Just how big is the yen carry trade?
Stephen Jen of Morgan Stanley insists not-so-big. Or at least that the large size of the carry trade cannot be proved – and that the impact of any recent unwinding has been overstated. That is his long-standing view. In early February, Jen wrote: “So far, our view is that most of the outflows from Japan are benign and stable in nature.”
John Dizard of the FT seems to agree. He notes that the Japanese authorities estimate that leveraged short-term bets on the yen only total $20-40b.
Others, though, seem to think the yen carry trade is now rather large.
Tim Lee of Pi Economics recently got some support from Jesper Koll of Merrill Lynch. He also puts the size of the carry trade at close to $1 trillion – and more importantly, says that various players in the market added about $300b to their position over the past year. The FT:
Merrill Lynch said on Tuesday that its research in the Tokyo interbank market showed that non-Japanese banks’ funding had surged from Y1,500bn to Y7,500bn since the end of 2005.
“This Y6tn ($51bn) surge in non-Japanese funding is a new development that, in our view, points to the true carry of yen-funded leveraged position build-up in the recent past,” said Jesper Koll, economist at Merrill Lynch, who believes “the absolute size [of this carry trade is] maybe around $1tn, with about a third of the positions built up in the past 12 months”
No surprise, many think that even after the market’s recent carry-unwind related wobbles, the are still a lot of leveraged short-yen/ long carry positions out there. Michael Woolfolk of the Bank of New York for one. Peter Garnham quotes Woolfolk in the FT:
“I suspect that what little carry trade unwinding was seen in the past two weeks was only the tip of the iceberg compared to the remaining trades that are still on the books”
Jim O’Neill of Goldman worries a bit more than Dr. Jen. He also puts forward a slightly lower – but still substantial – estimate of the size of the yen carry trade, roughly 5% of Japan’s GDP, or a bit under $250b:
Jim O'Neill, the bank's chief global economist, said investment firms playing the "carry trade" had been caught on the wrong side of huge leveraged bets against the Japanese yen. "There has been an amazing amount of leverage on currency markets that has nothing to do with real economic activity. I think there are going to be dead bodies around when this is over," he said. "The yen carry trade has reached 5pc of Japan's GDP. This is enormous and highly risky, as we are now seeing."
One approach to estimating the size of the carry trade – and the approach that I think Goldman uses — takes Japan’s basic balance of payments surplus (the sum of the current account, net FDI flows and net portfolio flows) as a good proxy for the size of the carry trade. Jamie McGeever of Reuters reports:
There's also an argument to be made that, in effect, Japan's broad basic balance surplus is a fair measure of the overall short yen position. That comprises the surplus on Japan's current account, foreign direct investment and portfolio flows, which is roughly 4.5 percent of the country's entire gross domestic product — or around $215 billion at current exchange rates.
Here is the logic of looking at the basic balance of payments, as I understand it. Japan has a substantial current account surplus, which implies a substantial, ongoing capital outflow from Japan. Yet, the balance of payments data doesn’t show large portfolio outflows. Indeed, if I am reading the Japanese BoP data right, Japan experienced a large net portfolio inflow in 2006, and the net portfolio outflow in 2005 was rather small. Japanese retail accounts seem to have a strong appetite for high-yielding foreign currency debt, but Japanese institutional investors seem a bit more cautious – and foreigners have been buyers of Japanese equities. The absence of net portfolio outflows implies that other – harder to track – flows have been the main vector bring Japan’s surplus to foreign markets. The BIS recently found that hedge fund returns correlate with the returns on certain simply long carry strategies.
One part of Merrill’s argument particularly interests me. If leveraged “long carry” bets are the way that Japan’s current account surplus is lent out to countries that need financing – or, for that matter, lent out to countries that don’t need financing, fueling their reserve growth and thus indirectly financing the US — thesize of the leveraged long “carry” positions has to keep on growing to sustain the global status quo. It is kind of like the situation with China’s reserves and quasi-reserves. China doesn’t just have to hold on to its existing dollars – it needs to keep on buying. Same with carry traders. If they weren’t adding to their positions, either Japanese institutions would need to but more foreign bonds and equities, foreigners would need to lose interest in Japanese equities (increasing the net outflow) or the Japanese government would once again need to borrow yen to buy dollars and euros (sort of like 2003 and early 2004).
