Contents

CDR Liquid Index data as at 28 May 2007

Source: Credit Derivatives Research

CDR Liquid Indices
The CDR Liquid indices represent the CDS levels of the most-liquid names in their respective markets and ratings classes. Liquidity is measured by the number of bid-offers a credit receives. Index values are simple averages of on-the-run five year CDS levels.

CDR Global Market Depth™
The CDR Global Market Depth Index is a daily measure of how many names are actively traded. Liquidity is measured by the number of bid-offers a credit receives. Index values are counts of the number of names that exceed CDR's Liquidity Floor.

CDR Global Market Activity™
The CDR Global Market Activity Index is a daily measure of activity within the global CDS market. Liquidity is measured by the number of bid-offers a credit receives. Index values are simple averages of total bid-offers of all names that exceed CDR's Liquidity Floor multiplied by CDR's Global Base Liquidity Constant.

Spanish synthetic surge?

New CDO could be the beginning of a trend

Lehman Brothers and Bankinter have begun marketing Castellana Finance, the debut Spanish synthetic securitisation of first-loss and residual pieces. The fully funded deal is only the third synthetic CDO to hit the Spanish market in recent years, but could signal the start of a long-awaited increase in such transactions in the country as Basel II capital requirements start impacting banks' balance sheets.

As Jose Ramon Tora, director at S&P in Madrid, observes: "There have only been two public synthetic transactions in the last five years in Spain. The idea was for the banks to obtain some capital relief, but it does not appear that any relief materialised because we have not seen any subsequent deals."

He continues: "The difference this time is that the issuer, Bankinter, has probably got a much clearer indication from the regulator in terms of capital relief. As such, it has created a precedent and paves the way for other banks."

Tora adds: "It's easy to see which banks have securitised more frequently in the past and obviously most will be holding their equity or first-loss pieces on their balance sheets. I'm sure many are already in discussions with their investment banks."

Meanwhile, Castellana Finance involves the issuance of €182.3m floating-rate credit-linked notes: €83.7m of triple-A rated 5.6-year Class As, €54.7m of double-A 12.45-year Class Bs and €43.9m single-A 13.28-year Class Cs. At the same time, the Irish-based SPV will issue €2.84m of unrated Class D notes.

The purpose of the transaction is to transfer to third-party investors the credit risk associated with the reserve funds and subordinated credit lines in 14 RMBS transactions retained by Bankinter. The transactions were originated by the bank between May 1999 and March 2007.

Castellana is expected to close in June with a legal final of January 2050. On the closing date, the issuer will deposit the note issuance proceeds as cash with Bankinter (as account bank provider) and enter into a €185.14m CDS with the bank, under which the issuer will sell protection on the reserve funds and the subordinated credit lines of the 14 RMBS.

Bankinter, as protection buyer, will make quarterly premium payments to the issuer. In return, the issuer must make credit protection payments to Bankinter if a credit event occurs, subject to certain conditions.

The transaction is similar to Britannia's Dovedale Finance, as well as Northern Rock's Whinstone Capital Management (Whinstone 1) and Whinstone 2 Capital Management, which closed in the UK in July 2006, November 2005 and June 2006 respectively. Dovedale references the reserve funds from seven standalone Leek RMBS transactions; Whinstone 1 references the issuer reserve funds from three standalone Granite RMBS transactions and the master trust; and Whinstone 2 references a portion of the Granite Master Issuer programme reserve funds supporting its de-linked issuance (for series 2005-1, 2005-2, 2005-3, 2005-4 and 2006-1).

The portfolio in Castellana Finance references four subordinated credit lines (allocated to the Bankinter 1 to Bankinter 4 transactions); five subordinated loans (Bankinter 5 to Bankinter 9); and five subordinated notes (Bankinter 10 to Bankinter 14). In contrast to the UK deals, each of these RMBS separately benefits from a de-leveraging mechanism over time. However, the WALs in Castellana are considerably longer than those in Whinstone (seven years) and Dovedale (two to three years).

S&P notes that the Bankinter funds have the option to redeem notes issued in the standalone transactions before legal final maturity. They can do so on or after the clean-up call date, once the aggregate balance of the notes is lower than 10% of the original aggregate note balance. In certain transactions, Bankinter issuers could exercise their option, regardless of whether this would result in a crystallisation of losses for Castellana Finance noteholders.

MP

LCDX expectations emerge

Index and its tranches could become CLO manager benchmark

As enthusiastic early trading of LCDX remains, dealers have this week continued discussions over standardised tranches (SCI passim) on the US loan CDS index. While no start date for tranche trading has yet been fixed, there are strong expectations for the products, not least in the context of the CLO markets.

A new report from JP Morgan CDO research in New York says: "Overall, we think markets will segment, with real money accounts focusing on rated CLOs and 'fast money' accounts - hedge funds, prop desks etc. - relatively more involved in non-rated index trading. However, given that the latter do participate in CLO equity and cash loan markets (around 20-40% of volume), we think there could be spill-over effects. We also think LCDX may begin to function as a quasi-CLO benchmark, even though there are significant differences between the two products."

Fundamentally, both tranched LCDX and CLOs are a leveraged position in high yield loan credit. However, the report notes: "Tranched LCDX (as currently discussed) would be considerably more leveraged, due to thinner tranches at certain attachment points and less subordination. For instance, the 15-100 is not comparable to a CLO triple-A, since it attaches where a single-A CLO would attach."

The report adds that the lack of principal/interest diversion mechanisms exposes senior tranches in the tranched LCDX to greater risk, but may benefit subordinate tranches in some cases (for example, no PIKing in distressed scenarios). "The static nature of the portfolio arguably exposes junior tranches to greater single-name risk, as there is no opportunity for a manager to mitigate loss through pre-emptive sales (or to generate trading gains). However, the shorter maturity of the LCDX (five years versus 8-10) is an advantage for its tranches, and the lack of manager, underwriter and rating agency fees is positive especially for the equity investor," it says.

