Early CDO Technology: How It All Started

Book: Structured Finance and Collateralized Debt Obligations | Author: Janet M. Tavakoli | Publisher: John Wiley & Sons (2008) | ISBN: 978-0-470-44344-6

Chapter 6 is the technical foundation chapter. This is where Tavakoli explains how CDOs are actually built - the legal structures, the credit protections, the taxonomy, and the historical roots of the whole industry.

It’s dense in places. But understanding these pieces matters, because the failures described later in the book trace back to specific structural choices and how they were abused.

True Sale Structures: The Original CDO Architecture

Up to the end of the 1990s, all CDOs used special purpose entities (SPEs), also called special purpose vehicles (SPVs). The SPE purchased a portfolio of assets from a bank’s balance sheet or trading books and then issued tranches of debt and equity backed by those assets.

These are called true sale structures because assets are actually sold to the SPE.

The SPE is bankruptcy-remote. That’s the key legal feature. It’s deliberately isolated from the credit risk of the bank that arranged the deal. The arranger can earn servicing fees, administration fees, and hedging fees, but otherwise has no claim on the SPE’s cash flows. If the bank fails, creditors can’t reach the SPE’s assets.

This isolation mattered enormously. It meant investors in CDO tranches were exposed to the credit risk of the underlying portfolio, not the credit risk of the sponsoring bank.

Later, synthetic structures changed this. In a synthetic CDO, no assets are actually sold. Credit risk is transferred via credit default swaps. Some synthetic deals use an SPE to issue notes linked to the credit risk, but others don’t use an SPE at all. The same portfolio can back much more risk in synthetic form, because you’re not limited by what the bank actually owns.

Credit Enhancement: The Many Layers of Protection

Credit enhancement is how CDO structures make lower-risk tranches out of higher-risk collateral. Multiple types are usually stacked into one transaction.

The main tools:

• Overcollateralization: More collateral than strictly needed to meet obligations. If collateral declines in value, the extra buffer absorbs losses before touching investor tranches.
• Subordination: Junior investors (equity tranche holders) take losses first. They absorb hits so senior investors don’t have to.
• Reserve accounts: Excess cash flow gets trapped here and held as a buffer.
• Cash flow triggers: If the deal deteriorates past preset thresholds, cash flow gets redirected - usually away from junior investors toward senior ones.
• Credit derivatives: Credit default swaps can be used to protect specific tranches.
• Credit wraps: Monoline insurers guarantee payments. More on this below.

The combination matters. A deal might have overcollateralization of 30 percent, a 3 percent reserve account, and subordination below each rated tranche. Each layer protects the one above it.

Letters of credit were common early on but fell out of use because so many banks were downgraded, making the LOC less valuable as protection.

Monoline Insurance: The AAA Backstop

Monoline insurance companies became a major feature of structured finance. They provide financial guarantees, also called credit wraps, which guarantee that investors receive their scheduled principal and interest payments even if the underlying collateral fails.

Under New York law, only monolines can provide these financial guarantees. At the time of the book’s writing, seven monolines were rated AAA, though several were already in danger of losing that rating. MBIA, Ambac, and FSA dominated the market.

A wrapped tranche can only carry the rating of the monoline providing the wrap. To achieve a triple-A wrap, a general rule was that credit enhancement had to equal five times the expected loss level. To keep a triple-A rating, Moody’s required a monoline to hold capital covering 130 percent of base-case losses and 100 percent of stress-case losses.

Wrapped tranches that would otherwise have been downgraded maintained their triple-A ratings purely by virtue of monoline guarantees. This worked until the monolines themselves got into trouble - which is a story Tavakoli covers in Chapter 17.

The difference between a financial guarantee and a credit derivative is important. A credit derivative buyer doesn’t need to actually own the underlying security or suffer a loss to collect. A financial guarantee requires an actual payment failure. This distinction became very significant in disputes involving Enron and JPMorgan.

CDO Classification: The Full Menu

Any future payment stream can be securitized. Tavakoli lists a wide menu: consumer receivables (auto loans, credit cards, home equity), commercial and industrial loans, mortgage-backed securities, equipment leases, aircraft leases, utility stranded costs, delinquent tax liens, emerging market debt, hedge fund of funds, private equity, nonperforming loans.

Even Formula One broadcasting rights and cable subscriptions have been securitized.

The main named types:

Collateralized Loan Obligation (CLO): Backed exclusively by loans. The first rated CLO backed by US bank loans came to market in 1990.

Collateralized Bond Obligation (CBO): Backed by a portfolio of bonds (secured or unsecured, senior or junior). The first rated CBO backed by high-yield bonds was brought to market in 1988.

Collateralized Mortgage Obligation (CMO): Backed by mortgage-backed securities. Ironically, CMOs are often excluded from CDO statistics even though they were the original model for the whole structure.

CDOs are further classified as:

• Cash or synthetic: Cash CDOs hold actual assets. Synthetic CDOs use credit default swaps.
• Arbitrage or balance-sheet: Arbitrage deals capture the spread between higher-yielding collateral and lower-cost CDO liabilities. Balance-sheet deals are primarily for regulatory capital relief.
• Market value or cash flow: The two main types of arbitrage CDO.

Market Value CDOs

Market value CDOs make up roughly 10 to 15 percent of the arbitrage CDO market.

