Hedge fund CFOs are one of the most recent innovations within the CDO market with the completion of the first transaction in June 2002. Unlike traditional CDOs, which are securitizations of bonds or loans, the assets securitized in hedge fund CFOs are the shares of various hedge funds. Therefore, hedge fund CFOs differ in important respects from traditional CDOs in regard to key risk factors. While the performance of a traditional CDO relies mainly on the credit performance of a pool of underlying bonds or loans, i.e., default and recovery rates, the performance of a hedge fund CFO depends on the net-asset value (NAV) of the underlying hedge fund shares at the liquidation date. The NAV of these shares depends on the performance of a potentially endless array of debt, equity, hybrid, or derivative positions, held long or short, and representing a diverse and often abstract assortment of financial risks and rewards. Below is a description of the main differences between traditional CDOs and hedge fund CFOs and how these translate when modeling these products for rating purposes. Over-Collateralization
Both CDOs (in particular cash-flow arbitrage CDOs) and hedge fund CFOs possess a structural feature known as over-collateralization (OC) triggers. The OC ratio is the amount of assets held by the special purpose vehicle (SPV) divided by the par amount of rated liabilities outstanding. An OC trigger is set at closing such that, if the ratio falls below the trigger during the life of the transaction, action will have to be taken in order to restore compliance. In the case of a CDO, this is achieved by diverting interest proceeds realized on the portfolio from the junior noteholders to repay principal on the senior notes until compliance is restored. For a hedge fund CFO, the OC mechanism differs from a CDO in two important ways. While the numerator of the OC ratio usually represents the par amount of the assets held by the CDO (save for CCC and defaulted assets which are often adjusted by their market price), the numerator of the OC ratio in a hedge fund CFO usually represents the current NAV of the hedge fund shares held by the CFO. This distinction is important because it makes the OC ratio of a hedge fund CFO more subject to market price movements than for a traditional CDO. Furthermore, in case the OC trigger is breached in a hedge fund CFO, the manager is obliged to liquidate portfolio assets and use the proceeds to redeem liabilities or buy cash assets until the OC ratio is once again in compliance. This also makes the hedge fund CFO dependent on the market price and liquidity for the assets it needs to liquidate while this is not true for a traditional CDO which is not forced to liquidate assets when the OC trigger is breached but simply needs to collect interest proceeds and redistribute them to senior noteholders.
Market value risk can be quantified by using a Monte Carlo simulation model that samples historic data provided by the manager, available hedge fund indices or a combination of the two. In a sampling approach, the data set consists of historic strategy returns organized by date. On each scenario of the simulation, a historic period is chosen to correspond to each future date in the life of the transaction and the historic strategy returns are copied into that date. The portfolio return is the weighted average of the strategy returns based on the strategy weights assumed for the portfolio. The sampled return for the portfolio in a given scenario allows calculating the NAV of the assets throughout the life of the transaction. If some assets need to be liquidated to restore compliance of the OC ratio, the liquidation value will be equal to the NAV of the assets calculated in the model for this scenario at that point in time.
Liquidity Risk
The second main difference between traditional CDOs and hedge fund CFOs is liquidity risk. This risk is limited in a traditional CDO because the assets pay a periodic coupon that is usually sufficient to service the interest on the notes. In a hedge fund CFO, however, the underlying assets typically do not pay current distributions in the manner of a periodic coupon. Therefore, the interest on the notes has to be paid by the liquidation of some of the assets. Usually these assets are not freely redeemable on demand but will instead be subject to initial lock-up periods of perhaps one year during which redemptions are not permitted and, following the lock-up period, to minimum notice periods and regular (but rigid) liquidation dates. Hypothetically, 15% of the portfolio may be recognized as being liquid on demand, perhaps having been allocated to a managed account; an additional 30% might require three months from the time of the redemption order for the structure to get back money; the next 25%, six months; and the remaining 30%, nine months to a year. Share redemptions may be further restricted by rules restricting the amount of assets of the fund that can be liquidated in aggregate for investor redemptions.