It is possible that LCDX, or tranches thereof, could be used as a benchmark for CLO managers - and the growing use of short buckets by managers is anticipated as opportunity sets expand (as, for instance, in the case of corporate CSOs).

But CLO spreads are unlikely to precipitously widen in a similar vein to structured finance CDO spreads following the introduction of TABX. In fact, structured finance CDO spreads had widened prior to the launch of TABX as ABX prices fell in reaction to deteriorating home equity loan fundamentals.

Loan credit trends - while marginally deteriorating in comparison to prior years - are still within or better than historical norms for leverage, interest coverage ratios and so on, and the economy remains solid. The JP Morgan analysts believe, however, that there may be an increased focus on capital structure relative value as market participants have the added ability to implement trade strategies and the benefit of transparent prices.

MP

Leveraged super seniors return

A number of LSS deals are expected to hit the market - among them is a first for the product type

Dealers report seeing some indications of super senior activity making a come back since last Wednesday, despite quiet trading surrounding the public holidays in the US and the UK on Monday. Such activity in the standardised correlation markets follows a rally in equity tranches, which has encouraged the preparation of a number of leveraged super senior (LSS) deals after a hiatus in issuance of the products.

Expectations are for at least two deals in Europe and as much as US$5bn worth of LSS issuance in the US. However, the equivalent delta notionals of LSS transactions are a fraction of the face value of a deal and consequently the activity had little impact on equity tranche prices even in thin trading.

Furthermore, because of the recent equity rally dealers were already short that piece of the capital structure. As a result, they had less of a need to buy equity protection to hedge the LSS issuance.

Notable among the deals being prepared is the first LSS of ABS to feature a spread trigger – Accolades I. The approximately US$180m transaction is issued by Starts (Cayman) Ltd, arranged by HSBC and managed by AC Capital Partners.

Moody's is the only agency to have rated the transaction's two classes of rated notes, both due 2037. The agency has awarded the US$120m Class A1 notes a provisional Aaa rating (with a leverage factor of 16.67x) and the €44m Class A2 notes a provisional Aa3 (35.08x).

The rated notes, together with an unrated euro-denominated equity tranche, are exposed on a leveraged basis to a portfolio of high grade diversified ABS. The US$2bn portfolio consists of euro, US dollar and sterling assets. The issuer is exposed to the portfolio synthetically via a total return swap undertaken with HSBC.

The reference portfolio will comprise of triple-A rated RMBS (up to a maximum of 70%), CMBS (30%), ABS and CDO collateral.

Market value triggers are used to reduce the exposure that HSBC has in its senior position as leverage provider. These triggers are breached if the market value of the reference portfolio drops below the outstanding notes' notional reduced by specific cushions.

At any time that the soft trigger is breached, reinvestment is prevented – thereby automatically de-leveraging the deal. But if the hard trigger is breached, investors have the option to either de-leverage the transaction by issuing additional principal, or unwind the transaction paying termination costs to the swap counterparty.

Moody's says that the modelling of this deal was based on a Monte Carlo simulation within the agency's CDOROM framework. The credit migration and spread evolution methodology developed for corporate LSS transactions was applied, with new parameters derived for the specific types of ABS securities that the transaction is exposed to. A bespoke component in the model was developed to capture the specific features of this transaction, including the market value triggers and sensitivity to FX rates.

MP

New frontiers

More Canadian fund managers to enter the CDS market and utilise new strategies

The investor base for Canadian credit derivatives is set to rise as more fund managers look to enter the market. One such fund is State Street Global Advisors (SSgA) Canada, which is soon expected to use credit default swaps for the first time in its Canadian investment portfolios.

"Our active mandates allow us to use credit default swaps and the firm is on the doorstep to start trading credit derivatives, which we expect will be in a matter or weeks," says Peter Lindley, senior portfolio manager, fixed-income group SSgA Canada in Toronto.

One of the ways that SSgA Canada will use CDS is as a synthetic diversification tool, to gain exposure to a credit not available in the cash market or in a particular maturity. More importantly, the firm will also use CDS in ways that are uncommon among many institutional investors in Canada to undertake basis trades and to short the market by buying credit protection.

"Dialogue with derivatives houses and clients suggests that no Canadian fund manager is really doing this," says Lindley. "It is our understanding that it is the shorting of credit by buying credit protection, that is somewhat unique, rather than the basis trades, which may be more commonplace."

It is understood that there has been some uptake of negative basis strategies by a couple of pension funds and hedge funds. However, it is uncertain how widespread this is among Canadian institutional investors.

A number of offshore and US hedge funds did short BCE on LBO rumours in the last two months, but trade sizes were small and this type of strategy is not prevalent among other investors. "I have not heard of a Canadian asset manager protecting his portfolio by buying CDS, if they have bonds they will just try to sell them," says, Edward Devlin, svp at Pimco in Newport Beach, California.

Lindley adds that, though using CDS as a strategic view is rare, he expects the strategy to gain traction, especially given the tightness of credit spreads across all markets. "We have the ability in our active strategies to go short through CDS. By buying protection, you can diversify a long-only index fund, which most pension funds and investors have," he says. "We have an asymmetric view of credit and so if you are good at selecting names that are going to have problems, then you are going to make a lot more money by being short them than you can by being long."

Lindley believes that LBO risk will be a continuing theme throughout 2007 and the volatility in spreads will provide greater opportunities to use credit derivatives. "CDS gives you an extra tool and versatility that you cannot always get with the cash market. If the market changes we can easily position ourselves in other ways, like basis trading," he says.

He is confident that the market is ready to take off in the short to medium term, but this is reliant on educating the end user, particularly the pension boards of the medium-sized pension funds in Canada, which he feels will be the main area of growth. "It is an ongoing process and a great opportunity. There is increasing familiarity and the market has a lot of potential to grow in the next couple of years," says Lindley.