In these deals, the manager actively trades the portfolio. Liabilities are paid from trading gains and interest income. Overcollateralization ratios - the ratio of asset market value to face value of liabilities - are monitored continuously. If ratios drop below required minimums, assets must be sold or replaced with higher-quality liquid assets. This is a trigger event.

Market value CDOs work best when the collateral pool consists of defaulted or distressed bonds that don’t generate predictable cash flows but have significant upside from active trading. A manager with a strong trading track record is essential.

The rating agencies monitor coverage ratios. Performance depends entirely on the manager’s ability.

Cash Flow CDOs

Cash flow CDOs make up 85 to 90 percent of the arbitrage market and were growing faster.

These work differently. The structure passes principal and interest from the underlying collateral pool directly through to investors. For the deal to work, cash flows from collateral must cover all obligations: investor coupons, principal repayment, fees. A properly structured cash flow CDO means investors only lose money if there are actual defaults in the collateral pool.

Cash flow managers are chosen for credit expertise, not trading expertise. Their role is to select the initial portfolio, make purchases during the reinvestment period, and maximize recovery if a credit defaults.

Some synthetic static CDOs dispense with the manager entirely because there’s a fixed pool with no trading. The CDS market lets you define maturities at inception, so there’s no reinvestment period to manage.

The Origins of US Securitization: Ginnie, Fannie, and Freddie

Most securitization technology came from the US mortgage market. Understanding its origins reveals structural patterns that get repeated across asset classes, often badly.

Three entities created the foundation: GNMA (Ginnie Mae), FNMA (Fannie Mae), and FHLMC (Freddie Mac). They set guidelines, pooled residential mortgages, and issued pass-through securities backed by those pools.

GNMA is a government agency. Its securities are backed by the full faith and credit of the US government. FNMA and FHLMC are privately chartered and government-sponsored, not formally guaranteed. The US market treated them as essentially risk-free. European investors were initially skeptical - they didn’t share the assumption that the government would never let Fannie or Freddie fail.

The cash flow risk in these securities came from the embedded call option: homeowners can prepay their mortgages. This sounds abstract, but it was financially brutal for investors who didn’t understand it.

Travelers Insurance bought GNMAs with double-digit coupons at high premiums in the mid-1980s. They got high current income - temporarily. When interest rates fell sharply, homeowners rushed to refinance. Prepayments sped up massively. Those premium securities began prepaying at par, destroying the coupon stream. Travelers took mark-to-market losses and lost the cash flow at the same time. The lower rates meant reinvestment at lower yields too. A late hedge using long-dated Treasury zeros helped partially, but the losses were substantial.

Collateralized Mortgage Obligations: The CDO Template

Investor demand for more structural choices led to CMOs. These carved up the mortgage cash flows into multiple tranches with different prepayment exposures.

The first CMOs had four tranches in a sequential pay structure. Tranche A received principal first. Once fully paid down, Tranche B received principal. Then C. Then the Z tranche, which didn’t receive any interest payments during this time - instead, the accrued interest helped pay down the earlier tranches. Investors in the Z tranche liked this because it eliminated reinvestment risk.

From this simple structure, Wall Street created floating-rate tranches, inverse floating-rate tranches, interest-only (IO) tranches, principal-only (PO) tranches, and PAC (Planned Amortization Class) bonds that tried to give investors predictable cash flows within a prepayment range.

Floaters needed caps. Every floater had to have an inverse floater. The inverse floater coupon moved in the opposite direction to rates, with a leverage factor. An inverse floater with three-times leverage meant that for every 1 percent drop in LIBOR, the coupon went up 3 percent.

IO tranches had no principal at all. Their value depended entirely on how long the underlying mortgages stayed outstanding. If prepayments sped up - because rates fell and homeowners refinanced - IO values collapsed. If prepayments slowed, IO values rose.

Tavakoli tells the story of a Morgan Stanley trader in 1993 who believed she could predict homeowner prepayments using the firm’s model. Tavakoli warned her the position was too large and the risk too concentrated. The trader dismissed the concern: she knew how to predict the unknowable unknown.

Interest rates fell. Homeowners refinanced at unprecedented rates. The price of those volatile securities plummeted. Before year-end, Morgan Stanley had lost significant money in that book. The trader was fired along with her boss.

The lesson Tavakoli draws: there is nothing wrong with taking risk. What investment banks cannot forgive is fooling yourself and your managers about how much risk you’re taking.

The Pattern Repeats

The Hammer story closes the chapter. A Japanese trading company investor saw Tavakoli’s two deals: a 14 percent yield option and a 15.25 percent yield option. The higher-yielding one was path-dependent and extremely sensitive to prepayments. If rates fell, the cash flows vaporized.

Tavakoli explained the risk clearly. The Hammer dismissed it. His Tokyo office thought rates were going up. He took the riskier deal.

A few years later, he called. His mortgage-backed holdings had sustained years of losses. By fiscal year-end March 1995, losses were $200 million. He lost his lifetime job.

The structural lesson that Tavakoli makes explicit: you can carve cash flows to meet any investor desire. But the patterns of risk appetite, overconfidence in models, and concentration in untested asset classes recur with every new instrument. The chapter ends with a direct warning that this same pattern will emerge again, with different instruments, in the stories that follow.

Previous: CDO Risk and Valuation: When Models Meet Reality

Next: Commercial Financial Services: An Early Warning Nobody Heard