The Merrill/ Goldman argument also jibes with the huge scale of recent inflows to places like Brazil and India – which is one reason why I find it relatively persuasive. But there is a risk that these flows are influencing me too much: the scale of “carry” positions in the Brazilian real ($12b according to a recent estimate) or in Turkish lira (foreign holdings of Turkish debt total 43billion lira, or around $30b) are small in the global scheme of things. That is why I am interested in Macroman’s contrary view. The folks he talks to seem to have different sense than the Goldman/ Merrill research shops.
And if anyone at Merrill reads this blog and wants to send me the report – my email address is brad dot setser at rgemonitor dot com
Brad,
A brief comment. According to Japan GDP data, net income from the rest of the world was 14.6 trillion yen in 2006 compared with 8.3 trillion yen in 2001. If all of this increase in net income from the rest of the world over the five year period is due to the domestic carry trade and if a this net income represents a 7% return on investment, then the total increase in the carry since 2001 would be about 90 trillion yen or about 750 billion US. This back of the envelope analysis is simplistic but does suggest that a reasonable about of the carry comes from domestic investors.
John Tofflemire
interesting point — hadn’t thought of looking at it from that point of view, but it makes sense. if leveraged investors from outside japan get most of the carry, they should also get most of the income.
three comments:
a) a lot of the income goes to the MoF … all those reserves it bought from 02-early 04. tis a carry trade of sorts … and the MoF is a domestic investor of sorts.
b) the average int. rate on US debt (I’ll be posting on this soon) is under 4.5% — v north of 6% back in say 00 — which suggests to me that the upward adjustment in japan’s income balance could rise further.
c) my personal definition of the “leveraged” carry trade focuses on the use of leverage more than on nationality — so japanese day traders who gear up/ japanese hedge funds would count, but not japanese real money.
obviously, though, there are different ways of thinking about this, and leveraged japanese investors may act very differently that leveraged foreign investors.
“Yen carry” traders have borrowed an estimated 40 to 70 trillion yen ($350-600 billion) and channeled much of the funds into commodities and stocks around the globe. In addition, Tokyo’s MoF has acquired $875 billion of foreign currencies through its intervention operations. About 65% of Japan’s FX reserves are in US dollars and 35% in Euros, so MoF can enforce a floor under the yen at any point of its choosing, and keep a lid on the “yen carry” trade through massive currency intervention.
DC — I don’t know mortgages, but i do know reserves … and well over 65% of Japan’s reserves are in $. Most estimates (Truman and Wang = the best) put Japan’s dollar share north of 80%. The 65/35 split is i think for the Bank of Japan’s “own” reserves. However, the MoF controls the bulk of Japan reserves, tho they are managed by the BoJ. And the $ share of the MoF’s reserves is generally put a lot higher.
a question: what is the source on the $350-600b estimate for the stock of the carry trade? I am trying to compile a good list of different estimates.
Hi Brad,
I pulled the “Yen Carry” trade statistics from a Google search and found the article by Gary Dorsch.
http://www.sirchartsalot.com/article.php?id=55
Whether his “yen carry” trade statistics are accurate or not, I certainly don’t know. Since the activities of Hedge Funds involved with the carry trade are opaque, I don’t think even the Japanese MoF knows the exact true scale of the massive currency transactions.
P.S. I registered for just posting on the RGE website but my account expires in 12 days. Can you see if you can make my account permanent? Thanks.
Regards,
It’s almost bedtime, but I have a few observations: the hidden vector in 2005-2006 can be very simply explained in two words: hedge ratios. In 2003-2004, Japanese lifers were paid to hedge 100% of the currency risk of their foreign bond holdings. From 2005 onwards, that has not been the case, so they have reduced their hedge ratios: a typically over the counter trade that is not captured in the portfolio data.
As Brad notes, the BOJ reserve ratios are utterly irrelevant. The MOF holds the vast, vast majority of its reserves in USD, and that is not going to change any time soon.