Devlin believes that more institutions are close to being operational and anticipates a couple of the larger pension funds in Canada will undertake credit derivatives in significant size in the next year. "It is a slow and gradual process and more about approval than education for pension funds as they need to make sure all the risk controls are in place, such as oversight and governance. Many have been looking at these markets for some time and are realising there is an investment opportunity in these structures," he says.

HD

Morgan Stanley adds to loans

The latest company and people moves

Morgan Stanley adds to loans
Robert Lepone, head of European distressed and leveraged credit trading at Morgan Stanley, has added two personnel to his team. Robert Campbell joins from his position as head of loan trading at Barclays Capital as an executive director in August. Andrew Jarman joins from Deutsche Bank as an associate.

Both will trade high-yield loans, bonds and derivatives and report to Lepone.

Calyon appoints two
Calyon has appointed two leveraged finance professionals to its US credit markets division, both in New York. The bank has recruited Mariano Goldfischer, who will be in charge of developing loan trading activity, including cash and derivatives trading as well as managing the loan TRS trading book.

Goldfischer previously traded high yield bonds, CDS and LCDS instruments at New York-based Strategic Value Partners. He will now report to Thomas Spitz, Calyon's head of credit markets & CDO trading Americas.

The bank has also hired Alexey Dronov, who will be part of the US-based cash CDO team with a primary focus on CLO structuring. He was previously in the cash CLO team at Morgan Stanley.

BNP Paribas loses traders
Julien Mareschal and David Hoffman, ABS/CDO traders in BNP Paribas' fixed income - European flow credit trading unit have left. Their destinations are not known.

Economides departs ABN
Marc Economides, an ABS trader at ABN AMRO, is believed to have left the bank for an unknown destination. Angela Capone is understood to have moved on to the desk to replace him.

Schulte leaves Citi
Scott Schulte is understood to have left Citigroup sales and trading in New York. His destination is not yet known.

XLCA hires for CRE CDOs and ABS
XL Capital Assurance (XLCA) has appointed Kenneth Cheng as a director, leading the firm's efforts in the CRE CDO space. He will report to Sohail Rasul, head of CDOs at XLCA.

Cheng comes to the firm after ten years at S&P. His last role at the rating agency was as a director and analytical manager in the CDO group, with responsibilities including the creation and updating of ratings criteria. Cheng also co-led the ratings effort for commercial real estate CDOs.

XLCA has also hired Russell Albers as an md responsible for commercial ABS transactions and David Beard as an md responsible for underwriting consumer ABS transactions. All three executives are located in XLCA's New York City office.

Prior to joining XLCA, Albers was a svp of ABS at Friedman Billings Ramsey, where he was responsible for the establishment and development of the firm's asset-backed finance group. Beard joins XLCA from Financial Security Assurance Inc., where he was a director in the mortgage finance group.

ClusterSeven makes more appointments
ClusterSeven, a provider of enterprise spreadsheet management software, says that in order to support the rapid expansion of its business, it has added several critical positions to the organisation.

This includes a number of sales and support staff located in both New York and London. An alliance and business development manager will look after building strategic and new relationships to strengthen the relevancy of solution sets across markets.

Mike Finta, the latest strategic hire, has accepted the role of global vp of product development and will be based in London. In this capacity, he will extend ClusterSeven's product and market reach globally and look into expanding the firm across other major hubs in North America. Finta joins ClusterSeven from Searchspace, a leading vendor of anti-money laundering software.

HD & MP

BlackRock launches innovative Valleriite

A round up of this week's structured credit news

BlackRock launches innovative Valleriite
Arranger Merrill Lynch International and co-lead manager IXIS Corporate & Investment Bank have begun marketing Valleriite CDO I. The deal will be the first hybrid cash/synthetic arbitrage CDO in the investment-grade space to be managed by BlackRock Financial Management.

According to the S&P pre-sale report, the asset structure will combine elements of cash CDOs (the purchase of bonds and loans with the proceeds of the note issuance) and synthetic CDOs (the selling of protection through a portfolio of CDS and total return swaps). The liability structure will combine funded and unfunded elements.

The transaction is split into two sets of notes totalling €190m and US$380m fixed- and floating-rate paper. The euro-denominated notes and the dollar-denominated notes will be issued in separate, segregated series of notes, and will be backed by separate collateral and independent asset portfolios.

The transaction uses a liquidity facility – provided by Coöperatieve Centrale Raiffeisen-Boerenleenbank (trading as Rabobank International), London branch – to cover losses arising from the synthetic exposures. The collateral manager can purchase cash obligations (bonds and loans), enter into total return swaps and/or enter into CDSs (either to buy or sell protection).

Valleriite CDO I will transfer the risk of the portfolio via the issuance of funded notes, with the remainder representing an unfunded tranche notional amount for which IXIS will be compensated using a credit charge amount payable by the issuer.

The transaction comprises: US$60m and €30m triple-A rate Class S notes; US$164m and €82m triple-A Class As; US$58m and €29m unrated Class Bs; US$20m and €10m unrated Class Cs; US$18m and €9m unrated Class Ds; and US$60m and €30m unrated subordinated notes.

Lehman launches total return CDS products
Lehman Brothers has launched CDX/iTraxx Total Returns, which are designed to measure the performance (funded and unfunded) of CDX and iTraxx indices. This new product will include total returns on both basket and tranche swaps, and will have historical data available from September 2005.

The total return calculation is performed as a monthly 'buy and hold' position that, at the beginning of each month, enters into a new at-market basket or tranche swap to sell protection on the CDX or iTraxx index. The position is then held throughout the month and its daily mark-to-market is calculated.

