would have massively borrowed in dollar to buy airplanes and then earn RMB revenues — RMB appreciation makes the deal that much sweeter. There must have been many such deals in anticipation of RMB appreciation. So yes even though RMB appreciation is a capital loss for the PBoC, part of the loss is a domestic transfer and net loss on a national basis may be more proportional to the cumulative (bona fide) trade surplus than to the C/A surplus.

  • Posted by Emmanuel

    I sometimes have an opinion or two to express, but when it comes to the composition of Chinese reserves, I shut up and listen to RGE. Jen is also proving to be somewhat off target with the Euro bouncing up at year-end.

    I would find it unconscionable if they are forcing state-owned banks to load up on Treasuries. On top of bad loans made to state-owned firms, they are made to purchase low-yielding sterilization bonds, and now Treasuries? And people wonder why the Chinese banking system is rickety.

  • Posted by Emmanuel

    And another thing: the FT article correctly points out that net foreign purchases of US Treasuries was negative in September 2006 (-.374B) for the first time since February 2003 (-3.210B). That ends 42 straight months of positive figures. That this interesting factoid escaped much attention is somewhat strange.

    Is this the start of a broader trend of portfolio diversification? We’ll have to see what happens in the coming months. While $.374B isn’t chump change, in January 2002 (-15.278B) and July 2001 (-11.494B) there were far larger negative figures.

    Let the fire sale begin!?

  • Posted by Anonymous

    Everybody knows about this. Amongst City scribblers the most widely shared resource is ignorance.

    You’re light years ahead.Don’t waiste your time.

  • Posted by bsetser

    HZ — I haven’t looked at domestic dollar borrowing recently, but most of it is financed by domestic $ deposits or offshore lines from the international banks — not from the the PBoC. The possible exception is dollars borrowed from the PBoC thru swap lines. The total amount is also small relative to PBoC reserves.

    I don’t think Jen is ignorant — he can be dead on at times. But in this case, I do think he ranged into an area (Chinese dollar holdings) without looking fully at the available data. We all do that at times.

  • Posted by Anon128

    Is it possible that China is limiting ‘official’ reserve growths by encouraging the corporate sector to keep their USD?

    If so, how does this, or would this, change the BWII dynamic?

    Thank you for the wonderful blog.

  • Posted by bsetser

    anon128 — I think it is quite likely. indeed, corporate dollar accumulation it is the only way I can make China’s balance of payments work for 2006. I don’t see many signs that RMB deposit growth has slowed/ Chinese citizens are less willing to hold their savings in RMB — which implies someone else is accounting for the “private” capital outflows in the balance of payments!

    the only real question in my mind is who in the corporate sector is keeping their USD. The banks clearly have kept their IPO funds out of China, whether in HK$ or US$. But someone else needs to be adding to their dollar balances as well.

  • Posted by The Tortoise

    I wonder if it ever occurs to financial gurus that The Peoples Republic of China could do more with that much paper than store or spend it. It is quite enough to destabilise the US economy, particularly since destabilisation, as a process, has the nuclear quality of becoming self-propagating once it reaches a critical point. But, of course, that would never be in the interests of The Peoples Republic of China. How many CEO’s have gone down convinced it wouldn’t be in a bank’s interest to pull the rug? In any event it shouldn’t be necessary, it could function better as a deterrent. A portion might, for instance, be used to give a rap over the knuckles and a clear warning in the matter of some future sensitive international issue. You hardly imagine the The Peoples Republic of China is going to stand by indefinitely letting the US repeat what it has done in the Middle East.