Now that sovereign CDS traders are about to reprise the role of Jason Bourne, and be hunted by international intelligence agencies just because under the not so wise advice of their prime brokers and preferred CDS salespeople, they dared to buy a minimum amount of $5 million in 5 year CDS of [Spain|Portugal|Greece], it is worthwhile to expose this sovereign CDS "thingy" once and for all. The following BofA research report [fn]Sovereign CDS are options too, like swaptions; February 12, 2010[/fn]will introduce not only the basics, but get into some of the more arcane concepts for those who feel that the need to roundhouse Spanish intelligence officers is about to reach boiling point (call it 30-bp spread induced synesthesia).
First, in theory...
Sovereign CDS are the favourite instrument used by investors to target countries with fiscal troubles and debt-rollover problems. Since the global crisis started three years ago, governments’ role has grown exponentially worldwide, thanks to the efforts of all Keynesians. The resulting transfer of risk from private to public sector finally puts a price on government credibility, which is being reflected in the rise of sovereign CDS levels globally. Back in the days when Enron/WorldCom caused a spike in credit spreads to a level implying a 10% cumulative 5y default probability for A-rated firms, few believed that one in 10 US investment-grade companies could close their doors in five years. Not only did historical evidence not support such a view, common sense seemed to go against it. What do we have now? Even after prospects for Europe backstopping Greece have risen significantly, the default probability for Greece is still pricing at 25%. Now, Italy is the new single A, with 10% default probability assuming 40% recovery. There seem to be few “invincible” countries left in the world. The market puts a 4% likelihood of a US default and that ranks the seventh safest in the world. Things go downhill from there with the next level being China matching UK, both at 7% chance of default.
To position for a default, there is little dispute that the sovereign CDS is the most direct and effective instrument. However, it is worthwhile asking what is the real likelihood of a sovereign default. Of all the 57 countries that have sovereign CDS quoted on Bloomberg (SOVR <go>), the average stands at 203bp, which implies a default probability of 16%, again assuming 40% recovery. We are looking at nine countries out of these 57 to default if the world does evolve according to these metrics. Is the market too pessimistic or are we missing something here? Think about what happened when the US market was pricing in a 10% default probability for all ‘single A’ names back in Enron/WorldCom days and what happened to those investors who bought protection at the peak.
Instead of default, may keep printing
No, we are not suggesting investors sell sovereign CDS protection blindly simply because we think the market may be too pessimistic about the world when looking at such high implied default probabilities. Indeed, worldwide government intervention financed by printing money may be so disconcerting that we think the sovereign issue will stay with us for an extended period time. However, there are several things investors may want to pay attention to, and the relative value play does exist.
Although we have seen multiple incidents in the past of sovereign defaults on foreign liabilities, it is quite rare for countries to refuse to make payment in full in nominal terms on their domestic-currency-denominated debt. However, for countries that rely extensively on external financing, there is a chance of default; because you can’t print money to pay back foreign liabilities. For other countries that also have high and increasing debt burdens denominated in domestic currencies, the strategy may be different. We should not expect default; instead, we should form expectations on interest rate levels, yield curve shapes and volatilities. Interestingly, the second camp spans a large set of countries, including those many are familiar with: US, Japan, Germany, China, and many others.
Linkage of rates, volatility and CDS
Interestingly, when an economy starts to emerge from a deep recession, inflation tends to stay low for quite sometime due to the large overcapacity to be absorbed. The absence of inflation supports relatively loose monetary policy. If uncertainty about the recovery persists, the government will also likely err on the side of providing sustained support to the economy through active fiscal policy. Putting the two together, we usually get a yield curve that tends to be steeper to reflect increasing concerns on the government’s fiscal discipline and financing capability. Hence, wider CDS should also mean a steeper yield curve. The correlation between the rate level and the sovereign CDS spread, however, is less certain since a weak economy tends to depress the level of interest rates while concerns over government finance tends to push it higher.
