Wednesday, July 29, 2015

I Will Be Back: The Greek Crisis Redux Version

Here from Kathimerini

SYRIZA’s central committee is due to hold an emergency meeting Thursday in an attempt to find a way to settle the growing rift within the party over whether the government should agree to a third bailout or not.

Tuesday, July 28, 2015

US Asset Flows + The Thing That is Chinese Stock Market

Flows into/out of/ within the US. Treasury released the TIC data last week (for May). This is how it looks (international flows in to the US)
International flows in to treasuries picked up this year since Q3 last year, mostly driven by private flows (as opposed to official, i.e. other central banks and sovereign wealth funds). The total official year-on-year growth in flows in negative territory for the first time (apart from a flirting with it early 2014). However total flows still positive, money continues to be pumped into treasuries by foreigners. Negative flows since March this year mostly driven by the financial centers, i.e. the UK, Belgium, so is China (Mainland) and Hong Kong put together. However, Caribbean (presumably a good proxy for hedge fund flows) has been quite positive.
On other assets, the corporate bond turned a corner and now in positive territory, while equities outflows continue.
On domestic front, the total equities flows to mutual funds flattened out mid last year, but allocation to world equities still strong. And bonds funds flows have seen a comeback since start of the year, after a minor hiccup Sep last year (which is interesting, as that coincides with a strong sell-off US equities). Also money markets flows picked up, perhaps pointing towards a more defensive stance among fund managers and other institutional allocators. Flow to bonds outside US (including ETFs) has diminished quite a lot since Jan, after a strong pick up last year around March (not shown here).
Key takeaways: 1) treasuries are yet to be become hot potatoes among investors in view of rising rates 2) too many folks missed out the strong sustained equity rally in the US and in no mood to get back 3) overall institutions seems to be holding most cash in recent times. This hardly makes a case for a sharp correction in US equities - not by a rate hike at least. A force from outside is a different matter altogether.

And on the later point, here is a comparison between NASDAQ Composite and Shanghai Composite from a long term investing point of view. This shows average growth of the cash indices vs. volume weighted price (VWAP) since beginning July last year (when the Chinese equity rally took off, click to enlarge).

The point is, Chinese equities are still up on YTD basis, even after the recent carnage (15% YTD and 83% if you invested start of last year). One way to estimate the damage is comparing the VWAP price with an average price. The scenario of VWAP being lower than average price is a really bad one, as that signifies volumes in selloff was larger and on a net basis offsets the gain during the rally. In plain English: more people lost money than gained. The chart above captures this. The VWAP and average lines are close to each other for NASDAQ composite but the VWAP is outperforming the average for Shanghai Composite so far. So as of now, the market is still up in China, and no hidden catastrophic pain than it appears. Of course two things can offset this conclusion: Who came in last? If it is the regular Joes (with a higher marginal consumption to income ratio) then it is bad for consumption part of the GDP. Secondly we have the recency effect, this kind of sell off is unnerving. Where do you invest in China if you are the little guy? real estates are not looking great, bank deposits is value eroding, and equities, they simply has taken you for a ride. Increase in central bank liquidity is not that effective in a risk-off scenario.

Upcoming bid data release: US Q2 GDP advance estimate and Employee Cost Index, both on this Thursday.