In addition to changes in spread levels, the performance impact of any up-front payments or default events (and recovery) experienced by the CDX or iTraxx swap will be calculated, allowing investors to consistently measure and compare the total return performance of the most popular structured portfolio credit products.

"Beyond standard benchmarking, the needs for CDX/iTraxx Total Returns will likely extend to new derivative products and will also facilitate the easy comparison of performance across different levels of the tranched capital structure," says Lisa Watkinson, head of Lehman Brothers' structured credit business development.

For a given CDX/iTraxx investment-grade series Lehman Brothers will publish total returns for 5-year, 7-year and 10-year maturities for both baskets and tranches. In addition, total returns will be published in the 5-year maturity for CDX.XO, iTraxx.XO, CDX.HY, CDX.EM, CDX.HVol and iTraxx.HVol.

First European inter-dealer PCDS trade
GFI says that it has brokered the first European inter-dealer preferred CDS trade. The counterparties to the five-year US$5m Aegon PCDS transaction were Lehman Brothers and BNP Paribas.

Simon Mayes, the trader involved from Lehman Brothers, comments: "Lehman has been trading PCDS since 2005 and, having been very active with our customers both in the US and Europe, it is great to see the product being traded by other European dealers. This will doubtless lead to a further increase in volumes."

A preferred CDS incorporates a credit event of non-payment of coupon or dividend on a preferred instrument. It allows the trading in synthetic format of lower levels of companies' capital structures.

Carador's NAV slips
Listed structured credit fund Carador announced its results for April last week and its un-audited net asset value per share has dropped by 0.6% to €0.9734.

The investment manager's report says: "The decrease was due partly to negative mark to market in some CLO positions. We believe this mark down does not reflect any fundamental changes in the positions and is based on new issue CLO equity being sold at a discount. Wider liabilities should result in low prices for new CLO equity as par IRRs do not offer attractive target IRRs."

Similarly, the report adds: "Seasoned CLOs, paying full dividends and enjoying lower financing costs, should enjoy positive mark to market. We expect prices to reflect this in the near future and remain buyers of older CLOs below par. We have unsuccessfully tried to purchase some additional equity at the lower mark-to-market valuations sent by some counterparties."

Now that the fund has a 12-month pricing history, the manager also confirmed the correlation of daily returns between Carador and other traditional and alternative asset classes. These correlation levels were as follows: ML Global HY & EM Index: 2.31%; ML Global Broad Market Corporate Index: -6.69%; LPX Composite Loan Performance Index: 7.05%; and the HFRX Global Hedge Fund EUR Index: 13.69%.

SEC votes to implement CRA Reform Act
The US Securities and Exchange Commission has voted to adopt final rules to implement provisions of the Credit Rating Agency Reform Act of 2006, which was enacted on September 29 2006 (see SCI issue 27). The Act defines the term "nationally recognised statistical rating organisation" (NRSRO); provides authority for the Commission to implement registration, recordkeeping, financial reporting and oversight rules with respect to registered credit rating agencies; and directs the Commission to issue final rules no later than 270 days after its enactment (or by June 26 2007).

"The goal of this new law is to improve credit ratings quality by fostering competition, accountability and transparency in the credit rating industry," says SEC chairman Christopher Cox. "The heart of the Act calls on the Commission to replace the barriers to entry that had previously existed. The replacement is a transparent and voluntary Commission registration system that favours no particular business model."

MP

Trading ideas - c'est la VIE

Dave Klein, research analyst at Credit Derivatives Research, looks at a negative basis trade on Veolia Environnement

Basis trades are often seen as the bread and butter of many credit traders. The systemic rise and fall of credit quality often exhibits itself quite differently in each of the cash and synthetic credit markets.

The basis, or cash-CDS differential, tends to move on the back of a number of technical and fundamental factors. The most notable, over the past few months, are the impact of the structured credit technical bid (driving supply for CDS protection) and loan issuance and its hedging (driving demand for CDS protection).

Neither of these major forces impacts the bond market directly - but clearly there is an arbitrage-based relationship between the two that needs to be kept in line. In this case, we look at a negative basis trade on Veolia Environnement (VIEFP).

Based on our survival-based valuation approach, the VIEFP's 5.375's of May 2018 bond is cheap to fair value. Exhibit 1 indicates the 'price-based' term structure of VIEFP and indicates that VIEFP's shorter-dated bonds are trading rich to fair value and its longer-dated bonds are trading cheap. VIEFP is active in the credit derivative markets and the 5's of May 2018 is a liquid bond.

Exhibit 1

Based on our analysis of bond cheapness and market activity, the 5.375's of May 2018 is viable for the long bond leg of our negative basis trade. We must now look at the actual spreads to judge whether we are actually trading at a negative basis, and, maybe more importantly, can position ourselves with positive carry.

Comparing the bond z-spreads (for the active bonds) with the CDS term structure, Exhibit 2 shows that the bond is indeed trading wide of CDS. Importantly, we must note that the bond is trading wide of its interpolated maturity-equivalent CDS (ten+ year) which reflects our second basis adjustment discussed above.

Exhibit 2

The VIEFP CDS curve is fairly 'well-behaved' and has no serious inflexion points. Additionally, we see solid two-way pricing on Bloomberg's ALLQ screens for the 5's of Jan 2011.

Given that we have identified the bond as trading cheap to the CDS market (from our survival-based framework) and its z-spread as being wide of on- and off-the-run CDS levels, we must ensure that we account correctly for any premium or discount at which the bond is trading.

As seen in Exhibit 3, the basis between June 2017 CDS and the May 2018 bond has stayed steadily negative over the past year (not surprising given the mismatch in maturities). The carry and positive roll-down of the trade are the main drivers.

Exhibit 3

Over the hedge
There are two significant risks that need to be hedged in a bond-CDS negative basis trade. The first is default, which can be accounted for primitively in the adjustments we suggested previously to the CDS leg notional amount. In a perfect world, we would suggest frequent re-hedging of the CDS leg to account for changes in the bond price, but this is impractical in reality.