What is more interesting is the relationship between the volatility and sovereign CDS spread. From a theoretical perspective, both capture the value of options. In both cases, one pays a premium and waits until the government runs into trouble with its finances. In this sense, the implied volatility and the CDS spread should see a positive correlation. The positive correlation should also express itself through market dynamics. Wider CDS reflects higher uncertainty which causes the market to whip around on each piece of positive or negative news. The increased realized volatility should also have an impact on the implied volatility. However, the relationship is more complicated. In addition to the usual factors impacting volatility, such as the mortgage dynamics in the US and structured products hedging in Japan, a government’s response to its financing difficulty may drive a wedge between swaption price and the CDS price. As we argued, for those countries with little external debt, printing money is always an option (typically a preferred one) for the government to avoid default. Thus, though interest rates may rise sharply, there may not be the default that allows CDS protection buyers to realize their reward. Thus, the correlation between the volatility and CDS spread may be weaker for these countries. On the other hand, for those countries with more external debt, the correlation should be higher, especially during periods of heightened sovereign risk.
The correlation matrix among sovereign CDS, volatility and volatility skew over the last 13 months reveals some interesting patterns. First, the correlation among the CDS is extremely high, reflecting the potential contagion effect which is prevalent based on historical observation in the emerging market crises of Latin and Asian cases. It also reflects the illiquid nature of sovereign CDS market. One needs to source multiple markets to build meaningful risk for a portfolio of reasonable size, forcing an order-driven correlation. Second, the correlation of volatilities is much lower and less certain. It ranges between -0.3 and 0.5. We try to choose the part of volatility surface which is relatively immune from mortgage and structured-hedging issues to isolate the effect of sovereign risk concerns driven by investor positions. We are not surprised by the dispersion of volatility correlation and believe it simply reflects the fact that factors affecting the volatility are too many to be removed cleanly. What is encouraging is the positive
correlation in 100bp-payer skew. They are all positive, though the magnitude is lower than that of the CDS. Similar to CDS, the correlation reflects investors’ unease over government actions that could potentially cause higher interest rates. What comes out as a surprise is the correlation between sovereign CDS and the volatility skew. They are negative! Even for the CDS spread with its own country/region’s volatility skew. It can be argued that if the sovereign risk rises, the interest rate is much more likely to rise, and there should be enhanced interest in buying high-strike payers to move up the skew at the same time as the CDS widens. However, for this to be true, we have to assume that the investor base is the same and they are equally likely to invest in both. Unfortunately, none of the assumptions seems to be true. The investor base does intersect at the macro camp. Beyond that, the swaption market has traditionally belonged to rates investors, until last year when the field attracted many equity and credit investors who were very concerned about hyperinflation spoiling their feast. The expansion of rates-volatility as an investment choice reflects one advantage the rates market enjoys – liquidity.
By using a shorter period to calculate rolling correlation, we found that the correlation between CDS and volatility is extremely unstable. For example, the correlation between US CDS and USD 2y30y volatility varies between -0.87 and 0.91 over the past one year. Similar results can be seen from EUR 2y10y volatility and German CDS. We view this as good news and there are some investment implications for both the macro-oriented and relative-value investors.
For macro-oriented investors, if the view leads to the conclusion that moneyprinting is unsustainable on a global basis, it is better to buy CDS protection and buy volatility through payers at the same time. Such a strategy should work even better for those countries with higher external debt dependence. Lack of correlation between the volatility and CDS at normal times would provide diversification benefit when both move sideways. In the event of a spike in sovereign risk, the volatility will rise with the CDS spread and positions will become correlated in favour of better portfolio performance.
For investors who are looking for relative value, however, we think that the default risk may have been overpriced at current levels, especially for countries like US and Japan. As such we would think selling CDS protection and buying payers may be a better trade. In this case, the correlation does not help that much due to the lack of it. However, in extreme cases where these countries face the prospects of immediate default, we expect the volatility will likely move up along with rates to offset the loss in CDS spread.
