Total Return Swaps, Synthetic CDOs, and Credit Indexes

The second half of Chapter 3 picks up where the credit default swap discussion ended and covers total return swaps, CDS pricing, synthetic CDO structure, equity TRORS, information asymmetry, pay-as-you-go templates, and the credit indexes that eventually let people bet against the subprime market.

CDS pricing: what goes into the number

Before moving to TRORS, Tavakoli lays out what should go into pricing a credit default swap. The conceptual starting point: credit default protection should trade near the spread of the reference security over a risk-free rate. If a bond yields 100 bps over Treasuries, roughly 100 bps of that is credit risk, and that’s the rough anchor for CDS pricing.

But CDSs tend to trade benchmarked against the asset swap market (spread over LIBOR), not the Treasury spread. Most banks benchmark funding costs to LIBOR, so this is the natural frame for credit derivatives.

The key inputs for a proper CDS valuation include:

• Credit quality of the reference obligor
• Credit quality of the protection seller (the CDS provider can itself default)
• Correlation between the obligor and the protection seller
• Probability of default and its volatility
• Joint default probability if obligor and protection seller are correlated
• Recovery rate in default
• Recovery rate of the protection seller (replacement cost of the CDS)
• Leverage of the structure
• Maturity
• Settlement provisions
• Supply and demand dynamics
• Any economic research that changes the credit view
• Urgency of needing the hedge
• Regulatory capital weighting of the protection seller

That last one matters in bank-to-bank transactions: the Basel rules give more regulatory capital relief when the protection seller is OECD bank-rated than when it’s an unrated entity.

Most models start with a binomial tree with these inputs and then adjust for soft factors including supply and demand. The math is not the hard part. Getting reliable inputs is.

Sovereign debt default language

Corporate CDS language doesn’t translate directly to sovereign credit. Tavakoli covers both.

For sovereign debt, a reasonable way to define a credit event mirrors the language in the bond prospectus itself. Typical conditions for triggering a sovereign CDS:

• Failure to make payment on public external debt in excess of a preset threshold (e.g., $25 million)
• Acceleration of debt in excess of a preset amount
• Declaration of a moratorium on payment of principal or interest
• The sovereign denying its obligations
• Action materially prejudicial to bondholder interests

Custom language can go further: you can include war or coup d’état as credit events if the specific risk you’re hedging requires it.

The Conseco controversy (2000) established that the same issues apply for corporate debt. Banks restructured Conseco loans – extending maturity by three months, adding increased coupon and covenants favorable to lenders. The credit rating only dropped to B1, not to default-level. Moody’s didn’t consider it a “distressed exchange” default. The loan price actually increased.

But CDS protection buyers delivered deeply discounted long-dated bonds (not the robustly priced loans) and received the difference between par and the discounted price. Protection sellers felt burned – they argued the banks, as loan underwriters, had inside knowledge of the Conseco situation and exploited the cheapest-to-deliver option by delivering bonds that nobody else would consider appropriate settlement.

The outcome: modified restructuring language entered the U.S. market to limit deliverable maturities after restructuring. Separately, language excluding restructuring as a credit event became common for index-based and purely speculative contracts. But banks hedging loans still need restructuring as a credit event to get regulatory capital relief – so the language tension remains.

Moody’s view: moral hazard risk exists when the sponsoring bank determines both when a loss event has occurred and what the deliverable will be. This risk is not something the rating agency can capture in its CDO rating models.

Synthetic CDOs: the structure

Before credit derivatives, assembling a CDO meant physically acquiring bonds. You needed to warehouse them, fund the position, hedge the market risk during accumulation, and deal with varied maturities.

CDSs changed this. A synthetic CDO references a portfolio of obligations through credit default swaps. No physical asset needs to change hands. You can create a specific maturity deal referencing whatever credits you want, regardless of what bonds actually exist in the market.

A generic static (nonmanaged) synthetic arbitrage CDO works like this:

1. The CDO SPE sells credit default protection to the arranging bank on a reference portfolio (typically investment-grade corporate names)
2. The SPE invests the proceeds in highly rated collateral (often AAA securities)
3. The SPE issues tranched notes to investors: super senior, senior, mezzanine, equity
4. The collateral earns income; the CDS premiums provide additional income; investors receive their tranche spreads
5. If credit events occur in the reference portfolio, losses are absorbed from the equity tranche upward

The super senior tranche – covering the very top of the capital structure – is what made synthetic CDOs so attractive. This tranche is so remote from losses that it can be funded extremely cheaply, or retained by the arranging bank with minimal capital. The premium from the CDSs on the reference portfolio, combined with the income from the collateral, made the economics work even with cheap senior funding.

