Andrew Rozanov of State Street knows a thing or two about how official institutions manage their money (that is a big part of his current job) and a thing or two about Japan (he was based in Tokyo for some time). In response to my previous post, he noted that much of the yen carry trade is now done through the swaps market – or off balance sheet. Hedge funds don’t really need to “borrow” yen and then buy higher yielding currencies – they do the same thing in the derivatives market, or pay a bank to do it for them. I have – with his permission – reprinted his comment below. It inspired a nice discussion.
UPDATE: State Street also seems to know a thing or two about the size of the carry trade. See the FT.
I took a number of things away from the discussion:
1) The obvious, what happens off balance sheet matters. The absence of strong growth in cross-border yen lending isn’t evidence that folks aren't betting on continued yen weakness – or that interest rate differentials will remain large enough to offset any uptick in the yen.
2) A trillion is probably a bit too high an estimate of the size of the yen carry trade, at least if we are talking about bets from the “leveraged” hedge fund community.
3) We shouldn’t forget about leveraged Japanese day traders … who, in aggregate, have become big players. Fx accounts with generous margin are the rage in Tokyo.
4) It is hard for competitive money managers to be on the sidelines and avoid putting on various kinds of carry trades right now.
Andrew Rozanov writes:
“as volatility remains low and carry keeps being profitable, for a commercially driven fund management shop it is becoming more and more painful to stay away from these trades, certainly if you are in competition with others in raising money from investors. Just plot the returns from a simple carry basket on Bloomberg's FXIP (I suggest using long GBP at 50% and long EUR at 50%, while short JPY 50% and short CHF 50%) -- and then look at the statistics below for one-year return, volatility and the Sharpe ratio. They are phenomenal! And what do investors more often than not look for when comparing and choosing hedge fund managers? Why, high and stable returns at relatively low volatility - strategies with very high Sharpe ratios! Never mind the higher moments and fat tails...”
I would note that volatility is also low for some emerging market currencies. Volatility in the Brazilian real/ dollar, for example, has collapsed – because the central bank has resisted market pressure for further appreciation in the real. The same thing incidentally happened in Turkey in 2005. Low observed volatility makes the real-denomianted carry trade even more attractive. And that pulls in even more money. It probably isn’t an accident that Brazil added over $5b to its reserves in January. There are a lot of different cross-currency carry trades out there that have attracted a lot of attention.
Andrew Rozanov’s comments are reprinted in full belowAndrew Rozanov:
I would like to commend those astute commentators who pointed out the direct relevance and central role of the forward currency swaps market in facilitating carry trades. It is arguably the easiest and most preferred method to put on a 'carry trade' by the leveraged community. It was described several years ago in a book written for a Japanese audience by a former Moore Capital trader, Ken Shibusawa, who currently runs his own hedge fund advisory firm in Tokyo (Japanese website for his company: http://www.shibusawa-co.jp/index.html).
According to him, essentially the press has got it all wrong. Yes, the underlying economics of a typical carry trade is to 'borrow' in low-yielding currency (e.g. the yen) and then 'lend' on in higher-yielding currency (e.g. the dollar), but when we discuss the actual mechanics of it, hedge funds and other highly leveraged players typically would not engage in any actual borrowing or lending. To the extent a hedge fund would consider putting on such a trade at all, it would most likely do it in the following way, or some variation thereof (we use US$/JPY levels as in Ken Shibusawa's example in the Japanese book):
Step 1. Buy US$ / Sell JPY in the spot market (say, at 120)
Step 2. Buy JPY/ Sell US$ in the spot market (again, at 120)
Step 3. Buy US$ at a discount / Sell JPY at a premium in the forward market (say, 3 months forward at 118.50)
Step 4. Buy UST in the spot market
Step 5. Borrow US$ against UST in the repo market
Some comments are in order.
1. A typical currency forward swap is step 2 + step 3. However, if that's all you do, you will be expected to deliver your US$ for settlement on T+2. But remember, we are looking at a hedge fund, who does not have the money to deliver and who wants leveraged carry. Thus, he needs to create this exposure with as little initial cash outlay as possible. The best way to do that would be to take an outright forward position in the market.
