Just how large is China’s subprime exposure?

Josef Joffe:

[Can Moscow and Beijing] mend a global financial system battered by subprime crisis?   Where are the central banks of Russia and China? 

Hmmm.    The fact that China and the oil exporters are running a large current account surplus does – in a lot of ways – help to explain why so much credit was available for so long for so many Americans looking to buy homes.   At the same time, I am not sure that Russia and China can really do much to solve the subprime crisis.  It, after all, stems from a combination of over-indebted US households that no longer can make mortgages payments and a crisis of confidence in US financial engineering.

But credit should still be given where credit is due.  Russia’s central bank has helped, in its own small way.  Its willingness to sell dollars and euros over the past two weeks helped foreign investors take some risk off the table, and in the process, helped supply liquidity to the market.   

China is also helping, though in a different way.   Its banks clearly will be absorbing some of the losses from subprime.   SAFE likely will absorb some losses too – though we don’t yet know how much.     

I have spent a bit of time recently trying to guess China’s subprime exposure.    The three big Chinese state commercial banks have disclosed about $12b (11.94b to be exact) in subprime exposure (BoC has the most exposure, but also the most foreign debt).    

These three banks collectively held – at the end of 2006 -- $155b in dollar denominated securities and $199b in foreign securities.  That works out to almost all of the $228b in “private” Chinese holdings of debt securities that show up in China’s net international investment position data.   

In addition to holding a fair amount of subprime exposure, they also likely have a lot of exposure to “prime” US mortgages, and probably a range of other US assets as well. 

But the $200 in debt held by the three large state commercial banks at the end of 2006 pales relative to the $1,066 billion that SAFE had at the end of 2006 -- let alone its current stash of $1,332 billion.

And it is very, very hard to figure out how much of that is in US mortgages.   The US survey data now breaks out holdings of mortgage backed securities from holdings of other asset backed securities.     But the US data is over a year old, and it clearly doesn’t capture all of China’s current holdings. 

How do we know this?   Simple:  the three Chinese banks have disclosed more subprime holdings ($12b) than China’s recorded holdings of all MBS in June 2006 ($9.45b).   The US data can be found at the very end -- p. 117/ Table 25 -- of the most recent US Treasury survey.  

It isn’t clear – at least to me – if China’s banks hold their US securities in China or Hong Kong.  But the combined total of the MBS held by China and HK at the end of 2006 -- $14.6b -- is only a bit larger than the exposure the Chinese banks have disclosed to US subprime debt.   Adding the data from Hong Kong in doesn’t solve the puzzle.   

Moreover, the limited scale of China’s holdings of MBS reported in the US data is at odds with the reporting of Keith Bradsher of the New York Times.  Bradsher, citing “people with knowledge of the central bank's trading,” has reported that China holds about $100b of private MBS.  It is also at odds with a fair amount of anecdotal data.  

The FT’s Alphaville reported about a letter from Kyle Bass of Hayman capital last week. Bass claimed::

I recently spent some time with a senior executive in the structured product marketing group (Collateralized Debt Obligations, Collateralized Loan Obligations, Etc.) of one of the largest brokerage firms in the world. …  This individual proceeded to tell me how and why the Subprime Mezzanine CDO business existed. Subprime Mezzanine CDOs are 10-20X levered vehicles that contain only the BBB and BBB- tranches of Subprime debt. He told me that the “real money” (US insurance companies, pension funds, etc) accounts had stopped purchasing mezzanine tranches of US Subprime debt in late 2003 and that they needed a mechanism that could enable them to “mark up” these loans, package them opaquely, and EXPORT THE NEWLY PACKAGED RISK TO UNWITTING BUYERS IN ASIA AND CENTRAL EUROPE!!!! ….  Interestingly enough, these buyers (mainland Chinese Banks, the Chinese Government, Taiwanese banks, Korean banks, German banks, French banks, UK banks) possess the “excess” pools of liquidity around the globe. These pools are basically derived from two sources: 1) massive trade surpluses with the US in USD, 2) petrodollar recyclers.

These two pools of excess capital are US dollar denominated and have had a virtually insatiable demand for US dollar denominated debt…until now. They have had orders on the various desks of Wall St. to buy any US debt rated “AAA” by the rating agencies in the US. How do BBB and BBB-tranches become AAA? Through the alchemy of Mezzanine-CDOs. With the help of the ratings agencies the Mezzanine CDO managers collect a series of BBB and BBB- tranches and repackage them with a cascading cash waterfall so that the top tiers are paid out first on all the tranches – thus allowing them to be rated AAA.  … This will go down as one of the biggest financial illusions the world has EVER seen. These institutions have these investments marked at PAR or 100 cents on the dollar for the most part. Now that the underlying collateral has begun to be downgraded, it is only a matter of time (weeks, days, or maybe just hours) before the ratings agencies (or what is left of them) downgrade the actual tranches of these various CDO structures.

That is a fairly dramatic claim – and one at odds with the relatively limited exposure of Chines to US mortgages reported in the US data.  So what is going on?