In the mid-2000s, European and Asian synthetic CDOs made up 80-90% of total CDO issuance in those markets (versus about 25% in the U.S., where high-yield bond supply was higher and cash deals were more common).

Total rate of return swaps (TRORS)

A TRORS (also called TRS – total return swap) is primarily a financing tool. The distinction from credit default swaps: a TRORS hedges both market risk and credit risk, not just credit default risk.

The setup: the total return payer owns the reference asset and pays the entire total return (coupon income, price appreciation, any other economic benefits) to the total return receiver. In exchange, the receiver pays a floating-rate fee – typically a spread to LIBOR or Euribor – to the payer.

The receiver gets synthetic long exposure with leverage. They don’t need to fund the purchase of the asset. They just post collateral (often a fraction of the asset value) and receive the economics as if they owned it.

In a TRORS:

• The payer retains legal ownership
• The payer has hedged market risk and credit risk until maturity
• The receiver has leveraged exposure without balance sheet impact
• If the reference asset defaults, the receiver pays the payer the loss (or takes physical delivery at the original price)

Timing can differ from the cash flows of the underlying asset. If the asset pays monthly but the TRORS settles quarterly, coupons must be reinvested – introducing reinvestment risk. This is why Tavakoli calls them “total rate of return swaps” rather than just “total return swaps” – to highlight this potential cash flow mismatch.

TRORS versus repos

Both TRORS and repurchase agreements provide leverage. The difference:

Repo (repurchase agreement): The asset owner sells the asset and agrees to repurchase it at a preset price on a preset date. The price difference implies a fixed rate – the repo rate. The seller is obligated to repurchase at maturity.

TRORS: There’s no pre-agreed repurchase price. The receiver is obligated to exchange payments based on the market value at maturity, not a fixed price. The receiver is exposed to the price risk of the reference asset. If the reference asset has a much longer maturity than the TRORS, this price risk at TRORS maturity can be substantial.

TRORS receivers who accept maturity mismatches do so because they want exposure to a reference asset for a shorter period than is available in the market, and they accept the price risk in exchange for flexibility.

TRORS pricing

There’s no standard pricing method. Tavakoli describes the range:

Lower bound: The rate the bank would charge for an unfunded revolver of the same maturity to this counterparty. This is a standby loan with lower expected loss than the TRORS reference obligation, so it’s genuinely a floor, not a fair price.

Upper bound: Whatever the market will bear.

In practice, generic TRORS price at a level that earns the bank a positive spread to its funding costs with adequate return on capital, while the receiving counterparty calculates whether the leveraged return is worth the risk. Comparison points: floating-rate assets of similar credit, asset swaps, and the return on unfunded equity transactions.

For AAA-rated MBSs, TRORS have been done at LIBOR plus 10 to 25 bps at three-month to five-year maturities. BBB assets have financed at LIBOR plus 25 to 50 bps for under three years. These are rough benchmarks – conditions change and supply/demand matters.

By mid-2007, prime brokers who had been aggressively competing for hedge fund TRORS business started reconsidering the generous terms they’d offered after several overleveraged hedge funds imploded.

Equity TRORS: disguised loans

This is one of the sharper examples in the book. Some banks enter into equity total return swaps with corporations that want cash.

How it works: the bank buys the corporation’s equity and pays the corporation the total return on that equity. In exchange, the corporation pays LIBOR plus a spread (often just 50 bps). The cash shows up on the corporation’s balance sheet as a capital injection (equity sale proceeds). The TRORS is off balance sheet.

Why would a BBB-rated corporation do this? Because borrowing this way costs LIBOR plus 50 bps versus LIBOR plus 250+ bps in the loan market. Huge savings.

The problem: if the stock price falls, the corporation has to post more shares as collateral for the mark-to-market loss. The bank receives more and more stock. If the stock price goes to zero – which happens – the bank is holding worthless collateral on what was effectively a loan.

Meanwhile, the original accounting treatment is questionable. The cash was booked as equity investment proceeds, not as a loan. The TRORS is a financing tool. It looks like Arthur Andersen’s kind of accounting.