The approach most frequently used by practitioners is to hedge based on the price of the bond. A bond trading at US$110 would mean we buy 1.1 times the notional protection as we bought face value of the bond. A bond trading at US$90 would mean we buy 0.9 times the notional protection as we bought face value of the bond. This approximation is, somewhat surprisingly, close to optimal in terms of a single CDS static hedge.

This default risk hedge amount – based broadly on the bond price change – is important to understand in that if the bond is trading a premium (over par) then we will be over-hedged in the CDS as the bond pulls 'down' to par over time. Correspondingly, if the bond trades at a discount (under par) then we will be under-hedged as the bond pulls 'up' to par as maturity approaches.

We have seen different approaches discussed as to how to solve this dilemma but we prefer to 'keep it simple'. A practical approach is to hedge the bond's default risk with the most liquid (and closest in maturity) CDS according to the average of current and final bond price adjustments.

Our initial hedge amount should be based on a minimisation of expected loss in default– we calculate the initial hedge as (Bond Price – Recovery)/(Par – Recovery). For a bond trading at US$110 and assuming a 40% recovery, the initial hedge amount should (110-40)/(100-40)=117%. We should buy US$11.7m protection for each US$10m of face value bond we buy.

This hedge will, obviously, be over-hedged as the bond pulls to par at maturity. We would want to hold a 100% weighting in the CDS as we get close to maturity. So we simply average our initial and final hedge amounts – (117-100)/2=108.5% - to arrive at our 'simple' initial hedge. We therefore buy US$10.85m protection for each US$10m of bond face value we buy. This hedge is, on 'average' our most optimal hedge.

We have a slightly more directional perspective on this hedge. Given that we are basically hedging the impact of default on our position, if we felt strongly that the credit was unlikely to default, then we would prefer to receive more carry for more default risk. Vice versa, if we felt that fundamentals were weakening and the credit was more likely to default, then we might prefer to be slightly over-hedged.

Utilising Gimme Credit's Credit Scores – which reflect the fundamental outlook for the issuer, we adjust our single CDS hedge. If the outlook is improving then we will prefer to hedge only 100% of face value to maturity. If the outlook is deteriorating then we would hedge at our maximum current hedge amount (117% in our example). If the outlook is stable then we will choose our mid-point optimal hedge (108.5% in our example).

Once this hedge is put on, we will monitor bond price levels and default risks to ensure that our hedge is still 'close' to default neutral. This is more reasonable than adjusting frequently and paying the bid-offers, and even more importantly can often provide more carry in the short-term on a credit that is a low default risk.

In our specific trade, the bond is trading at €101.87 (best ask) and so our maximum hedge would be (101.87 – 40)/(100-40) = 103%. Given how close to 100% this number is, we choose to transact in regular sizes and hedge with 100% CDS.

Adding in mid-dated CDS to balance the hedge is possible and we are happy to discuss the use of a second CDS position with any investors who prefer to be more 'perfectly' hedged. We suggest the single CDS to bond 100% hedge and a less frequent but vigilant re-hedging program. We can generate a forward-based price projection for the bond based on the CDS curve – to show how we expect the bond price to drop (premium) or rise (discount) until maturity – which may help some investors with their hedging strategies.

The second (and often overlooked) hedge is the simple interest rate hedge on the bond. Our basis trade is positioned to benefit from any convergence of the credit risk perspectives of the cash and CDS markets and further, the bond is cheap on the basis of its CDS curve (only credit-risk based).

We therefore must ensure that our bond is hedged against interest rate movements and the impact these will have on the price of the cash instrument. Note that the CDS is exposed to interest rate movements but only in the discounting of cashflows and this interest rate sensitivity is minimal (and we ignore it in this case).

Investors could choose to asset swap the bond to minimise the interest rate sensitivity but we suggest otherwise. The asset swap has a number of shortcomings (not the least of which are lack of liquidity, premium/discount bond price errors, and default cash flow timing mismatches). We suggest buying the bond outright – benefiting from the most positive carry – and hedging interest rate risk on a portfolio basis.

However, many clients have asked for an interest rate hedge to be integrated into our negative basis trades and so we incorporate a duration-matched govvie hedge. This type of hedge, while good is not perfect since we are still exposed to changes in swap rates.

Given the cheapness of the bond and relatively wide basis, we suggest this negative basis trade (long bond and long protection) as a default-neutral way to pick up positive carry, positive roll-down and potentially realise the bond's relative value differential.

Risk analysis
This position is default-neutral for the life of the CDS. There is a slight maturity mismatch since the bond matures in a bit less than eleven years. This does not concern us as we expect convergence long before our CDS position matures.

We do, notably, understand that many investors would still prefer to understand the interest rate risks associated with this position. Exhibit 4 shows the overall and key-rate sensitivities for the bond that can be used to specifically hedge any residual interest rate risk and offers the investor some more insights into the price sensitivities of the position.

Exhibit 4

The trade has reasonable positive carry – given the current levels. This carry cushion protects the investor from any short-term mark-to-market losses.

Entering and exiting any trade in these maturities carries execution risk, but this is not a major risk with VIEFP CDS in this maturity as it is increasingly liquid. Our largest concern is execution in size in the bond markets.

Liquidity
Liquidity is a major driver of any longer-dated trade – i.e. the ability to transact effectively across the bid-offer spread in the bond and CDS markets. Our data on liquidity, created from the volume of bids, offers, and trades we see each day, provide us with significant comfort in both the ability to enter a trade in VIEFP and the bid-offer spread costs.

VIEFP shows reasonable liquidity in the ten-year maturity and bid-offer spreads are narrowing to around four basis points. Recent bids and offers for the May 2018 VIEFP bonds have shown good availability in both directions. We recommend the bond leg of the basis be worked first, looking for a price below €102.43 (our model fair-value).