First, and most obviously, the US survey data is now a year out of date.   It misses $400b in Chinese reserve growth (roughly $370b adjusting for valuation effects) and an additional $63.6b in private Chinese debt purchases in the second half of 2006 that shows up in the Chinese balance of payments data (some of those private purchases may have been reversed in the first half of 2007, we don’t yet know).  The acceleration in Chinese reserve growth over the past year coincided with an increase in China’s risk appetite.   The purchase of debt securities by China’s banks also picked up in the course of 2006 – and a lot of these purchases were financed using funds borrowed from the PBoC through fx swaps -- 

As a result, China -- both the banks and SAFE -- likely added substantially to their holdings of US mortgage debt over the past year.   That doesn’t bode well for China – at least not if it bought newly issued MBS and above all new mortgage-backed CDOs.   The 2006 “vintage” of subprime CDOs hasn’t received rave reviews from the critics.   A lot of those mortgages were taken out by people who umm perhaps stretched the truth in order to qualify for a loan to buy a home that they expected to rise in value, aided by mortgage-brokers looking for another commission … 

Second, after looking at the data on the reported holders of US MBS in the middle of 2006, I would guess that the US data isn’t picking up the ultimate owners of a lot of MBS.   Two give away: the single biggest holder of MBS was the Cayman Islands ($72.4b of $340.9b total) and the Caymans, Bermuda and Jersey combine to account for about 30% of total foreign holdings.   

My guess is that a lot of CDOs are legally set up a Cayman island funds.  Their purchases of US debt show up as purchases from the Caymans.  But the CDO then sells tranches – effectively its own bonds – to investors globally.    Those sales wouldn’t show up in the US data.    Even the survey is only picking up one leg of the overall trade.

Another $120b of MBS were held in the UK, the Netherlands, Luxembourg, Belgium and Ireland.   Some of those MBS were held by conduits set up by European firms – and US banks who prefer to show profits (and perhaps now losses) in Europe for tax reasons.   But perhaps some CDOs are also managed out of these locations – all of which offer relatively favorable (I think) tax treatment for offshore investors.

Certainly China’s $9.5b in recorded holdings of MBS – and Japan’s $19.8b in recorded holdings – seems a bit low.   China and Japan have large current account surpluses to invest in global markets.   The Caymans – and for that matter Europe – doesn’t.

In the past, I have used China’s holdings of all “corporate” debt as a proxy for its MBS holdings – largely because the rise in Chinese holdings matched the anecdotal reports that China had emerged as an important buyer of “private MBS.”  After looking at the data more closely, I clearly got this wrong – the majority of China’s recorded holdings of corporate debt are not MBS.  See the Table 25 at the back of the 2006 survey; too bad the monthly TIC data doesn’t provide a bit more of a breakdown between different categories of corporate debt.

But I am also relatively confident that the US data understates China’s holdings.   There is just too much evidence that China was a force in the MBS market over the past year to believe the US data.

Bradsher’s estimates make sense to me.   Significant purchases of MBS over the past year that don’t fully register in the US data would also help fill in the huge gap between recorded Chinese purchases and any plausible estimate of the increase in its dollar reserves.    Some of London purchases and Hong Kong purchases are also likely making their way to China.   No one else has bought $450b of debt over the past year …

But even if SAFE has $100b of “private” MBS and say $20b in subprime exposure that doesn’t represent that large a share of its portfolio.   It probably holds close to $900b in Treasuries and Agencies and another $300b in relatively safe euro, pound, Canadian dollar, and won-denominated debt.   And on a mark to market basis, the rise in the value of China’s Treasuries and Agencies likely offset much of the fall in the value of its “subprime” CDOs.    Say the average duration of China’s holdings is around 3 years.   And say US interest rates fell by 20 bp because of the subprime crisis.   That implies a 0.6% increase in the market value of China’s holdings from interest rate changes – or a $6b rise.

China’s true exposure to the subprime crisis doesn’t come – in my view – from its direct holdings of CDOS that hold subprime CDOs. 

No, China’s exposure comes from its concentrated bet on the dollar – and the risk that the US policy response to a slowing economy and investors aversion to US debt will combine to put pressure on the dollar going forward.    In my view, analysts who rely on the TIC data to infer China’s holdings end up understating the dollar share of China’s portfolio rather significantly.   China now holds a lot more than $635b of US debt.   Just look at the survey data – is shows close to $700b in Chinese holdings over a year ago (p.50/ chart 19).   The survey revised China holdings up by $90b in mid-2005 and in mid-2006 and likely will do something similar once the mid-2007 comes out.

Further dollar weakness consequently would be a problem for China.  It implies more pressure for RMB appreciation – and growing financial risks for the central banks and the investment agency, which have financed their purchases of both dollar and euros by borrowing in RMB.    SAFE has a lot more exchange rate risk on its balance sheet than credit risk.

None of this though is to deny the sting China may feel from losses on its holdings of CDOs.   It likely bought at the wrong time, just as the US housing market turned south.   And while subprime CDOs may not amount to that large a share of China’s overall portfolio, they could account for a decent share of recent flows.

And that poses a key question:   will SAFE realize the mark to market gains on its treasuries by selling some and using the proceeds to buy riskier assets – perhaps at attractive prices?   Poland’s blogging deputy central bank governor – Dr. Rybinski --  thinks China and others central banks with a lot of cash should do so.   That would help add liquidity to segments of the market that now lack liquidity.   And the central banks might be able to buy some assets at a nice discount.

Or will China retreat from risk after buying a generation of CDOs that offered more risk than return?

My guess is that there are a lot more investors in the US and Europe now hoping to sell illiquid CDOs – including additional CDOs stuffed with subprime debt -- to China, Russia and the wealth funds of the oil exporters than there is demand for such exposure …  But that is just a guess.

UPDATE: SAFE now reports that it has zero subprime exposure -- apparently, it only bought "prime" MBS.