Tavakoli draws an ironic parallel: when hedge funds lend against equity collateral using so-called “toxic convertibles,” they’re vilified for potentially causing stock dilution. When banks do the exact same thing via equity TRORS, they describe it as performing a public service.

The incentive structure is also worth noting: TRORS trading desk heads earn bonuses tied to fee income. They have every incentive to do as many deals as possible and recognize accrual income. The risk (that stock prices fall, collateral becomes worthless) is someone else’s problem later.

Information asymmetry and moral hazard in loan markets

TRORS and CDSs on loans create a specific problem: the lender has inside information.

When a bank originates a loan and then enters into a CDS or TRORS referencing that loan, the bank retains voting rights as the legal owner. The bank may know things about the borrower that aren’t public. The protection seller counterparty doesn’t.

There’s usually a gentlemen’s agreement that the lender will vote according to the protection seller’s wishes if contacted. But if the bank can’t reach the counterparty in time, it votes with the majority of other lenders – who may also have better information.

More broadly: if an underwriter is accumulating assets in a warehouse and becomes aware of material changes in those assets (fraud, deterioration, changed underwriting standards), that information must be disclosed. For private placements with risk disclosures that include caveats about novel asset classes or departed from historical underwriting practices, Tavakoli’s clear view: sophisticated investors should not buy until they’ve done adequate due diligence on default probability and recovery.

This warning was directly applicable to subprime-backed CDOs in 2005-2007.

Pay-as-you-go (PAUG): CDSs on ABS

Standard CDS documentation was designed for corporate bonds. Asset-backed securities have different cash flow structures that require different contract language.

ISDA introduced the PAUG (pay-as-you-go) template in 2005 for RMBS and CMBS, and in June 2006 for CDOs. The idea: replicate the cash flow profile of the cash bond using a CDS contract.

The ABS-specific issues:

• RMBS bonds are subject to “available funds caps” (AFC) – interest payments are capped at the weighted-average mortgage rate net of expenses
• CMBS bonds have “WAC shortfalls” with similar capping
• CDO bonds have “pay-in-kind” (PIK) shortfalls when interest is not paid in cash

Unlike standard CDSs, PAUG protection sellers make floating payments when these shortfalls occur. Protection buyers may reimburse these if the shortfall is later recovered. Protection buyers can choose whether to call a credit event or a floating amount event.

PAUG credit events:

• Failure to pay principal
• Write-down
• Distressed rating downgrade (CCC or below)
• Failure to pay interest (for CDO reference obligations only)

PAUG floating amount events:

• Interest shortfalls
• Principal shortfalls
• Write-down amounts

Settlement is complicated by illiquid secondary markets for structured finance securities. Physical delivery of all or part of the notional is an option if a credit event occurs; cash settlement for illiquid products is notoriously difficult to price.

Tavakoli’s concerns about PAUG: restructuring language for mortgages is problematic (hard to distinguish genuine restructurings from delaying tactics to avoid recognizing losses), the contract introduces substantial basis risk, and floating amount manipulation was becoming an issue in the subprime market.

Credit indexes

By 2006, a set of credit indexes linked to CDSs had emerged, tracked primarily by Markit Partners.

The ABX.HE Index (asset-backed home equity index) launched in early 2006, linked to rated slices of subprime residential mortgage loans. The ABX.HE BBB 06-2 – linked to 2006-vintage subprime mortgages – became the key instrument for people wanting to bet against or hedge the subprime market.

If you want to buy protection (bet on decline), you short the index. As the index price falls, reflecting declining value of the underlying subprime bonds, the short position gains. Hedge funds used short ABX positions for what Tavakoli calls “wildly profitable speculation” – profitable because the index correctly anticipated massive losses in subprime-backed bonds.

The TABX launched February 14, 2007, composed of standard tranches on the ABX.HE Index.

Corporate bond indexes also developed as reference portfolios for bespoke CDO tranches and other products. These indexes standardized previously custom deals and added liquidity, but also made it easier to create large synthetic exposures quickly – with consequences for how much hidden risk accumulated in the system.

Chapter 3 closes the toolkit section. Next, Chapter 4 covers the history of how the CDO market actually evolved – from a few billion in the late 1980s to the peak of the credit cycle in 2006-2007.

Previous: Credit Derivatives and Default Swaps: The Basics

Next: CDOs and How They Shaped Global Capital Markets