This risk can also be modeled by implementing a module within the simulation model to test fluctuations in the liquidity profile of the fund-of-funds portfolio and its effects on available liquidity for payments. The liquidity module accommodates several liquidity baskets for the portfolio with assigned door-to-door liquidity times. As redemptions are ordered from the most liquid baskets, the liquidity module keeps track of how quickly the most liquid baskets can be replenished and used for further payments. And, if necessary, will register non-payment of an obligation that arises before liquid funds are adequately replenished and if external liquidity support, such as an external credit facility is nonexistent or exhausted. For conservatism, one may assume the investor has chosen the least opportune time during the redemption cycle to give notice, for example, the day after the last date on which it would have been eligible to receive cash at the soonest liquidation date. If a fund requires at least three months notice for redemptions, but fulfils redemption orders on the last days of March, June, September and December of each year, an investor placing such an order on April 1st may not receive money until Sept. 30. As such, the worst-case door-to-door redemption time is six months (less one day).
Benchmarking
The third main difference is performance benchmarking. While traditional CDOs can easily be benchmarked against a default curve, which represents the historical default frequency of an asset with a given rating over a certain period of time, finding an appropriate benchmark for a hedge fund CFO is more delicate. One has to assign a strategy to each hedge fund in the portfolio and then compare the hedge fund to the relevant index. For example, the CSFB Tremont-Index is broken down into 13 sub-indices, each sub-index representing a hedge fund strategy. This classification by strategy is important because different strategies can have substantially different risk/returns characteristics. It ensures that the assumed returns sampled in the model will match as closely as possible the characteristics of the portfolio.
The fourth difference is the potential of negative correlation between interest rates and hedge fund returns, i.e., the risk that interest rates and the NAV of the portfolio will move in opposite directions. An increase in interest rates, together with a reduction in the NAV of the portfolio would affect a hedge fund CFO in two ways: first, the rise in interest rates would increase the interest burden as CFO notes are typically floating rates instruments. On the other hand, the value of the assets would decrease, making it more likely for the CFO to be unable to pay the coupon due to the noteholders. This risk is less relevant for traditional CDOs because they are not relying on the NAV of the assets to pay coupons to noteholders. In a CLO, an increase in interest rates would usually be neutral for the rated notes since both assets and liabilities would typically be floating-rate instruments. Therefore, the increase in interest due would be offset by an increase in interest collected on the assets. In a CBO, an increase in interest rates would adversely impact the performance of the transaction because of the increase of interest due to the noteholders while the interest collected on the assets would be unchanged. This would typically be mitigated by an interest rate swap the CBO put in place at closing and it would only affect one side of the balance sheet. That is the liabilities side of the CDO, while the asset side would be unaffected. In a CFO, both sides of the balance sheet would be affected. Therefore, this risk would affect a CFO more dramatically than it would affect a CDO. This risk can be quantified by implementing a stochastic interest rate model and correlating the random component of this model with the random component of the model governing the NAV fluctuations using Cholesky factorization.
Leverage
Lastly, the use of excessive leverage is often mentioned as a risk factor in hedge fund investing while it is already captured in a traditional CDO by the rating of the underlying assets. In a hedge fund CFO, the transaction portfolio might become systematically over-levered for any number of reasons, including a lack of transparency or oversight. It is therefore important to discuss with the asset manager in order to gain comfort with management practices in regard to monitoring the aggregate level of risk, including that arising from leverage, across the fund-of-fund portfolio. Certain hedge fund structures may incorporate structural volatility tests that indirectly address leverage-related risk; these tests cause the OC triggers to be raised in the event of increased portfolio volatility.
This week's Learning Curve was written by Benjamin Tolédano, associate director, and Michael Romer, senior director, in the alternative investments team at Fitch Ratings in London.