Wednesday, July 22, 2015

Cross Asset Correlation

Update on the cross asset correlation as the market settles down for another summer.
Overall we have a minor uptick in rates level vs. fly correlation and a minor downtick in rates level vs. slopes. Commodities and FX to yield curve slope picked up as well (and also between themselves). Inflation to rates level and slopes also picked up. Table below shows rolling 13 weeks factor correlation across assets (click to enlarge).
For EUR, the short end slope (2s5s) correlation is considerably strong to overall slope (5s30s). Unlike USD and GBP and much like JPY. In fact most of the correlation in EUR resembles JPY. Given the level of the rates, it is no surprise. However things that stand out are the 10s30s (much more correlated in EUR than JPY). And 10s30s is again the major difference in correlation structure between USD and GBP (both of which are poised for rate hikes after years). Click to enlarge charts below
Some other observations:
  1. 1) The 5s10s has strongly correlated to rates levels and slopes in GBP (more like EUR and much less like USD), which I think is going to start to reduce.
  2. 2) The recent peak in 5s10s30s fly to rates levels and slopes in GBP should go down from here
  3. 3) The positive correlation between EUR 5s10s30s to 10s30s does not make sense, and perhaps an anomaly. Position for a correction.
  4. 4) the 2s5s10s recently flipped in correlation to level and on a path of recovery back to positive territory (esp. to 5y area). Position to take advantage of this move.
  5. 5) Given who moves first, Fed or BoE, we should see a breakdown in correlation between front end rates and flies with respect to rates levels (e.g. 2s5s and 2s5s10s w.r.t. 5y). Given Fed is the favorite to make the first move, the higher correlation between 5y to 2s5s in USD compared to GBP does not make much sense. This supports a 2s5s steepener in GBP vs. USD (which I recommend for reasons other than pure correlation, see here).
Cross assets, we have seen a re-correlation between rates and break-even inflations, esp. for USD (and also for EUR).GBP continues on a weakening, although levels are higher comparatively. On equities, the global utilities remain strongly negatively correlated to yields (esp. USD yields).  Also S&P shows a very recent spike in negative correlation to rates levels and slopes, much unlike DAX or Euro Stoxx. in FX, JPY shows a much reduced correlation to USD rates, and EUR shows a pick-up in correlation to rates. In commodity space, the correlation to USD rates to broad-based commodities are dwindling, while the same for GBP has picked up. FX and equity correlation is strongest for EUR, and not much for USD or GBP. And interesting a strong correlation between AUD and MSCI world material sector index. And more interestingly the EM vols (JPM EM, JPMVXYEM Index on Bloomberg) has been strongly negatively correlated to US equities lately.

Wednesday, July 15, 2015

Blend It Like Beckham: An RV tool for Trading Skew and Vol of Vol

(Below is from rates perspective, but applicable to any asset class in general.)

the market price of skew in rates can be captured through the concept of implied and realized blend. The concept in principle is quite simple and intuitive. Usually in rates markets, the underlying (i.e. forward rates) is modeled as a normal process. That is, the absolute change in rates are independent. The other extreme is lognormal, which implies percentage changes (instead of absolute changes) are independent random moves. The actual realized dynamics can be anywhere in between. Let's assume the forward rate dynamics is as given below (sorry for the clumsy picture!)

Here beta is the blend parameter.

This captures the normal dynamics mentioned above, for a value of beta = 0, and lognormal for beta = 1, and any realistic dynamics for different values of beta.

It is interesting to see what it means in terms of skew pricing. For example if we are talking about skew measured in terms of normal (bps) volatility (as usually done for rates markets), a normal dynamics would mean (without any stochastic volatility assumption) a flat smile. Whereas a beta close to 1 would mean a volatility smile which is higher for payers (higher as yields goes up), and similarly a beta less than 0 means a receiver side skew. This is evident from the model above: e.g. for beta = 1, as yields go up, the ATM implied vol is priced higher, giving rise to a payer side skew.

Now just like implied and realized vol, this beta, or blend can also be thought of having two different sources of estimation. One from the current prices of options at different strikes. This is easy to extract, by taking the price of a (tight) collar around the ATMF and solving for the value of beta which prices the payer and receiver vols correctly (or rather their difference, i.e. the collar price), given the current ATM vol. This can be thought of as the implied blend.

For the realized blend, just like realized volatility, we can take historical ATM vols and regress against the ATMF yields. Note beta here defines the value at which the right-hand side of the model above becomes a random move. So running the above regression and changing the value of beta, we can pick the value for which we get the least R-square (worst fit). This value of beta can be thought of as the realized blend. Just like realized volatility, it will be sensitive to the amount and period of history we look back at.

Now armed with this two measure we can draw some interesting observations. To begin with, the values directly compare the realized vs. implied skew and can be used to find relative value in skew trades. Imagine a scenario where we have seen a high realized beta and a low implied one. This means the markets prices the receiver skews more favorably (implied), although in recent history volatility went up as yields went up (realized). If we assume this past behavior will persist, then it makes sense to buy a payer and sell a receiver (buy a collar or risk-reversal overall) and delta hedge. If yields indeed go up, the portfolio will be long gamma, and the implied vols will be higher than what was priced in. If the yields go down, the portfolio will be short gamma, but implied vols will be lower than priced in.