I have also heard/read a substantial amount of talk about private equity being funded in yen. I have done a fair amount of research on this matter, and the conclusion is: it ain’t happening. P/E deals are generally funded in local currency debt to match liability with asset, and there was absolutely no suggestion from the people I spoke to (who are in an excellent position to know) that PE funds are swapping their liabilities into yen. Quite the contrary; the pushback I received was that PE firms are aggressively agnostic about currencies and as such do not try to take directional bets: hence the Asset-Liability matchup.
Yen carry trades not defined clearly makes this debate confused
Gheorghius — my definition would be leveraged bets on yen weakness or stability against currencies with higher interest rates. that excludes real money portfolio flows. I also focus on fixed income flows, but I can see a case for including someone who say borrows yen to bet on brazilian equities rather than Brazilian local currency debt.
but you are certainly right — and this is something that I agree with Dr. Jen on — the term yen carry trade includes a vast number of different things. Sometimes it includes real money flows leaving Japan b/c of low Japanese rates. But my focus would be on those borrowing yen at low rates … not those who already have yen and are leaving japan to get higher rates, if that makes sense. I am open to other definitions.
Macroman — my sense from reading various press reports is that the japanese lifers haven’t been adding to their portfolios that quickly. tis true that changes in the way they hedge their existing stock of $ claims might generate some interbank flows (I would need to think about it), but it the potential scale here really big enough to accout for the needed $200-250b outflow in the BoP data (taking into account the observed portfolio inflows)?
DC — don’t worry. here is what is going to happen — your account will expire in 12 days. your broad access to the RGE site will disappear. but expired accounts still work for blog registration/ blog commenting (see NR’s post), so you can still comment here.
this is a byproduct of the way the blog registration process was set up/ piggybacked off the site’s existing registration process. however, i share your view that it is rather confusing, since a lot of people likely will have the same concern you expressed — it seems like the expiration of your account should limit your acces to blog. but it is a seperate clock, one for the rest of the site. I don’t think I am going to be able to convince the powers that be to change the current approach though.
bottom line: don’t worry, everything should still work on the blog even after your trial subscription expires. if it doesn’t email me.
Brad, I think there are two issues conflated here.
#1: how much is invested from Japan externally in the hope and expectation that the yen will weaken, thereby permitting one to tolerate a lower rate of return,
vs.
#2: how much of any investment, internal or external, is on margin.
I think that currency risk is the major risk, even greater than that of recession. So, I keep some cash in Euros. These earn about 3% annually. I could keep it in dollars and earn 5%. But the currency shifts have added several percent to the value of the investment. This is, by the general definition, carry trading. But there is zero risk to the general public because none of the money is borrowed. If the Euro suddenly drops, I lose money.
Japanese are fearful of being wiped out by local disaster. One has to live there to understand just how deep is the fear of earthquake and fire. To the extent that the carry trade is of Japanese domestic source, I think it reflects geographic diversification.
Again, I think the real issue is how much of the carry trade is borrowed money.
anonymous — agreed. which is why the notion of looking at the basic balance — which includes portfolio investment — makes sense. if lots of “real” rather than borrowed money is leaving japan looking for yields, it should show up as a portfolio outflow, reducing the surplus in the basic balance. looking at income payments as toffelmire suggests is another way of getting at the same issue.
the merrill study apparently found a surge in global banks borrowing in yen — which suggests borrowed money, not real money. and the absence of net portfolio outflows also suggests a signigicant role for borrowed money. though if a domestic japanese investor borrowed yen to buy $ bonds on margin, that should show up as a portfolio outflow –
i am still trying to understand the data. the absence of stronger portfolio outflows from japan (and on the US side, stronger recorded Japanese purchases) puzzles me a bit, given everything else that has been written.
The fact Japanese retail flows so desperately reaching for yield denied them by domestic monetary policy, are, as some posit, rather important, and [perhaps] aliasing some portion of the overall YEN carry phenom should provide little comfort to markets in general, and to those dollar_bulls/yen_bears alike. For Japanese retail are notoriously some of the most mimetic, and least contrarian investors in the known universe, albeit with a most unprofitable lag (call it LILO - last-in, last-out). They may, on dollar dips, provide support on the way down, only lately entering a trade they should be exiting, then some time later, upon discovering their folly (and capital losses) en-masse exit the positions at pain thresholds much departed from their entry points. I can say this with some authority having profited handsomely and more or less continuously from them for the better part of two decades. There are few as reliable telltales as the apparition en-masse of Japanese retail in a trade….
