CDOs and How They Shaped Global Capital Markets
Chapter 4 is where Tavakoli steps back from mechanics and tells the story of how the CDO market grew. The chapter is titled “CDOs and the Global Capital Markets” and it covers roughly 20 years of history – from the junk bond era of the late 1980s through the explosive synthetic CDO growth of the 2000s to the beginning of the unraveling in 2007.
Understanding this history explains a lot about why the financial crisis happened the way it did.
Credit derivatives: from billions to trillions
The scale of growth is the first thing to absorb.
In 1996, the global credit default swap market was $100-200 billion in size. By 2002, Morgan Stanley estimated it at $2.4 trillion. By 2004, London had captured more than 50% of global CDS trading, and ISDA estimated the market at $8.42 trillion. One year later: $17.3 trillion. In 2006: $34.5 trillion. By year-end 2007: $62.2 trillion. Tavakoli notes that $80+ trillion was projected for end of 2008.
These are notional amounts, not risk-weighted figures. The actual economic exposure is smaller. But the scale still illustrates what happened: a market that barely existed in 1996 became the dominant feature of global credit markets within a decade.
The CDO market grew alongside. Credit derivatives drove CDO expansion because they made synthetic securitization possible – and synthetic securitization was cheaper and more flexible than cash securitization.
The old paradigm: junk bonds and the S&L crisis
The CDO market actually started in the late 1980s, before credit derivatives existed. The precursor was the collateralized bond obligation (CBO) – a CDO backed by high-yield bonds.
The context: junk bonds were yielding 13-20% in the late 1980s. Meanwhile, U.S. savings and loans had figured out they had an implied government guarantee. If they bought highly leveraged residuals from CMO transactions and high-yield bonds, and things went well, they kept the profits. If things went badly, they could essentially hand the S&L back to the government. This moral hazard contributed to the S&L crisis.
The Resolution Trust Corporation was charged with liquidating the failed S&L portfolios. The fall of Drexel Burnham Lambert (Drexel was the primary junk bond underwriter) added further cheap supply to the market. Insurance capital regulations tightened, reducing demand. High-yield bonds were cheap and plentiful.
Arbitrage CBOs were the solution for investors who wanted diversified exposure to this cheap supply. These were actively managed cash deals with leveraged market value structures.
Then supply dried up. High-yield prices recovered through the early 1990s, yields fell, and deal volume declined. The simpler cash flow CDO structure reignited investor interest after 1995. That was the old paradigm.
The paradigm shift: synthetics and the super senior tranche
By the late 1990s, the market had fundamentally changed.
The European market illustrates why. The U.S. bond market was approaching $600 billion by 1999. The European bond market was only about $35 billion – too small to support cash CDOs. The multicurrency nature of Europe added further complexity. The conventional wisdom was that Europe would never be a significant CDO market.
Credit derivatives changed this. Instead of physically accumulating a portfolio of bonds, a synthetic CDO could reference any credits through CDS contracts. It didn’t matter if the bonds were actually available in European markets. You could build a CDO referencing U.S. investment-grade corporate names from Frankfurt or Paris just as easily as from New York.
The key innovation: the super senior tranche.
In a cash CDO, the senior tranche had to be funded – investors had to buy it. The senior tranche of a cash deal typically required a spread of LIBOR plus 30-40 bps.
In a synthetic CDO, the risk above the senior (mezzo and equity) tranches is the super senior. This tranche is so remote from losses – protected by all the layers below it – that it can be retained by the arranging bank with minimal capital, or provided to monoline insurance companies for a tiny premium (single-digit basis points). The bank keeps the super senior, earns a small fee, and uses the freed-up capital for other activity.
This was the “arbitrage” that made synthetic CDOs so attractive. The super senior tranche made up 80-90% of most synthetic CDOs. Funding it cheaply made the equity and mezzanine tranches look much more profitable by comparison.
The market in numbers: 1995 to 2006
Before 1995, virtually all CDOs were cash deals. By 1999, the CBO/CLO market was around $120 billion – a roughly 50x increase in just four years, primarily driven by synthetic CDOs.
The introduction of the euro in 1999 created a wider single-currency asset base in Europe and a broader investor base for euro-denominated CDOs. Issuance reached around $200 billion in 2001, or $325 billion counting single-tranche CDOs (STCDOs). In 2002: $250 billion, or $600 billion with STCDOs.
By 2002, synthetics made up more than 75% of the CDO market. In Europe and Asia, synthetics were 80-90% of total CDO issuance. In the U.S., only about 25% were synthetics – because the U.S. had more high-yield bonds and leveraged loan supply to fuel cash deals.
Defaults also spiked in 2001 and 2002: over $180 billion in defaults that year. High-yield was hardest hit, but investment-grade wasn’t immune. Investment-grade bond defaults in 2001-2002 exceeded the cumulative total of the previous 20 years. CDO downgrades grew 318% in 2002, with 74% in the high-yield sector.
This created a problem for cash CDOs backed by high-yield collateral, but actually reinforced the synthetic CDO market. Synthetics could be built using exclusively investment-grade credits, which hadn’t defaulted at the same rate. The synthetic CDO arbitrage remained viable.
Until it wasn’t. Deals stuffed with “investment-grade” names like Enron and WorldCom – both of which defaulted in 2001-2002 – demonstrated that ratings weren’t the same as safety, and that concentrated exposure to names that happened to carry investment-grade ratings could wipe out equity and mezzanine tranches.
