US 20060106705 A1 Abstract Computer technology for substantially optimizing portfolios of multiple participants is disclosed. Preferably the portfolios of such multiple participants comprise fixed income instruments. The disclosed systems and methods include using at least one computer system for storing digital data representing portfolio holdings of multiple parties and, in particular, for each participant storing in the computer memory data representing constraints with respect to the desired portfolio. The method and system comprise optimizing using an optimization engine portfolio and constraint information of multiple participants so as to generate a set of trades that would substantially optimize participants portfolios with respect to a known objective.
Claims(69) 1. A computer method for adjusting portfolios of fixed income instruments of multiple parties comprising:
storing in memory of at least one computer digital data representing portfolio holdings of multiple parties; storing in the memory of at least one computer digital data representing constraints that define trading requirements of the parties; converting, using at least one computer, the digital data representing the portfolios of multiple parties and the digital data representing the constraints of the multiple parties to optimization digital data adapted for processing by an optimization engine; and optimizing using at least one computer the optimization digital data so as to generate a set of trades among the parties that rebalance the parties' portfolios in accordance with parties' constraints such that the portfolios are substantially optimized with respect to a predetermined objective. 2. The method of 3. The method of 4. The method of 5. The method of 6. The method of 7. The method of 8. The method of 9. The method of 10. The method of 11. The method of 12. The method of 13. The method of 14. The method of 15. The method of 16. The method of 17. The method of 18. A computer method for adjusting portfolios of fixed income instruments of multiple parties, comprising:
storing in memory of at least one computer digital data representing portfolio holdings of multiple parties, the multiple parties comprising two or more affiliated parties, the portfolio holdings comprising at least one fixed income instrument held by at least one of the two or more affiliated parties; storing in the memory of at least one computer digital data representing constraints that define trading requirements of the parties, the defined trading requirements comprising distinct trading requirements for each of the two or more affiliated parties; converting, using at least one computer, the digital data representing the portfolios of multiple parties and the digital data representing the constraints of the multiple parties to optimization digital data adapted for processing by an optimization engine; and optimizing using at least one computer the optimization digital data so as to generate a set of trades among the parties that rebalance the parties' portfolios in accordance with parties' constraints such that the portfolios are substantially optimized with respect to a predetermined objective; wherein the digital data representing the constraints includes digital data representing system constraints stored in the memory, and wherein said system constraints comprise constraints designed to reduce the likelihood of trades between the two or more affiliated parties. 19. The method of if the at least one fixed income instrument is not held by any party other than the two or more affiliated parties, then said system constraints comprise a constraint that prevents the two or more affiliated parties from buying the at least one fixed income instrument, and if the at least one fixed income instrument is held by at least one party other than the two or more affiliated parties, then said system constraints comprise a constraint that requires the amount of the at least one fixed income instrument bought by the two or more affiliated parties to be less than the amount of the at least one fixed income instrument sold by all parties other than the two or more affiliated parties. 20. The method of 21. The method of 22. The method of 23. The method of 24. The method of 25. The method of 26. The method of 27. The method of 28. The method of 29. The method of 30. The method of 31. The method of 32. The method of 33. The method of 34. The method of 35. The method of 36. The method of 37. The method of 38. The method of 39. The method of 40. The method of 41. The method of 42. The method of 43. The method of 44. A computer method for adjusting portfolios of fixed income instruments of multiple parties, comprising:
storing in memory of at least one computer digital data representing the portfolios of the multiple parties; storing in the memory of at least one computer digital data representing constraints that define trading requirements of the parties; converting, using at least one computer, the digital data representing the portfolios of the multiple parties and the digital data representing the constraints of the multiple parties to optimization digital data adapted for processing by an optimization engine; supplying first pricing information for the fixed-income instruments in the portfolios of the multiple parties; optimizing using at least one computer the optimization digital data and the first pricing information so as to generate a first set of trades among the parties that rebalance the parties' portfolios of fixed-income instruments in accordance with the constraints that define trading requirements of the parties such that the portfolios are substantially optimized with respect to at least one predetermined objective; communicating the first set of trades to each of the multiple parties; receiving approval of the first set of trades from each of the multiple parties; supplying second pricing information for the fixed-income instruments in the portfolios of the multiple parties, said second pricing information comprising prices quoted by traders of an intermediary entity that facilitates trades among the parties that rebalance the parties' portfolios of fixed-income instruments; and optimizing using at least one computer the optimization digital data and the second pricing information so as to generate a second set of trades among the parties that rebalance the parties' portfolios of fixed-income instruments in accordance with the constraints that define trading requirements of the parties such that the portfolios are substantially optimized with respect to at least one predetermined objective; and executing the second set of trades at the prices quoted by the traders of the intermediary entity. 45. The method of 46. The method of 47. The method of 48. The method of 49. The method of 50. The method of 51. The method of 52. The method of 53. The method of 54. The method of 55. The method of 56. The method of 57. The method of 58. The method of 59. The method of 60. The method of 61. The method of 62. The method of 63. The method of 64. The method of 65. The method of 66. The method of 67. The method of 68. The method of 69. The method of Description This is a continuation of application Ser. No. 08/963,605, filed Oct. 31, 1997, now allowed. This invention relates to computer technology for optimizing portfolios of multiple participants and, in particular, for optimizing portfolios of fixed income instruments. It is well known that computer technology can be effectively employed for financial applications. It is also known to employ computers that execute optimization programs, such as programs based on linear programming techniques, so as to achieve financial goals. For example, computer technology that analyzes and optimizes a portfolio held by a given entity is known. Computer systems have also been employed as an intermediary in transactions where multiple parties desire to trade