Just what is in China’s portfolio? And do foreign central banks think Agencies are Treasuries?
China's decision to set up a new investment authority with a mandate to invest in a range of assets — not just traditional reserve assets — has attracted a lot of attention. But as the Bank of New York notes, China hasn't yet revealed a lot of the key details about the new authority, including how it will raise the money to finance the purchase of its foreign assets and how quickly it will build-up its assets.
The announcement of the new investment authority though has had one useful byproduct: it has produced a flurry of information about the content of China's existing portfolio.
Keith Bradsher of the New York Times – reporting out of China – notes that China has roughly $100b in U.S. mortgage backed securities. That seems right to me.
People close to the State Administration of Foreign Exchange, which is controlled by the People's Bank of China and manages the country's reserves, estimated that the agency already held about $100 billion worth of American mortgage- backed securities. That may seem to be an unusual place to invest foreign-exchange reserves, but it was selected in the hope of achieving better yields than those on U.S. Treasury securities. None of these mortgage-backed securities is said to be tainted by the subprime loan portfolios that have recently fallen sharply in value.
Bradsher also notes that China holds $600b in Treasuries.
Experts estimate that it holds another $600 billion or so worth of U.S. Treasury securities that it lends actively to generate extra profit, as well as at least $200 billion worth of euro-denominated bonds. The remainder is thought to be held in bonds denominated in yen and other currencies.
That seems off to me – at least the $600b Treasury estimate. The US Treasury’s own data puts China’s actual holdings of Treasuries at around $350b.
The TIC data — which is the source for the most recent Treasury estimate — tends to understate China’s actual holdings. The survey data is better — but it is now dated.
June 2005 is roughly $360b of Chinese reserves ago (more like $330b of reserves ago if you adjust for valuation gains). When the next survey comes out, China's current holdings may be revised up. The last survey led the US to increase its estimate of Chinese Treasury holdings by over $50b (see the series break in June 2005). But it won’t be revised from $350b to $600b.
So where did the $600b number come from? Here is one theory: a lot of central banks seem to consider high rated Agencies to be a very close subsistute for “Treasuries.” Or they at least don’t always differentiate between the two when talking to the press and public.
One example. Russia’s investment guidelines for its oil stabilization fund allow for investment in a wide range of highly rated fixed income securities (with relatively short maturities). It is often reported that Russia correspondingly holds a lot of Treasuries. However, the US data suggests Russia overwhelmingly holds short-term Agencies. Agencies accounted for $61.956 of $61.966b in recorded Russian short-term holdings in the June 2005 survey (see p..66). Ergo — be careful. Sometimes the distinction between Agencies and Treasuries seems to get a bit blurred.
The last (June 2005) survey put Chinese holdings of treasuries at around $300b — $277b of long-term treasuries and $21b of short-term Treasuries. At the time, Chinese holdings of Agencies totaled around $192b. Since then the US data shows Chinese holdings of Treasuries have risen by $50b, to around $350b. Purchases of long-term debt have actually been a bit higher, but China has reduced its short-term holdings a bit. The flow data indicates another $57b (roughly) in agency purchases – enough to bring the total up to $250b.
$350b of Treasuries and $250b of agencies is, well, $600b. Exactly the number Bradsher reported.
That could be a coincidence. But it might not be. In the past, the PBoC sometimes seems to have drawn on the US data to describe the composition of its portfolio. That way, it avoids giving out any state secrets.
$600b in Treasuries and Agencies plus $100b in private mortgage backed securities puts SAFE's dollar portfolio in the $700 range – which would be a bit lower than I would estimate. Remember, China’s central bank holds $1070b or so. $700b works out to a bit above 65% of China's portfolio.
China's actual dollar holders would be a bit higher — as the $700b total doesn’t include China’s onshore or offshore dollar bank deposits, its holdings of bonds issued by the World Bank and its holdings of dollar-denominated emerging market debt. Even so, I suspect $700b is on the low side for China's actual claims on the US.
Richard McGregor of the FT — who usually is well informed — reports that around 75% of China's reserves are in dollars. Think $800b. That seems about right to me.
Simon Derrick expects the Europeans will tell the Chinese not to diversify, since they don't want an even higher euro. I rather suspect the Europeans quietly have been delivering that message to China for some time now.
What explains the gap between the $700b in Chinese holdings that emerge from the TIC data and the $800b estimate that emerges from a more top down estimate?
Simple: The TIC data probably understates China's purchases of US debt since June 2005.
Since the last survey, the TIC data indicates China has bought around $165b in treasuries, agencies and mortgage backed securities (labeled corporate debt in the US data). Its short-term claims also fell by $15b over that period – for an overall increase in recorded US holdings of US debt of $150b. China’s reserves increased from $710b to $1070b over that period – and increase of $360b ($330b after adjusting for valuation changes). My personal view is that over $150b of that $330b increase went into US debt.