Fundamentals
While this trade is technical in nature and not necessarily impacted by fundamentals, we must confirm there are no surprises in the short-term that could cause divergence between cash and CDS markets such as LBOs or major M&A activity. Generally, our negative basis trades are based on the assumption that the bond market is mis-priced relative to the CDS curve, rather than an expectation of general curve movements.

While we do not maintain a fundamental outlook on VIEFP, we note that our MFCI score is negative, with spread/rating, spread/default probability and spread/leverage factors all indicating VIEFP is trading tight to its rating group. This can be interpreted in two ways. First, that the CDS will sell off bringing VIEFP back in line with its rating cohorts. Second, that the market is anticipating a ratings upgrade bring VIEFP's rating in line with its CDS. FY 2006 figures on VIEFP exceeded expectations with revenue increasing almost 12% over the previous year and operating income increasing almost 17%.

Summary and trade recommendation
We look across the pond and find a negative basis package on VIEFP. With the VIEFP bond trading a bit cheap to fair value, we feel a default-neutral negative basis trade is an excellent opportunity to earn positive carry, roll-down and the potential profit of a return to fair value. We recommend a slightly under-hedged basis package using a single CDS static hedge to pick up 10.5 basis points of carry.

Buy €10m notional Veolia Environnement 10 Year CDS protection at 38.5bp and

Buy €10m notional (€10.2m cost) Veolia Environnement 5.375% of May 2018 bonds at a price of €101.82 (z-spread of 49bp) to gain 10.5 basis points of positive carry

Sell €10.2m notional (€10.1m cost) DBR 4.25% of July 4, 2017 at a price of €98.885 to hedge bond interest rate exposure

For more information and regular updates on this trade idea go to: www.creditresearch.com

Copyright © 2007 Credit Derivatives Research LLC. All Rights Reserved.

Note: This article is intended for general information and use and does not constitute trading advice from Structured Credit Investor (see also terms and conditions, below, section 12).

Synthetic CDOs: made to measure - part 1

A two part guide to valuing bespoke and non-standard tranches by Richard Huddart and Domenico Picone of the structured credit research team at Dresdner Kleinwort

When pricing a CDO tranche, default correlation is, of course, a key input. Whilst the expected loss of a portfolio does not depend on any correlation input, the expected loss of a tranche does. This is because a change in default correlation changes the shape of the loss distribution, defined as the probability density function of x, where x is the proportion of the portfolio lost by maturity.

When default correlation is high the probability of joint defaults is increased, and hence senior tranches price in more risk. Hence we say that senior tranches are short in correlation, as they lose in MtM terms as correlation increases. At the same time, higher correlation also increases the probability that the portfolio will experience no defaults. Given that equity (0-K%) tranche losses are capped at detachment point K, the equity tranche expected loss falls when correlation increases. Hence we say that equity tranches are long in correlation. In short, higher correlation allocates more probability to the tails of the loss distribution.

Tranche pricing - an abstract view
The following diagram summarises the key model inputs required to price a synthetic CDO tranche. Of these the correlation input is the only one not given to us from either the tranche definition or the interest rate and credit markets (we make a typical recovery rate assumption). If we want to price a tranche, we therefore need to make an assumption on correlation, or more realistically we would turn to the liquid 'standardised' tranche markets for help. By treating liquid tranche spreads as an input rather than an output, we can solve for 'implied' or 'compound correlation', a la implied volatility in option theory. 'Implied' correlations in some sense then give us a basis for pricing non-standard tranches or tranches of bespoke portfolios consistently with the liquid market.

What is base correlation and why is it useful?
The shortfalls of 'compound' correlation
The one-factor Gaussian Copula model has become the 'Black Scholes' of the synthetic CDO world in spite of its many deficiencies and a wealth of research into more sophisticated models. The most well publicised of these deficiencies is the model's inability to match (or even reasonably match) all tranche prices with the same correlation input. When backing out implied correlations from standard tranche prices, we therefore get a different value for each tranche – giving us the famous compound correlation 'smile'.

? Equity tranches tend to be priced with lower correlations as the market at this end of the capital structure is driven by protection sellers, who will demand a (relatively) large spread for taking on this risk (equity is long in correlation).
? Similarly, senior tranches are priced with higher correlations, as the standard Gaussian Copula model finds it difficult to allocate enough risk higher up the capital structure.
? There tends to be a lot of pressure on mezzanine tranches (e.g. 3-6% at 5y or 7y), as a result of bespoke CDO issuance, meaning that these tranches are priced with lower correlations.
Other issues with compound correlation as an approach to pricing include:
? Mezzanine tranche behaviour lies somewhere in between that of equity and senior tranches – and hence they are not strictly long or short in correlation. This means we may get multiple solutions for compound correlations of mezzanine tranches, or even none at all.
? The compound correlation is defined for given attachment and detachment points. Added to the fact that it does not appear to be monotonic, it is not clear at all how one could interpolate in any way to price tranches with non-standard attachments/detachments.

Base correlation – every tranche is an equity tranche
Base correlation is a variation on the compound correlation approach that provides us with a more tractable way of consistently pricing non-standard tranches. It does this by addressing the final two problems with compound correlation that we just discussed. The framework is as follows:

? Every tranche (K_L, K_U) can be expressed as the difference in two equity (or base) tranches. More specifically:
Tranche (K_L, K_U) = Tranche (0, K_U) – Tranche (0, K_L)
? To price any tranche therefore, we simply need to be able to price equity tranches.
? We know that equity tranches (0, K) are strictly long in correlation, and hence there always exists a unique correlation for every detachment point K. This is called the base correlation at detachment K.
? By applying a bootstrapping technique we can back out implied base correlations for liquid detachment points K in the standardised tranche market. For example, the 3% base correlation is just the same as the 0-3% compound correlation, whereas to calculate the 6% base correlation we need to use the 0-3% and 3-6% tranches to imply a price for the 0-6% tranche and then solve for implied correlation. It should be clear at this point that the 6% base correlation is the same as the compound correlation of the 0-6% equity tranche. Further bootstrapping allows us to calculate base correlations at higher liquid detachment points.
? The base correlation 'skew', defined as the base correlation against detachment K, is implied by the market to be increasing and approximately linear. This suggests that some type of interpolation may be appropriate for pricing tranches with non-standard attachment/detachment points.