And now, in practice:
When Germany comes to the rescue
News on Euro sovereigns continued to shape performances in all European asset classes this week. Somewhat more concrete developments on the political front allowed for significant tightening in peripheral spreads as well as temporary pickups in risk appetite. The odds of a complete disaster scenario have declined but nevertheless, concerns regarding the fundamental differentiation between core and non core Europe are likely to subsist in the long term, in our view. In addition, open questions on the exact nature of the support and its implementation (including the resulting cost to be bared by core countries) are likely to maintain pressure on both core and peripheral spreads and keep volatility close to current levels.
The ECB can pursue its gradual exit
While some are hoping to hear about a possible extension of the relaxed ECB collateral rule at the March meeting, the reality is that the ECB is rather heading in the other direction, having already in mind the exact steps to remove its excessive liquidity provision. In comments made on 11 February, Weber provided a more specific description of the “gradual exit”, noting that first the maturity of outstanding liquidity will be reduced then that the “most likely next exit issues will be a gradual return to normal tenders in longer term operations”, meaning 3M LTRO operations will move from full-allotment back to fixed amount/variable rate.His comments suggested on the other hand that unlimited allocation via the 1W MROs should stay in place for longer, inline with our view that liquidity will remain abundant in the system in 2010.
A normalisation of sovereign credit curves is triggered
Inline with the usual positive relationship between spread levels and CDS curve slopes, the substantial tightening of CDS spreads/ bond peripheral spreads following first the headlines on the preparation of a potential aid package for
Greece was accompanied by a steepening (ie normalisation) of CDS and ASW curves.
The reduced probability of an imminent default prompted a stronger tightening in short term CDS spreads. The Spanish 3s5s CDS curve has now disinverted, ie the 3y Spanish CDS is now tighter than the 5y, but most impressive is the steepening observed in Portugal and Greece (6.5bp and 4.6bp respectively – see margin table). Going forward, we might see a short term breakdown of the positive relationship between spreads and slopes as the reaction this week from European officials will likely keep away concerns of an imminent disaster scenario and additional spread widening is therefore likely to be greater on the longer end, resulting in a steepening of CDS curves.
Bond ASW curves have also experienced some normalisation, with a stronger ASW richening of short term bonds. The 2y10y ASW curve remains however inverted in Greece (see Chart 14). The shape of ASW curves in peripherals is not only function of the risk assessment but also depends on distribution of coming supply. In Greece, the new 10y to be syndicated by Greece in February is likely to be the next key driver of the curve.
And signs of a transfer of risk can be expected
The next development in our view is a relative widening of German, French and other core countries’ CDS relative to peripheral CDS. If a concrete decision on Greek financial aid is taken at the Ecofin meeting on 15-16 February (not the base case scenario according to latest headlines), a guarantee for Greek bond issuance for instance, we expect the CDS market to react in a similar way as when sovereigns set up guarantees for their banks (see charts next page). The current high risk associated with peripheral countries would be partly transferred to those now deemed safe with the extent of the move depending on the involvement of each country in the rescue package. On the other hand, if no concrete announcement is made, it is also likely that CDS spreads for core countries will continue widening, as risks of a weakening of the overall Eurozone increases.
On the bond side, a concrete announcement would likely result in a compression of peripheral spreads as German bonds would likely sell-off (both due to a return in risk appetite and as a result of speculations on increased German borrowing needs to finance the Greek bailout) while peripherals would rally on lower credit risk.
And dissecting the CDS-bond basis...We sure hope the Spanish spies read this to realize how unfounded their prosecution of CDS traders is
Basis at much lower levels than in Jan-Mar 2009 Although most sovereign CDS spreads broke their Feb-09 record highs in the past two weeks, this did not translate in new highs for the CDS-bond basis. The latter was rather well contained for all countries (see chart 17 with SovX vs cash and chart 18 with single name basis, based on bond spreads to OIS).