There is another wrinkle: The entral bank isn’t the only source of Chinese dollar holdings. Central huijin (the holding pen for the PBoC’s stake in Chinese banks) holds another $60b in total assets, presumably mostly in dollars. The state banks have been forced to hold the funds they have raised offshore offshore – and they presumably are primarily in US or Hong Kong dollars. And through various swaps the PBoC placed $14b with private banks in 2005 and probably another $30-$40b in 2006 …
Central Huijin got its dollars before June 2005, so its assets should have been counted in the survey – my calculations compare China's roughly $530b in reported assets in the survey to $710b in reserves and $60b with Central Huijin, for a total of $770b. But the dollars from the bank IPOs and various swaps generally came after June 2005, so they in some sense shoudl be added to the increase in reserves.
If all of China's additional “reserve like” foreign assets are added to the SAFE's holdings, official China has north of $1200b in foreign assets — or an increase of around $400b since June 2005.
If 75% of those are in dollars, that works out to around $900b.
Ergo — I would say that $700b is the lower bound of China's "official" dollar holdings, and $900b is probably close to the upper bound.
All this is just an attempt to square Bradsher’s reporting with my own numbers – both to test Bradsher and to test my own understanding. Bradsher clearly got a bit more access to the SAFE than most.
But I think there is a broader point here too – one that Richard Duncan has made in the past.
Fannie and Freddie do a lot of things. But one of the things that they do particularly well is transform US mortgage lending into assets that central banks can comfortably call reserves. Their credit guarantees make long-term US mortgages more attractive. An agency guarantee gets rid of default risk – and reduces the central bank’s need to worry about all sorts of pesky details about the US mortgage market. Interest rate risk remains … but that is something central banks think they understand.
And for those central banks that don't want interest rate risk (and pre-payment risk) , the Agencies also finance a portion of the portfolio with short-term debt. Here the agencies also intermediate between foreign central banks demand for short-term reserve assets and the US household sector’s desire for long-term financing …
I am not sure the US really has a comparable advantage in financial intermediation. London – and for that matter several other European financial centers – are doing rather well at borrowing funds from the world at low rates to invest at higher rates. The US by contrast is struggling to attract as much as it needs to cover its current account deficit. But I am pretty confident the US has a comparative advantage at turning home loans into securities that fit nicely into a traditional central bank portfolio.
The latest flow of funds data shows that central bank purchases of Agencies have matched their purchases of Treasuries in both 2005 and 2006.
Then again, lots of central banks now seem interested in holding a rather less traditional portfolio. In 2006, that meant more agencies and for the adventurous, more “private” mortgage backed securities. In 2007, it likely means dipping their toes in the equity markets …
A last administrative point: as I noted at the end of last week, RGE now requires that all those who want to comment register. I do not want this to hinder the discussion at this blog, which I value very highly.
Those registering have two options — one is to just register to participate in the comments section of the blog (there is a box in the registration process that says "blog comments only"), the other is to register for a trial subscription to RGE and to sign up for RGE's biweekly email as well as to register for the blog. The blog registration will remain in effect after the trial subscription expires. Registered users can still post comments anonymously on the blog. RGE also generally doesn't accept yahoo or gmail emails for registration purposes. If you have questions/ difficulties, please email me at brad dot setser at rgemonitor dot com.
On the lesser point of reserves invested in agency bonds, I’d differ with the comment by Bradsher “None of these mortgage-backed securities is said to be tainted by the subprime loan portfolios that have recently fallen sharply in value.”
That’s naive. Lenders who were increasingly lax at the lower end of the mortgage scale have been increasing lax at all levels of the credit spectrum, including commercial lending. Different departments -residential, high end residential, commerical condo, commercial warehouse, industrial - all compete for rank within each lending institution. To think Fannie Mae and Ginny Mae sorted through all the packaged MBS with a microscope and noticed this is unrealistic. Aggressive in one area, agressive in all. Department heads compete with each other for growth in volume of loans made. It’s human behavior in the real world, IMHO.
Don’t agencies come with a credit guarantee tho, so the risk is with fannie, not the PBoC? The PBoC/ SAFE only need to worry about credit risk on the portion of their portfolio that is in “private” MBS, n’est pas?
The rationale of a credit guarantee seem to me to be somewhat of a moot point. As accruing evidence appears to indicate, spillover from the subprime to higher quality issues, and the credit markets in general, appears to be emerging. If this continues, and I see no reason why it will not, even if default risk is “guaranteed against”, there will be growing valuation risks if the subprime financial contagion grows.
Ryan — i see your point for private MBS ($100b plus of the PBoC’s portfolio), though the PBoc/ SAFE presumably hold the best not the worst credits. But i don’t see it on Agencies. Fannie should worry — not the PBoC (plus i think the agencies require documentation and a lot of other things, so they were scooping up the worst credits to back their securities). With the agencies, there is an issue of pre-payment if interest rates fall on the back of a broader slump, but unless my understanding of agency securities is way off, there just isn’t much credit risk for the end buyer.
Brad,
With all due respect and appreciation for your work, I’m not so sure you do see my point. Although, it certainly could be me that does not see the point.
The frame of reference suggested by your analysis does not seem to take into consideration the systemwide implications of what, as your colleague Nouriel recently described it, may be a financial “train wreck”.