? It should be noted that a tranche priced in the compound correlation framework relies on one correlation assumption, whereas a tranche priced in the base correlation framework relies on two (different correlations corresponding to attachment and detachment points). When calculating tranche sensitivities (or Greeks) such as spread delta, etc, different results may therefore be obtained from the two different approaches.

More a mapping than a model – but that will do for us!
To the extent that different (0,K) tranches must be priced using different base correlations, we see that the base correlation framework is essentially just a mapping or a pricing 'rule', rather than a model. For a model to be consistent, we would expect that all tranches could be reasonably priced using the same calibration (i.e. the same correlation input).

Another reason that we may consider base correlation to be only a pricing rule is that the framework cannot model skew dynamics – that is to say, the shape of the base correlation skew changes when spreads change. However, for pricing bespokes base correlation can be extremely useful because:

? Base correlation in the Gaussian Copula framework has become a sort of 'market standard', in spite of the continued development of more sophisticated models.
? In pricing a bespoke our focus is more on consistently translating liquid tranche information into bespoke tranche spreads, rather than aiming to capture skew dynamics etc.
? The upward sloping nature of the base correlation curve suggests some kind of interpolation to be a natural approach to pricing tranches with non-standard detachment points.

Some drawbacks
There has, however, been much discussion regarding the drawbacks of base correlation as an approach to pricing, and this is one of the reasons that the development of models has accelerated so much over the past couple of years. Some of these include:

? Its naïve use can lead to negative expected losses being implied on some mezzanine tranches.
? Base correlations are only unique for a given set of detachment points. That is to say, base correlation for a more senior tranche can differ if it is computed by bootstrapping from different detachments on more junior tranches.
? Base correlation shifts are not necessarily intuitive – for example, the PV of a senior tranche will fall when compound correlation rises (reflecting the fact that the tranche is short in correlation), however, there may be no corresponding rise in base correlation at the detachment point. This can be explained by the fact that the base correlation is heavily influenced by activity in the whole (0, K_U) region, whereas compound correlation is a measure implied more specifically from the (K_L, K_U) tranche alone.

We touch on some of these issues as this report progresses.

Pricing bespoke tranches

Given base correlations implied by the market, how can we use this information to go about valuing bespoke CDO tranches?

Three main issues
Essentially there are three main issues to address here. They are:

Non-standard tranches
How can we price a non standard tranche of a liquid basket (such as iTraxx or CDX) at a liquid maturity (eg the same maturity as the on-the-run 3y, 5y, 7y and 10y baskets), given that we have a set of base correlations relating to the detachment points of the standard tranches?

Non-liquid maturity
How can we price a tranche (standard or non-standard) of a liquid basket at a non-liquid maturity, given that we know the tranche prices (base correlations) relating to the liquid detachment points and maturities of the same basket?

Bespoke portfolio
If we have a bespoke portfolio, with certain characteristics in common with liquid portfolios, how can we value a tranche of this bespoke portfolio?

In this report we shall focus primarily on the first and third of these problems, which essentially restricts ourselves to pricing non-standard and bespoke tranches with liquid maturities. The first question is essentially a problem of interpolation and extrapolation of the base correlation skew(s). The third asks how we can adapt information from the liquid markets in order to account for differences between the liquid portfolio(s) and our bespoke portfolio. We will tackle these challenges in turn, culminating in a framework for valuing bespoke CDO tranches.

A note on the term-structure of correlation
We leave the second issue, namely that of dealing with tranche prices at non-liquid maturities, for a forthcoming report, as this requires an understanding of some kind of term-structure of correlation approach in order to ensure that tranches are priced consistently across maturities. We believe that such term-structure approaches deserve a more in-depth treatment, particularly as it is important to understand their interpretation, rationale and limitations. All of the approaches we consider in this report can be adapted to incorporate the term-structure of base correlation.

Interpolating and extrapolating base correlation at a given maturity

Suppose that we want to value a 4-8% tranche based on the iTraxx Europe portfolio. In order to do this we obviously need to take the market base correlation information into account, but 4% and 8% are not liquid detachment points, and so we have to do a little work in deciding what correlation assumptions to use.

Some market participants note that the base correlation skew is approximately linear, and so simple linear interpolation of the base correlation skew is common. For example, if we wanted to price a tranche with an 8% detachment referencing the 5y iTraxx pool on 19 April 2007, simple linear interpolation would suggest using a base correlation assumption of 28.5%, as is illustrated in the following chart.

It should be noted, however, that in spite of its simplicity there is no theory that suggests linear interpolation to be an appropriate approach to this problem, and its naive use can potentially lead to arbitrage opportunities arising. We now explain how this may be the case.

The cumulated expected loss profile must be concave and increasing
Let us define the market-implied (risk-neutral) cumulated expected loss at detachment K as:

Cumulated EL at K = Expected Loss of tranche (0,K) expressed as a % of portfolio notional

If the market is efficient enough to remove all arbitrage opportunities (assuming no frictions, such as transactions costs), then it must be the case that the cumulated EL as a function of detachment K is both increasing and concave, as shown in scenario A. To see why this must be the case, we discuss the possible alternatives (scenarios B to D) and see that they permit riskless profits. We then argue that simple linear interpolation of the base correlation curve may not be arbitrage free.

