LITMIR.BIZ

Скачать книгу

section. There are four other costs associated with replacing managers, none of which are affected by the past performance of the managers. These include forgone interest on dormant cash, forgone excess returns on uncommitted cash, administrative costs of closing out one position and opening another, and the market impact of liquidating one position and starting a new position. The common factor driving the first two costs is that there are several leads and lags when making a decision to take money from one manager and placing it with another. The last two costs depend on the strategy being considered and the experience and resources of the investor.

      To understand the first two costs (i.e., forgone interest on dormant cash and forgone excess returns on uncommitted cash), consider the lags that exist in the process of replacing one manager with another. The first lag represents the time it takes to review a manager's results and make a decision about redemption. This might take anywhere from a few days to, in some complex cases, several months. The second lag represents the time between the notification deadline for a withdrawal and the moment when a net asset value is struck. This can take several weeks. The third lag is the time between the striking of net asset value and the receipt of the first round of cash. The fourth lag represents the time that passes between the receipt of the first round of cash and the final round of cash. The last lag is related to the time between when the entire position is liquidated and the cash is returned to the investor and when the cash can be allocated to a new manager.

      The first cost arises from the third and fourth lags and is associated with the liquidation and the forgone interest on dormant cash. This cost is borne by the investor, and depends entirely on a fund's practices with regard to interest payments on cash balances. Industry practices vary a great deal, but it is not uncommon to find that funds do not pay interest on the value of cash balances. In these cases, the cost of forgone interest depends on how quickly cash is returned to the investor.

      The second cost is associated with the last lag and represents the opportunity losses associated with liquidation and reinvestment. These losses stem from the intervals during which investments are not committed to enterprises that promise returns in excess of market interest rates.

      The third cost is related to transaction and administrative fees. Closing out old positions and opening new positions will entail administrative fees and due diligence costs. These costs will vary by investor. Experienced investors may have long lists of managers to choose from, and therefore due diligence costs may be relatively small. Also, since due diligence is a relatively fixed cost, the impact will depend on the size of the position. That is, this could be a significant cost for investors who have a relatively small allocation to this asset class.

      The final cost will depend on the liquidity of the positions that have to be closed and opened. If the positions are not liquid or the strategy being considered does not have a large capacity, then the market impact of liquidating the old positions and creating new positions could be costly, and will be borne primarily by the investor.

      2.1.4 Three Observations on TAA and Portfolio Reallocation Costs

      Given the preceding discussion, it appears that it will be very hard to make a convincing case for tactical asset allocation using alternative asset classes. However, a few observations may provide a basis for tactical asset allocation.

      First, by focusing on a few asset classes, the manager might be able to develop separate forecasting models for each and therefore generate forecasts with independent errors. While FLOAM presents IC and BR as somewhat independent parameters, they tend to be dependent in practice. For example, it is highly unlikely for a manager to have the skill to forecast returns on a large number of independent securities. Notice that the key word here is independent. This means that the manager applies one or more models to a set of securities that are not highly correlated, and therefore the forecast errors are independent from each other. This is a very strong requirement that is unlikely to be fully satisfied. In other words, there is a negative relationship between IC and BR. The more markets to which the manager tries to apply her skills, the less accurate the forecasts are likely to become.

      Second, the information coefficient tends to be much higher when applied to asset classes than when applied to individual securities. The random returns on individual security prices contain a significant amount of noise, which makes forecasting models less accurate. On the other hand, available empirical evidence suggests that expected returns on various asset classes or portfolios of securities behave in a more predictable way through various market cycles.11

      Finally, while TAA may be difficult and costly to apply to alternative assets, TAA can be applied to the traditional portion of the portfolio, where derivative products can be used to significantly alter a portfolio's characteristics without having to redeem allocations to certain illiquid funds. For example, suppose the equity beta of a diversified portfolio of traditional and alternative asset classes is given by β_Port. This can be estimated by regressing the historical returns of the portfolio against the returns on an equity benchmark. The portfolio manager can reduce the equity beta of this portfolio by selling equity futures contracts. In particular, since the beta of a portfolio is equal to the weighted average of the betas of its assets, we have:

(2.3)

      Here, β_New is the new beta of the portfolio, which could serve as the target by the manager, (F/P) is the ratio of the notional amount of the positions in the futures contracts to the size of the portfolio, and β_Futures is the beta of the futures contract with respect to the equity benchmark used to calculate the beta of the portfolio, β_Port. The beta of the futures contract is typically equal to one. An investor can engineer a new beta for the portfolio by adjusting the level of futures contracts, F.

      APPLICATION 2.1.4

Suppose the beta of a diversified portfolio against the S&P 500 Index is 0.9. The portfolio manager wants to increase the beta to 1.2 because of improving economic conditions. If the market value of the portfolio is $500 million, what notional position does the portfolio manager need to take in the S&P 500 futures market with a beta of 1 in order to achieve a beta of 1.2? What position would have lowered the beta to 0.4? Inserting the known values into Equation 2.3:

      That is, the manager has to take long positions in $150 million worth of S&P 500 futures contracts. If the manager had decided to reduce the equity exposure to 0.4, the futures position would have been:

      In the second case, the manager has to establish a $250 million short position in S&P 500 futures contracts to lower the beta to 0.4 from 0.9.

      Application 2.1.4 highlights an actual advantage that TAA may have when applied to broad asset classes: Portfolio managers may use liquid futures and swap markets to effectively implement the ultimate objective of TAA, which is to alter the portfolio's exposures to various sources of risk. This issue of factor exposures and factor investing will be further discussed later in the chapter.

      2.1.5 Keys to a Successful TAA Process

      As just illustrated, TAA has to overcome significant costs and barriers if it is to be applied to portfolios of alternative asset classes. However, TAA can also be applied to such a portfolio not through the actual sale and purchase of illiquid alternatives but through the use of futures contracts to alter the portfolio exposure to sources of risks. This section discusses the TAA process.

      2.1.5.1 The TAA Process and Return Prediction

      The key component of a successful TAA process is the development of sound models that can consistently forecast returns across asset classes. This point may appear to contradict the basic tenet of the efficient market hypothesis that asset returns are essentially unpredictable. There is strong evidence that returns to asset classes are indeed predictable (Pesaran 2010). What is less clear is whether this predictability is the result

Скачать книгу