When considered in isolation, the attributes associated with an agency security would lead to an analysis such as your offering. However, it appears to me, and I believe it is not likely to be a point missed by the Chinese finance ministers, that the spread of a subprime collapse is likely to have implications that will permeate every sector of the global economic system.
If we consider that a guarantee is only as good as the guarantor in the context of federal government with staggering, and growing, unfunded liabilities, the possibility of a subprime collapse also carries with it the possibility of some type of governmental intervention to remediate. At the very least this creates an added dimension of uncertainty, and consequently, increased risk, increased risk premia, and a potential devaluation of current security holders.
A prudent financial response the a holder of concentrated positions in these securities would be to diversify away, preferably into asset which hold strategic operating values, even if the financial dimension were to suffer. This appears to me to be the Chinese response, and a very intelligent one at that from what I can see.
Ryan — NR is a bit more bearish than I am. But i don’t really think your broader point (big unfunded mandates) is relevant here — as far as i know, the agencies tend to be in the the documented, fixed rate bit of the mortgage market not the more exotic bits, and supposedly haven’t been underpricing their credit guarantees. in a sense, the fed’s efforts to reign in the agencies may have limited their exposure — as it helped pave the way for the boom in ‘private” mbs. I don’t think China has to worry about credit risk on its agency/ treasury position.
If i were China i would be much more worried about another kind of concentration risk — namely concentration in $ assets. While the unfunded mandate problem is still a ways off (soc. sec is cash flow positive and will remain so for 10 years and so on), the US trade deficit is big and immediate — and it implies further $ depreciation over time. that concetration is what should worry chinese policy makers.
their problem? they still basically peg to the $, and don’t want to change quickly — which forces them to keep on buying $.
Sorry I wasn’t more clear - you’re right on what PBoC should worry about. What else it should worry about is that MBS generally including those guaranteed by Fanny Mae are shakier than believed, and that value of Agencies in the market will be affected by that shakiness. Also the US banking system has exposure to MBS used in part as Tier One Capital. China’s guarantees (in dollars) may not be worth as much as they thought, like the MBS and CDO’s. Main worry here is squarely in the US.
For someone who’s uninitiated, can you explain what is the difference between Agencies and Treasuries? Is what we call Agengices mainly Fannie Mae and Freddie Mac?
Also, is there any reliable data on what is the total amount Fannie and Freddie are likely to lose in the expanding subprime crisis? Are there any reliable data for Wall Street losses?
Finally, what makes TIC data on non-Chinese purchases any more reliable than chinese purchases?
Agencies are Fannie and Freddie — basically, the government sponsored agencies that are active in mortgages (but of course are stand alone entitities that the US government would never back in pinch). Treasuries are the unsecured marketable debt of the US government.
No clue on Fannie/ Freddie exposure. As indicated above, i think it should lower than the exposure of the street, because a) Fannie was forced to curb the expansion of its balance sheet by various treasury iniatives and b) I don’t think either were big in the exotic parts of the mortgage market but rather did boring old fixed rate with documentation of your income mortgages and the like.
The TIC data tends to get total purchases right — it just tends not to capture the true buyer, so the geographic distribution is off as is the official private split. Basically, the tIc suffers from custodial bias, so it overstates purchases from the private sector (as some central banks use private custodians) and overstates purchases from financial centers (London most famously). The TIC also will count purchases by say a london fund manager with a mandate to invest for a foreign central bank as “private” when it really is official assets under private management.
the one thing the TIC doesn’t capture at all is the sale of US debt already abroad to another foreigner — i.e. a german pension fund sells to a Chinese state bank. But it should capture the volume of net foreign purchases accurately, as it is based on the reporting of transactions, just not the ultimate purchaser.
Are you sure that this is not a case of journalists thinking that agencies are MBS, Brad? It would be surprising if the Chinese really did hold $100bn of MBS directly, as opposed to holding them indirectly via bonds issued by the Federal mortgage agencies.
I am no expert in MBS, but my understanding is that a buyer of them gets delivered a bundle of many small pieces. This and the complexity of the pre-payment risk makes them difficult for central banks to handle. The only central bank I know to buy MBS directly reckoned that the benefits were roughly matched by the costs, and was reviewing whether to continue.
Most central banks do buy the bullet bonds issued by the agencies though. This leaves the agencies to deal with the MBS, but still gives the central bank a pick up. Since the Fannie Mae benchmarks and Freddie Mac reference notes are highly liquid, this pick up supposedly reflects the credit risk of the agencies themselves. While they cannot officially admit it, most central banks are happy to take this risk, because they believe that the US authorities would bail the agencies out if they got into difficulty.
Rebel — i think Bradsher has it right. Chinese holdings of agencies are far larger than $100b — and recently the TIC data has shown large Chinese purchases of US corporate debt, which generally is thought to be “private” MBS, along with large purchases of agencies. His $100b number matches the number I get from following the TIC flows + the survey data. SAFE is from what i have heard the central bank that has been most active in the MBS market.