Some further observations: firstly, the co-movement of blends vs. rates can allow us to conclude about higher orders than skew, in this case about vol of vol. The blend (both implied and realized) will depend on the absolute level of yields. For very low levels of rates, a further rally will mean limited scope for yields to move downwards (by zero bound condition for nominal rates), where as the moves upward is unbound. This means in normal condition the yields and the realized blends should move in the opposite direction (vol stabilizing environment). If they move in same direction, that reflects uncertainty and high vol of vol scenarios (vol blow-up environment). This directionality can be compared with the current vol of vol priced in (say in a strangle or a fly) and use this method to spot relative value in vol of vol.

Here are some charts for the 10y swap point for USD (click to enlarge).

And for EUR (click to enlarge, all data from Bloomberg)

Notice how the market spent most of the recent times constantly re-pricing in USD under increased uncertainties as vol picked up (see here). In fact last October and early this year we have seen highest level of vol of vol for US rates since Lehman crisis. For EUR, initially it was normal. Only after introduction of a negative depo rate and imminent QE, blend caught up in lock-step with rates with a jump in vol as well.

Interesting time ahead.

Sunday, July 5, 2015

In Pictures: Greece Referendum

Greece has a highly concentrated populations, majorly around the Attica region

Which is reflected in the highly concentrated economies as well

These figures are relevant, as most of the upside and downside will be concentrated, no matter how the Greeks vote today. Most opinion polls are also around Athens so far (including this one from Bloomberg). Based on the last elections and the parties that are campaigning for "No" vote, here is the "No" vote share for each region.

Finally, given the perception that the young population more likely to vote "No", here is the population pyramid (from here). The old outnumber the young.

Brace for some rough rides.

Thursday, July 2, 2015

Move Over Greece: The Down-Under Version

Since the surprise start of the week, and large rally in rates, rates are almost back to the levels it lost, with steeper curve, especially in Euro. Break-evens are little budged (except a sort of swing in sterling) and dollar marginally stronger and Euro marginally weaker. Only equities still nursing the wounds so far.

I think my optimism about UK turned out to be justified so far. Data-wise US, UK continue on strong data flow. At the same time, I think we are focused on Greece more than it deserves and missing other big thing in the bargain - the unraveling of the Chinese stock markets. This will definitely have a strong influence on the Australian rates markets. I have argued before that the AUD curve is to steep and has to flatten. Since then it flattened from 100+ to 70bps level and now back to 105 (in 5s30s).

We can propose a simple model of the factors that drive the slope in line with similar models for Euro and USD. The explanatory variables are the terms of trade (CTOTAUD Index on Bloomberg), RBA policy rate, and the China current account balance (GDP is also somewhat significant, but not included, inflation not much significant). The estimates are as below.

The model fits as below (click to enlarge)

Post-crisis there is a change in unconditional expected level of slope, which is higher that what it used to be before 2008. But also a couple of changes interesting: a higher sensitivity to China current account balance (as percentage of GDP) and RBA policy measure. The RBA part is as expected. On the China sensitivity, one conjecture can be that current account balance shrunk in China primarily as exports reduced, along with a reduction in imports. That directly impact Australia, and builds in pressure for bull flattening.

This, along with the current directionality of rates with slopes means a very asymmetric positioning for a AUD flattener. If we see a strong return of job growth and commodities inflations, we will see a more hawkish RBA, leading to a flattening. If we see a further worsening in sentiments from China, but no deterioration in inflations, the correlation should lead to a bullish move. A separate estimates shows for each 1 percentage point reduction in China current account balance (% of GDP) leads to a 30bps rally in 5y and 17bps in 30y in AUD.

Given this, and the recent steepening and sell-off in AUD rates, it is a good opportunity to enter a trade. It can be either a flattener (probably balanced by a 5y point to take care of the directionality) or it can be a 5y or 5y5y outright receiver. And all of this will have a positive carry. Which can be used to pay for a sell-off protection in either EUR, GBP or USD long end.

All good, then there is this one possibility of course.