Eyeballing some of the balance of payments data (including January 2007 preliminary), it looks like flows in the order of the sum of the current account surplus plus net portfolio inflows are being recycled through the Japanese banks. The banking system also shows an enormous outstanding (i.e. cumulative) net foreign asset position. Perhaps recycled dollars are exchanged for yen deposit liabilities, but it seems impossible that domestic banks would run outright short yen positions (i.e. carry trade positions) of such size. Perhaps much of the net foreign asset position is swapped into yen, while still classified as foreign? This would push natural short yen positions outward.
Hm. Comment lost in transmission.
Briefly: Brad, I see now in comments where you addressed the issue of borrowed money vs. investment flow.
If we look at the various estimates, I think they perhaps break into three categories: (1) total Japanese holdings denominated in dollars, including governmental and corporate reserves, (2) long- and short-term external Japanese investment excluding corporate and government holdings (Jen’s figures are based on retail bonds), and (3) actual short-term speculative flows, based on borrowed money. If I understand correctly, the latter figure is estimated by both Dizard and Koll to be in the tens of billions.
With his $1T testimate, Koll is pointing out, I think, that if the dollar shows a consistent trend of weakening against the yen, long players will rethink their positions. At that moment, up to $1T in dollar-denominated assets held by Japanese could come into play. But of course, so could many more dollars of Chinese-held and Saudi-held and European-held assets.
Could this be simple differences in what is being measured?
it is the more than differences in what is being measured — koll’s measure of offshore yen lending is a lot higher than the BoJ/MOF number cited by dizard. and both in a sense are leaving out the real money (institutional and retail) flows that Cassandra mentions, which are driven by low Japanese yields but aren’t really leveraged. koll gets a bigger number i think (I just got the paper and need to read it carefully, not just look at the summary) by assuming that the borrowing in yen just covers margin requirements on derivatives (i.e. you have to post some margin, but can gear up off balance sheet) to various leveraged players real exposure to the yen is much much larger.
I agree that the banks wouldn’t want to run (or typically be allowed to run) a big open fx position — but couldn’t some of their NFA be yen lending to non-residents? that leaves them matched, and puts the currency mismatch on the books of the offshore investor (i.e. the levered carry trader).
While I will readily concede that Cassandra has superior experience with and knowledge of Mrs. Kobayashi and the rest of Japanese retail, I would nevertheless offer two observations.
Insofar as s/he is correct about the poor investment choices of Japanese retail, I would humbly submit that the most glaring error currently is retaining 50% + of household assets in non-yielding deposit accounts. As such, diversifying away from such deposits is correcting, rather than committing, an error.
Moreover, real-ish time data from Japanese leveraged retail FX platforms reveals a stunning sophisitication. Rather than being the herds that many of us have come to know and love, they have demonstrated a clear propensity to sell yen when it goes up and buy it back when it goes down. Insofar as these flows represent the levered-money carry trades that seem to be the focus here, these traders’ behaviour is actually quite encouraging. Gaitame.com I believe is the name of one leading (japanese language only, sadly) retail FX broker.
MM - Since I concur that reality on the ground is such that Japanese savers have everything to fear in respect of Japanese’ authorities efforts to impoverish YEN savers, the retail flow away from Nippon is indeed, as you highlight, rational.
However, rational as the outflow might be, I might point out that there is an archetypical genre of horror film where the recurring plot has the attractive heroine, sweaty & dirty, twisted ankle, torn clothes revealing perhaps a bit too much cleavage, having the escaped the clutches of seemingly heinous villain #2, finding safety and comfort with a seemingly reliable trusted friend of the opposite sex, only to find out (with the accompaniment of very melodramatic music) that he is NOT to be trusted, and in fact, is the axe-murdering heinous villain #1.
(PS - I like your musings, MM, though I must tell you I think I was somewhat unfairly “dissed” by your posse in pointing out the direction of the now-requited tail risk in mid-Feb).