Let us consider first of all scenario B – an implied cumulated EL profile that has an area of convexity between detachment points K₁ and K₃. If we take detachment K₂ such that two tranches (K₁, K₂) and (K₂, K₃) are of the same width, then we find that the expected loss priced into tranche (K₂, K₃), given by EL₂₃= EL₃-EL₂, is larger than that priced into tranche (K₁, K₂) (EL₁₂= EL₂-EL₁). Given that the two tranche widths are the same, it must then be the case that the spread paid on tranche (K₂, K₃) is bigger than that of tranche (K₁, K₂). But this means that if we were to buy protection on tranche (K₁, K₂) and sell protection on the more senior (K₂, K₃), then we would earn a positive carry. This is a free-lunch, given that we cannot lose on the more senior leg until we have gained on the junior leg. A no-arbitrage argument thus tells us that it should not be possible for the implied cumulated EL profile to exhibit convexity.

Scenarios C and D are more trivial cases. Following an argument similar to that with scenario B, it should be clear that any two tranches of equal width imply the same expected loss given a linear cumulative expected loss profile (scenario C), and hence in such a case it would also be possible to sell protection on a more senior tranche whilst at the same time buying protection on the same notional of a more junior tranche. This would be a zero-cost position with a strictly non-negative payoff, thus implying an arbitrage opportunity. This tells us that not only must the expected loss profile be concave, but it must be strictly concave.

Finally, scenario D, which has a flat cumulative expected loss profile for large values of K, implies zero expected losses for more senior tranches. An efficient market would never allow protection to be purchased at zero cost, and hence scenario D is ruled out as a possibility. Our discussion here thus concludes that the cumulated market-implied expected loss profile must be increasing in detachment points K and strictly concave.

So, what is wrong with linear interpolation?
Simple interpolation of the base correlation skew across detachments does not necessarily guarantee this behaviour, and so this should be borne in mind. The following two charts show an example of this for 5y iTraxx on 31 January 2007. In the left hand chart we have constructed the cumulated EL profile implied by linearly interpolating on the base correlation skew. This in fact leads to areas of convexity, such as the one highlighted around the 6% detachment point. This is perhaps clearer in the right hand chart, where we break this down into tranchlet expected losses, as it can be seen that more risk is priced into the 6-7% tranchlet than the 5-6% tranchlet.

Cubic spline interpolation
In order to eliminate such arbitrage possibilities it may then make more sense to focus on constructing a smooth, increasing and strictly concave cumulated loss function, rather than trying to interpolate base correlation in some way. Once such a smooth function is created, base correlations can then be 'backed out' from our model at intermediate detachments.

To do this we first of all need to calculate the cumulated EL at each liquid detachment point. This can be computed by summing individual tranche risk-neutral expected losses (normalised as a proportion of the entire portfolio) and will give us five points in the iTraxx or CDX market. A shape-preserving interpolation function such as a cubic spline can then be fitted to these five points in order to give us our required interpolated cumulated EL function.

We have applied this procedure to our iTraxx example and the new interpolated cumulated expected loss profile is shown in black on the following left hand chart. This now satisfies our no-arbitrage conditions and this effect is much clearer when we look at tranchlet expected losses once more – it is now the case that for every pair of tranchlets considered, the more senior has a lower expected loss.

Extrapolation
The question of what correlation assumption should be used to price a (0, K) tranche where detachment K is less than 3% is a somewhat more challenging problem. In this region it is not possible to interpolate expected losses or base correlations, as no liquid detachment point below 3% is traded. There have been attempts to grow a market for trading 'tranchlets', or thinner slices generally situated within the 0-3% region, however progress has been relatively slow – and the main reason for this has been our precise issue of how to price them. Tranchlet prices can only therefore be seen as a start to gathering pricing information inside this equity tranche region.

If one notices that the cumulated EL on tranche (0, 0), i.e. a tranche with zero width, must equal zero, then it could be suggested that we could continue to interpolate expected losses between this point and the 0-3% using the method just discussed. This would, however, be an extremely naïve approach, especially given that the equity tranche is so exposed to idiosyncratic risk (or independent defaults), which is not so much a factor for more senior tranches. As a result, the market will tend to price in more risk to the 0-1% tranchlet (being the most exposed to idiosyncratic risks as the first-loss tranchlet) than would be implied by such cubic spline interpolation, with the spline being unable to differentiate at all between idiosyncratic and systemic risk.

Of course there are certain things we do know about the cumulated EL profile in the 0-3% region. For example, we know that it must be smooth, increasing and strictly concave in detachment K. In an effort to move on from base correlation and eliminate the problem of the correlation smile, several other models, such as the random factor loadings model (Andersen and Sidenius, 2004), the Composite Basket model (Tavares et al, 2004), the Normal Inverse Gaussian model (Guegan and Houdain, 2005 and Kalemanova et al, 2005) and the Implied Copula (Hull and White, 2006), are now being used. Where as the base correlation approach (in, say, the Gaussian Copula framework) is a one-parameter model that requires a different calibration of the parameter to price each tranche (hence the correlation skew), these alternative models are able to provide a reasonable fit to different tranche prices with the same single calibration, however at the expense of more parameters. Where such a model is calibrated to the iTraxx market, it may be reasonable to use this calibration to give an indication of the loss density within the 0-3% region¹.

In any case we note that all of these approaches are to a certain extent 'ad-hoc', and so no one in particular should be taken as a hard and fast rule for valuing (0, K) tranches that detach in the 0-3% region.

Notes
1. The equity tranche price tends to be the most problematic to match for a model which is calibrated to price all parts of the capital structure with the same set of parameters, and so as a starting point it should be assumed that the model can produce a 0-3% tranche spread in line with the market.

© 2007 Dresdner Kleinwort. All Rights Reserved. This Research Note was first published by Dresdner Kleinwort on 14 May 2007.