2. An astute observer will have noticed that the cumulative effect of steps 1 through 3 is precisely to create such an outright forward position, whereby our hypothetical hedge fund is now short the yen and long the dollar at 118.50 on a 3-month horizon (in other words, the hedge fund promised to pay 118.50 in 3 months time for an asset which is currently trading at 120 - and the lower the US$/JPY volatility, the higher the chance that the forward contract will have converged at or around the original spot price of 120).
3. But if a hedge fund wants an outright forward position, why go through the hassle of steps 1-3? Well, it all has to do with costs and liquidity. Apparently, it is cheaper and easier to put on such a position in the currency swap market rather than as an outright forward - as the market supporting corporate and 'real money' hedging activity, it is the most liquid.
4. Notice how the trade is structured in two separate legs. There is the FX market leg, where the hedge fund puts on a position to earn the carry from the interest rate differential in the two short-term markets. Then there is the UST leg, which is a completely separate transaction: longer duration US Treasuries are purchased to earn a higher yield, but they are financed in the repo market (of course, this example assumes a 'normal' upward sloping yield curve in the US). So if (and it IS a huge if) all goes as planned in the US$/JPY and UST markets, the total carry earned on the trade is the sum of FX market carry (i.e. short term interest rate differential) and UST market carry. Another way of putting it: the hedge fund earned money from two major risk sources: currency mismatch risk and duration mismatch risk.
5. In his book, Ken Shibusawa makes an interesting observation: many press commentators focus on what they can trace in the spot markets (i.e. step 1 and step 4), and often do not pick up on simultaneous transactions in derivative markets (i.e. combination of step 2 & 3, step 5). Thus, the tendency is to interpret this as hedge funds borrowing yen and selling yen spot, thus procuring dollars and then investing these dollars in long maturity UST). While the underlying economic logic and principle is the same, the mechanics are totally different.
6. Ken Shibusawa quips in the book that while much press attention at the time was focused on global macro hedge funds, it was probably relative value shops that were particularly keen on putting on massive yen carry trades in 1997-98, following the relative success of their "Japan risk premium" carry trades in the previous years.
One final comment from me: as volatility remains low and carry keeps being profitable, for a commercially driven fund management shop it is becoming more and more painful to stay away from these trades, certainly if you are in competition with others in raising money from investors. Just plot the returns from a simple carry basket on Bloomberg's FXIP (I suggest using long GBP at 50% and long EUR at 50%, while short JPY 50% and short CHF 50%) -- and then look at the statistics below for one-year return, volatility and the Sharpe ratio. They are phenomenal! And what do investors more often than not look for when comparing and choosing hedge fund managers? Why, high and stable returns at relatively low volatility - strategies with very high Sharpe ratios! Never mind the higher moments and fat tails...
And, for those of us without the financial sophistication to understand why anyone would both buy US$ for yen and sell US$ for yen in the spot market, Andrew noted:
Let me try and answer your question about why go through steps 1 and 2 by considering the other options.
1. If you do only step 1, you are going long USD and short JPY in the spot market. There are two problems with this approach as far as leveraged carry trades go. First, it's not leveraged - you will need to come up with the entire notional amount of JPY to settle this trade in two working days (T+2), so you're not really 'borrowing' anything. Secondly, you're not really trading for carry. Instead, you're putting on a directional bet: you buy US$ at 120 and hope it appreciates (or JPY depreciates), but you're not picking up any interest rate differentals. Incidentally, this type of trade typically favours high volatility environment - remember, volatility cuts both ways, so if you want to make money from a large swing up in the dollar (or swing down in the yen), you would be implicitly betting on high, not low vol. Carry trades, on the other hand, are typically low vol trades: any additional depreciation of the yen would be icing on the cake, but your main bet is that the world stays stable and predictable while you earn your carry on a leveraged basis (i.e., interest rate differential multiplied by leverage).
2. What about taking only steps 2 and 3 - i.e., the foreign exchange swap? The good news is this gives you carry: you buy 120 yen for every dollar now (on T+2) and at the same time contract to sell the newly acquired yen in 3 months' time, such that 118.50 buys you a dollar - in other words, you sold the dollar now at 120 and bought it back in 3 months at 118.50, earning the carry in the process. But the bad news is that, just like in the first comment above, you do not get any leverage, since you will need to provide an initial cash outlay on T+2 to settle the leg in step 2.