Ah, Cassandra, you flatter me…I can only aspire to having a “posse.” In a sense, though, I think we were both right.
You (correctly) pointed out the short-option payoff profile of the carry trade. This has, ex-post, proven to be correct.
My contention was that positioning was not substantial enough to generate a 1998-style liquidity vacuum. Ex-post, this has also proven to be correct, as one could have sold USD/JPY 3% from the top the day AFTER A-shares and the SPX melted lower. While some of the other yen crosses have experienced more drawdown, the unwind has been relatively orderly in comparison with past episodes in 1994, 1997, and 1998.
MM -
I would counter-contend that positioning IS, in fact, sufficiently substantial to generate a 98 event. Importantly, however, the most meaningful positioning is of course “Official” in nature, for the moment emboldening both the spec side and retail flow sides of the market, and so I agree with you that unwinding of non-official positions (in the absence of official unwinding or signalling of similar intention or reserve accumulation policy change) will be more restrained.
But make no mistake, there is a huge fat tail that remains ensconced on the side of yen strength, and every spec going to bed each day with short JPY/USD is faced with the horror of waking up, or receiving the dreaded 300am phone call saying blah blah policy change 15 handles no trade vaccuum blah blah. Perhaps Munch was, or had a close friend who was a Macro Spec??!?
I defer to your expertise on calculating economic statistics, Brad. But usually when experts disagree by 1 or 2 orders of magnitude, there is some difference in methodology.
As for the foreign financial institution borrowing from a Japanese bank in yen, yes. That sounds like one likely source of mismatch. But then the catastrophe, if any, will probably impact the country of origin of the foreign financial institution first, since they will probably have to liquidate local assets and convert them into (strengthening) yen.
So I see the risk of this “carry trade” to be more to the US market than the Japanese.
anonymous — Dizard, citing the Japanese authorities, and Koll are trying to measure the same thing — speculative positions financed in yen. I am not sure how the BoJ/ Mof estimated $20-40b. Koll looked at foreign bank borrowing in the japanese interbank markey, using BoJ data. it has gone up significantly. And Koll showed that foreign banks borrowing yen in japan seem to be lending those yen to NY and london, not lending them inside japan … if memory serves, Koll estimates that foreign borrowing in the interbank market now totals around $80b — which is more than the estimate in Dizard.
But there is a difference in methodology. Koll argues that the “visible” yen borrowing in the interbank market basically provides the yen to meet margin requirements on a much larger derivatives position — hence a much larger estimate. See the material from Rozanov that I posted here in the past …
Re: why hasn’t there been a bigger move if there are large short-yen positions, i think the answer from those who believe that the yen carry trade is big would give is that the move to date hasn’t really forced a massive unwinding of short-yen long-carry positions (some, sure, but not huge), e.g. a lot of positions remain outstanding even after the events of the past couple of weeks. that seems to be the view of the guys at the BoNY, who argue that lots of short yen positions remain on the books of lots of players. Are they right — I don’t know. But it strikes me as a consistent story.
Cassandra, I will readily concede that if SAFE publicly announces ‘we are selling 200 billion USD/JPY at best, where’s the bid?’, we may get the sort of price action that you describe. Failing that (unlikely, in the literal sense)outcome, I just don’t think we’re going to get that kind of move.
Recall that in early 2006, after a sustained period of yen weakness in 2H 05, the central bank of Russia announced one Monday morning that they were going to incerase the yen weight of their portfolio. USD/JPY, trading 119.60 at the time of the announcement, didn’t regain those levels for a year.
However, the downdraft was hardly a freefall; although USD/JPY did trade 10 figures lower over the next few months, the market had gotten itself long yen in aggregate by about 115.
In terms of fat tail events, I am personally much more concerned about the structured credit market than I am USD/JPY (not to say that a collapse in the former wouldn’t generate drawdowns in the latter). While my ignorance of the CDO market is profound, my perception is that many of the holders of this stuff will be forced sellers in the event of downgrades/defaults, whereas Mrs. Kobayashi is unlikely to get the equivalent ‘tap on the shoulder’ in such a timely manner.
Thanks for your comments, Brad.