The fact that five syndicated auctions took place in January could be one of the reasons behind the strong sell-off in peripheral bonds and therefore the tighter basis compared to Feb-09. Also, this crisis being more driven by fundaments (e.g. high deficit ratios, fears of unsustainable debt) rather than a global increase in risk aversion, it is reasonable to expect a more pronounced reaction in the bond market. Long term bond holders find more reasons to be concerned.
Since 11 January, the basis declined for Greece and Ireland (see Table below) while it increased most in Portugal and also Germany and France. Demand for bonds in the latter two held very well as a result of a flight to quality and liquidity.
Sharp jump in short dated CDS flattened basis curves
Up to the middle of this week, a very strong sell-off in the short end of peripheral curves resulted in a flat bond term structure for some countries (chart 14). However, it remains that the inversion of CDS curves was more dramatic, resulting in a flattening of basis curves across countries. So why the sharp jump in short dated CDS curves? Part of the movement was due to a sudden rush to hedge default risk – some banks were using short dated CDS to cover their country exposure. In market panics, CDS spreads react faster than bonds. Short dated CDS was the easiest (and in some cases the only way) to hedge country risk. As mentioned above, short dated CDS spreads have tightened this week, but for most countries the basis remains wide in the short end.
1) Tightening of German basis We believe in a tightening of front-end German CDS-bond basis. If Germany announces a concrete intervention to help Greece, its CDS should widen (see transfer of risk above) but German bonds could suffer to a greater extent. They will become less attractive in the higher risk appetite environments while also being pressured by heavy supply in the 2y and 5y.
2) Cross country: We like to go long Belgian basis versus Italian one in the 5y. Another trade to play a normalisation is to go short the French basis versus the Dutch one. France has cheapened much more than other AAA countries on the CDS side (see Table above), either due to fears regarding the exposure of French banks to Greece or its high rollover risk. Yet if the situation regarding Greek banks deteriorates we believe it should start to impact further on Dutch credit as well, while on the rollover risk, we highlight that Netherlands has a similar 27% of debt to refinance by year end.
Drivers of the CDS-bond basis
Theoretically, both CDS spreads and yield spreads to risk free rates represent premiums received when taking an exposure to the risk of default of an entity. However a basis can exist in practice between CDS spreads and bond spreads.
Difficulty to short in cash market (increasing the basis)
Sourcing of bonds in the repo market could be difficult. Banks tend to buy CDS instead of shorting cash bonds in order to hedge their credit exposures or to express a negative view on a single name. This pushes the basis up.
Difference in relative liquidity (basis increasing with high bond liquidity)
The basis will move in such a way that provides extra compensation for investors in less liquid segment/markets, favouring countries with very liquid bond markets. This is also true, when it comes to sectors; liquidity in CDS market concentrates on maturities in 5 and 10years, whereas in the bond market, liquidity is already high in the very front end and can extend to the 30y.
Issuance patterns (mixed effect, usually decreasing basis for govies)
Banks involved in bonds syndication tend to buy protection in CDS markets during the issuance, causing the basis to widen (mainly observed in the corporate side). On the other hand, new bonds (especially govies) are often issued at a higher yield in order to attract sufficient interest during auctions, depressing the basis as a result. This was for instance the case at the launch of the 5y Greek bond at the end of January and at the Portuguese 10y syndication on 10 February.
Funding issues (decreasing the basis)
Not all investors are able to borrow at repo levels. Some of them may find it easier to obtain credit exposure by selling CDS than by being long the bond. This makes CDS spread relatively low versus bond spreads (to OIS / repo). This is particularly true at year end, when balance sheet reductions are working against cash positions and contributing to the cheapening of the bonds versus CDS (off balance sheet instruments).
Difference in players
The divergence in the types of investors in each market, with bonds mostly taken up by pension and insurance funds, while banks and short term investors are more active in CDS.