3. Why not forget steps 1 and 2 altogether, and just go for step 3? This is entirely legitimate and certainly doable. In the FX market, such a transaction is called an 'outright forward'. You agree with your counterparty bank to go long USD and short JPY based on today's spot price, but with settlement in 3 months' time - you will negotiate and agree the spot price, and then the bank will adjust it by the interest rate difference. No cash flows occur on T+2, you have effectively secured a leveraged position, and you've bought US$ at 118.50 in 3 months. Voila! The only reason you would consider doing something a bit more convoluted (like going through steps 1 to 3 above) is if it helps you save costs and earn even more money. And in competitive markets like FX you get the best prices and execution where the liquidity is the deepest. And the liquidity of spot and FX swap markets has always been (and I believe still is) much better and deeper than in the outright forward markets. Just have a look at Table B1 on page 5 of the BIS Triennial Central Bank Survey at the following link (last survey done in 2004):
Hellasious and Tmcgee also chimed in; Hellasious noted the growth in outstanding yen forwards. Tmcgee argued that these are notional numbers, so they only give a loose idea of real exposure. He also noted the role of Japanese fx margin traders – something that the FX team at Citi also likes to highlight.
Hellasious and Tmcgee also chimed in; Hellasious noted the growth in outstanding yen forwards. Tmcgee noted that these are notional numbers, so they only give a loose idea of real exposure and noted the role of Japanese fx margin traders – something that the FX team at Citi also likes to highlight.
I urge everyone to visit the BIS site and look up the relevant FX forward data for yen. The amounts have jumped very substantially from $2.3 trillion to $3.8 in 5 years. A big part of the increase has come in just 6 mos., going from $3.1 to $3.8 trillion between Dec. 2005 and June 2006 (latest available data).
I think $1 trln is a slightly ridiculous number, even if you include the sheer Japanese foreign asset holdings that don't really constitute carry trades per se, just a hunt for higher yields. Also, the BIS derivatives data gives some clues on potential size of carry trades. But I believe they also are all notional numbers, like the US OCC data, which means they're always growing at a fairly rapid pace -- because both sides of the trade are counted. The funny thing is now people are freaking about the size of the carry trade even though one of the supposed culprits of the unwinding of carry trades last april/may (incorrectly) -- a drop Japan's monetary base from the BOJ ending quantitative easing -- is still shrinking 21% yr/yr. and other BIS data -- reporting banks' cross-border positions vis-a-vis all sectors, by domestic and foreign currency -- shows no explosion of yen liabilities among major global banks.
Reporting bank yen cross-border positions vis-a-vis all sectors (blns USD) liabilities:
Domestic currency: December 04: $240.9; Sept 06: $235.9
Foreign currency : December 04 $433.1; Sept 06 $457.6
We know the IMM speculative yen short data, so we have a vague since carry trades are popular. But as the Wachovia anlaysis points out, the evidence is far from compelling. ….
Also, no one seems to mention the role of Japanese FX margin trades in the carry trade. Check out data from the Tokyo Financial Exchange, which gives just a small peek into the explosion of margin trading here. These margin traders are quick to take profits when cross/yen hits new highs, but also buy the higher-yielding currencies (like the Aussie, on serious leverage) on any sharp pull-backs. And since so much money is waiting in Japan for moments of yen strength to invest abroad, it tends to put a floor under not only dollar/yen, but also euro/yen and sterling/yen …
BSetser edits: Data presentation edited for clarity, punctuation added.
And Cassandra always has a thing or two to say about the yen and yen carry trades.
Even if Tmghee is right and total carry trade exposure is significantly under $1 trillion, it scale of the carry trade seems to have increased significantly recently. And a lot depends on how the capital outflow from real money Japanese accounts over the past few years should be interpretted. Is it international diversification, money that is unlikely to return to Japan should relative interest rates change and/ or the yen start to appreciate? Or is it a bit more fickle, and perhaps be inclined to act a bit like leveraged money should conditions change?