Single-tranche CDOs and hidden bank risk
By the mid-2000s, a significant portion of synthetic CDO activity was in single-tranche CDOs (STCDOs).
In an STCDO, the bank arranges the deal but only sells one tranche – typically the mezzanine – and doesn’t place any other tranches with outside investors. The bank retains everything else: the super senior, the other mezzanine tranches, the equity. It then delta-hedges the retained positions by selling credit protection on a “vertical slice” of the portfolio.
Here’s the problem: the delta hedge is based on estimated correlations and credit quality. The correlation models are debatable. The bank is holding substantial residual risk – equity risk, senior risk, correlation risk – without fully recognizing it.
Tavakoli makes a point that bank managers consistently missed: reporting STCDO risk only on the basis of the single sold tranche dramatically understates the actual exposure. The correct frame is the full notional amount of the underlying portfolio, because that’s what the bank’s retained positions are referenced to.
“Risk is not measured in billions or trillions of notional amounts. Risk is measured as value at risk.” The sold mezzanine tranche is one horizontal slice of a much larger portfolio. The unsold tranches represent the rest. Reporting only the mezzanine tranche misrepresents the exposure.
By 2006, when credit spreads were tight, the return on the implied equity risk retained by arrangers was less than 10%. When spreads had been very wide, that same equity could return 60%+. The deterioration of returns wasn’t obvious from looking at the mezzanine tranche alone.
Who bought the super senior?
Most of the synthetic CDO volume wasn’t going to traditional CDO investors. Where did it go?
- Arranging banks retained much of the super senior, often without proper recognition of the risk
- Monoline insurance companies provided guarantees on super senior tranches for tiny premiums – premiums that didn’t adequately compensate for the remote but real tail risk
- Structured investment vehicles (SIVs) held junior super senior and mezzanine AAA tranches, funded with short-term commercial paper
- Constant proportion portfolio insurance (CPPI) and constant proportion debt obligation (CPDO) products absorbed further structured product exposure
- Hedge funds became major buyers of equity tranches, often with significant leverage
As a result, the actual risk of the CDO market was distributed in ways that were difficult to track. It was concentrated in institutions – banks, monolines, SIVs – that either didn’t recognize the risk or couldn’t adequately hedge it. When the underlying collateral (particularly in mortgage-backed CDOs) started showing losses, the cascade was severe precisely because so much risk was hidden.
The three categories of synthetic CDOs
By the mid-2000s, synthetic CDOs fell into three main categories:
Balance-sheet CDOs: Banks use these primarily to reduce regulatory capital. The bank owns a portfolio of loans or bonds and buys CDS protection on the portfolio through a CDO. The risk transfers; the bank’s regulatory capital requirement falls. In the mid-2000s, some arbitrage-driven versions used mezzanine tranches from multisector CDOs as the reference collateral.
Static synthetic arbitrage CDOs: A fixed reference portfolio is selected at inception. No active management. The deal runs to maturity with whatever credits were in the original pool.
Managed synthetic arbitrage CDOs: An active CDO manager can substitute credits in the reference portfolio, within preset guidelines. This introduces the CDO manager’s judgment – for better or worse.
What the rating agencies missed
Moody’s data showed CDO issuance growing at about 27% annually from $14 billion in 1996 to $158 billion in 2006, for total 2006 issuance of $329 billion. But this data only captured rated tranches.
Unrated STCDOs, bespoke tranches based on CDS indexes, privately placed deals – none of this was captured. Tavakoli estimates market size in full notional terms was dramatically larger than rating agency data suggested.
Her estimates of outstanding market size, accounting for full notional amounts through 2006, were substantially higher than the rated-tranche-only figures. The subsequent write-downs announced in 2007-2008 validated the larger estimate as the more meaningful measure of real exposure.
The feedback loop
Hedge funds as CDO equity investors created a specific feedback loop. They could take equity positions with very little capital down – sometimes as little as 5-15% of the notional amount, with leverage provided by the investment bank.
When the bank is doing this kind of transaction with an opaque hedge fund or thinly capitalized CDO manager, the only thing the bank should count on is the initial collateral. If the hedge fund blows up, the bank is left holding a leveraged position in CDO equity – the most subordinated, first-loss tranche – often without adequate capital to absorb the losses.
Synthetics made it possible to take leveraged exposure throughout the capital structure – equity, mezzanine, senior – with minimal upfront capital. This contributed directly to the global credit crunch of 2007-2008.
What changed everything
Tavakoli’s chapter is explicit about the inflection point: synthetic CDOs were the dominant driver of CDO growth, and the creation of the super senior tranche was the mechanical reason they were so attractive.
But the super senior’s appeal was based on two assumptions that broke down:
- The underlying reference portfolio would remain well-diversified (not concentrated in correlated assets that default together)
- Recovery rates and default probabilities would stay within historical norms
Both assumptions failed when the reference portfolios shifted from diversified investment-grade corporates to concentrated exposure to mortgage-backed securities with fraudulent origination and impossible underwriting assumptions.
The market was well-designed for the risks it had historically encountered. It was catastrophically poorly designed for the risks that accumulated in 2005-2007.
Next: Chapter 5 gets into CDO risk and valuation in more detail – what the models are actually measuring, why they fail, and what proper due diligence looks like.
Previous: Total Return Swaps, Synthetic CDOs, and Credit Indexes