Tales of leverage

Regulators are proposing the introduction of leverage caps on banks. However, Satyajit Das argues leverage achieved through derivatives does not always show up in traditional leverage measurements

There is widespread recognition that the benign conditions of recent years encouraged a sharp increase in leverage within the financial system. Responding to the high levels of debt that contributed to the current financial crisis, the Turner Review - a report on global banking regulation published in March and written by Adair Turner, chairman of the UK Financial Services Authority - proposed a limit on the absolute leverage of financial institutions. This has been echoed by other regulators.

However, the proposals do not fully recognise that new financial techniques have supplanted lending as the primary source of liquidity and leverage creation. As US humorist and actor Will Rogers drolly observed: "You can't say that civilisation don't advance, for in every war they kill you in a new way."

Traditional leverage, in the form of borrowings, remains important. A major area of growth has been collateralised lending, where holders sell assets against an agreement to repurchase them at a future date. The increase in hedge fund and prime brokerage activity fuelled explosive growth in this area and led to an expansion in the range of assets against which funding can be raised. Substantial sums could be raised against almost any type of security or instrument, complementing the long-established government bond repo markets.

The borrower posts an initial margin, or 'haircut', and promises to post more cash if the value of the asset declines. Favourable regulatory rules, optimistic views of liquidity (the collateral must be sold if the borrower fails to pay), and faith in the models used to set the initial margin drove aggressive use of collateral, increasing available liquidity and leverage.

Derivatives have also contributed to the sharp rise in leverage in several ways. The first is the 'derivatisation' of lending. For example, the purchase of $10 million of shares would typically require a commitment of cash. However, an investor can instead enter a total return swap (TRS). Under the terms of the swap (see figure 1), the investor receives the return on the share (dividends and increases in price) in return for paying the cost of holding the shares (decreases in price and the funding cost of the dealer). The TRS requires no funding other than any collateral required by the dealer - substantially less than the $10 million required to buy the shares outright. The result is that the investor acquires the same exposure as buying the shares but increases his or her return through leverage. In effect, the loan to the investor secured over the asset has been repackaged as a derivatives transaction, enhancing the leverage.

The second way derivatives can be used is to create embedded loss leverage, so the potential gain or loss is increased following a given event. Two examples - digital options and credit leverage in collateralised debt obligations (CDOs) - illustrate the idea.

Under a normal option to buy shares at $100 (the strike), if the shares at expiry of the option are above $100, the gain is equal to the share price minus $100. If the share price is $110, the purchaser of the option makes $10. In a digital or binary option, the parties can agree that if the share price is above $100, the option payout is $25. If the option is in-the-money (above the strike price of $100), the gain to the buyer is fixed at $25, irrespective of whether the share price is $100.01 or $200. In effect, for a relatively small move in the share price (from $100 to $100.01), the option buyer gains $25 (25% of the value of the share), effectively embedding tremendous sensitivity (that is, leverage) to price movements.

Similarly, options with conditional operation, such as barrier options that knock in or out, can rapidly increase or decrease exposure and leverage depending on price movements. The popular accumulator structures in equity and currency markets that have been problematic in recent times are examples of this phenomenon.

The equity tranches of CDOs offer an example of loss leverage in credit markets. A typical CDO may consist of a $1 billion portfolio comprising $10 million exposures to 100 corporations. The equity investor assumes the first 2% ($20 million) of losses on the portfolio. Assuming a loss of $6 million if any corporation defaults (recovery rates are 40% of $10 million), the equity investor is taking the risk of the first three defaults. In contrast, if the trader invested $20 million in the entire portfolio ($200,000 per corporation), then three defaults in the portfolio would result in the investor losing $360,000 (a loss of $120,000 per company, calculated as $200,000 adjusted for 40% recovery rates, times three). For three losses, the equity tranche investor's leverage to defaults is 56 times (if there were three losses, the investor loses the entire $20 million invested in the CDO equity, against $360,000 in the diversified portfolio).

By reducing the tranche width, leverage can be increased even further. For example, if the equity tranche is reduced to $10 million, the loss leverage is around 83 times. In contrast, a regulated bank can leverage to a maximum of 12.5 times.

Many innovations were motivated by increased leverage. For example, CDO-squared transactions offered higher returns to the investor by increasing leverage. CDO-squareds also used credit concentration to increase returns. In a CDO-squared, this takes the form of overlap risk - the same names may be present in the original CDOs.

The use of collateralised lending and derivatives to provide hidden leverage was augmented by 'layered' leverage. This refers to the practice of the same underlying cashflows being leveraged and releveraged through a series of interrelated structures.

The risk transfer model, for instance, frequently entailed layers of complex leverage. Banks institutionalised leverage in a bewildering array of off-balance-sheet structures - asset-backed securities (ABSs), residential mortgage-backed securities, commercial mortgage-backed securities, asset-backed commercial paper, conduits, structured investment vehicles (SIVs) and CDOs. These structures allowed the financial system to create multiple levels of borrowings, using the value of other underlying debt and cashflows as collateral.

Figure 2 sets out the daisy chain of risk. Banks originated loans that were subsequently securitised. Financial engineering was used to convert risky assets, with the help of rating agencies, into high-quality (AAA/AA) ABSs. The ABS notes were then sold to conduits, which issued highly rated commercial paper to money market investors. Alternatively, the high-quality ABS tranches were sold to SIVs, which then issued their own AAA rated debt to fund the purchase. Some SIVs even purchased highly rated debt from other SIVs in an astonishing chain of risk. High-quality ABS tranches were also sold to hedge funds that leveraged the AAA rated assets by up to 20-30 times via banks willing to lend against the value of the securities.

For example, hedge funds used leverage to enhance returns on subprime debt. Figure 3 shows how a hedge fund used $10 million to take the risk of the first $60 million of losses on an $850 million portfolio.

Increasing the amount of potential gain or loss for a given event and using complex chains of risk are now routinely used to create leverage. As Wells Fargo chief executive John Stumpf observed: "It's puzzling why bankers have come up with these new ways to lose money when the old ways were working so well."

The use of these techniques is poorly understood. Importantly, it does not always show up in traditional leverage measurements that are focused primarily on the level of borrowings. The additional liquidity and leverage creates complex chains of risk and moral hazard in markets that have proved problematic as prices corrected in the global financial crisis. It is another 'unknown unknown' of modern markets.

Current regulatory proposals do not recognise fully the issues of derivatives leverage - both disguised lending and embedding leverage - and the interrelationships between different structures in the risk transfer markets. The proposed limits on absolute levels of leverage and greater financial disclosure will do little to address the problems in financial markets that led to the present crisis. A further risk is that implementing absolute leverage limits will merely force changes in the methods by which leverage is created.

The current regulatory debate seems informed by John Kenneth Galbraith's observation: "Between human beings there is a type of intercourse which proceeds not from knowledge, or even lack of knowledge, but from failure to know what isn't known."

Satyajit Das is the author of a number of books on derivatives and risk management, including The Swaps & Financial Derivatives Library: Products, Pricing, Applications and Risk Management. He is also the author of Traders, Guns & Money: Knowns and Unknowns in the Dazzling World of Derivatives.