specific equity instruments. In such computer applications, optimization may be employed to facilitate trading of an equity of interest. However, the inventors are not aware of computer technology developed for trading holdings of multiple participants, where a computer acting as an intermediary processes entire portfolios of the participating entities and generates trades that optimize portfolios for a desired result, particularly for portfolios of fixed income instruments. Portfolio-based trading, for example, exists in the equities market, where investors may buy or sell a portfolio of stocks on an aggregate basis. The investor provides a statistical description of the portfolio, usually including how closely it tracks the S&P 500 index, the sector distribution of the portfolio, and a measure of the diversification of the portfolio. The broker then commits to trade the portfolio of unknown stocks for a fixed fee at the prevailing market price at a pre-arranged point in time, typically the market daily close. Because the broker only knows the “statistical” composition of the portfolio, the investor feels more comfortable that the broker is unable to affect the closing prices. Because of the statistical relationship between the portfolio and the index, the broker feels comfortable that the investor cannot unload a portfolio of unattractive securities. An important component of such a transaction is the independent price of equities contributed by the public transaction records of the equity markets. The vast majority of fixed income transactions are performed on a principal basis where the broker takes the opposite side of the transaction from the investor. The lack of adequate fixed income transaction records and the broad range of structures and maturities of fixed income instruments creates a significant barrier to developing the confidence on either side of the transaction that pricing is fair. Thus, it is desirable to provide a system that employs unbiased pricing and reassures the investors that the transaction is a fair deal. Further, it is desirable to provide computer technology that supports such fixed income transactions and, in particular, enables multiple parties to participate in the transactions. In particular, it is desirable to develop computer technology that would allow multiple investors to specify constraints on their portfolio holdings and, within those constraints, allocate by the optimization computer process fixed income holdings to individual investors participating in the transaction. As noted, in general, optimization techniques for financial applications are known. For example, Adamidou et al., Optimization methodologies relating to financial applications are surveyed in H. Dahl, A. Meeraus, and S. A. Zenios, Such publications on financial engineering do not teach computer technology that enables multi-party portfolio trading in fixed income instruments, wherein computer-driven optimization aids in rebalancing portfolios of multiple participants. Yet, there is a need for such technology. For example, there is a need to provide computer technology that enables multiple investors to recognize the economic benefits of selling bonds at a price below the price originally paid thereby obtaining a tax deduction. Accordingly, there is a need to develop technology that would enable investors to exchange portfolio holdings so as to substantially maximize the tax deductible loss. It is believed that technology for such portfolio trading between multiple parties that enables them to substantially optimize trades so as to substantially maximize tax advantages has not been developed by others. Although the system and method of the present invention relates to computer technology applicable to a wide array of portfolio optimizations in trading among diverse parties, the preferred embodiment relates to a computer system and method that provide a capability of taking advantage of refunds on taxes paid within the previous three years by maximizing book losses on trades of multiple participants. The preferred embodiment provides technology that enables trades as swaps among multiple parties while keeping the trades out of the market. The advantage of swapping between portfolios of participating firms versus transacting in the open market is that large scale trades can be executed without adversely affecting the market trading. In addition, the specific preferred embodiment enables swap members to buy discount bonds as replacements, which may be problematic in the open market but provides further, two advantages. The computer technology of the preferred embodiment facilitates a solution to a multi-party book-loss optimization. In general, the input to the computer system of the preferred embodiment comprises a set of bond portfolios owned by a group of firms, and the output comprises the set of trades which substantially maximizes the participant firms' total book losses. The implementation of the preferred embodiment avoids churning (i.e., buying and selling the same security) and wash sales (i.e., buying and selling a sufficiently similar security) and, therefore, reduces a risk of degeneracy in the process of maximizing book losses. In addition, individual firms typically have portfolio composition constraints that must remain satisfied in any intermediated transaction implemented by the system. Such constraints may include fixed market value of holdings within given sectors and maximum holdings of given names. The implementation of the preferred embodiment provides means for satisfying such constraints. Although a particular implementation of the preferred embodiment relates to producing tax deductions, a person skilled in the art will realize that it can be generalized to allow different participants to have different objectives and still produce multi-party portfolio-based optimized transaction. Furthermore, as will be understood by a person skilled in the art, extensions are possible where the participants provide prices at which they would be willing to buy or sell rather than using uniform prices provided by the intermediary entity, as in the preferred embodiment. In general, a person skilled in the art will appreciate that the invention can be extended to accommodate differing views among the participants on the economic attributes of the fixed-income instruments in their portfolios. The invention will be better understood when taken in conjunction with the following detailed description and accompanying drawings, in which: The following detailed description of the preferred embodiments is organized as follows: first, computer architecture of the preferred embodiment is disclosed. Next, a specific illustrative application addressed by the technology of the preferred embodiment is described. Thereafter, software programming developed for implementing the illustrative application of the preferred embodiment is disclosed. Computer architecture of the preferred embodiment is depicted in The data representing portfolio information and specific client constraints is provided by the clients participating in the transaction to the workstation Client portfolio data and their constraints are then translated into a uniform format discussed below and entered into a front end module, symbolically illustrated as The workstation In the preferred embodiment, data obtained from external sources includes: bond indicatives (e.g., coupon, maturity, etc.) from EJV, Electronic Joint Venture (EJV) Capital Markets Services (http://www.ejv.com) 1996, and Bloomberg, Bloomberg L. P. 499 Park Ave., NY, N.Y. 10022 (http://www.bloomberg.com) 1996, databases, insurance company holdings from the Capital Access FINCOM database, Capital Access Corp. Mountain Heights Center, 430 Mountain Ave. Murray Hill, N.J. 07974 (http://www.interactive.net/˜cac) 1996, sector descriptions from EJV and Fact Set, FactSet Research Systems Inc., One Greenwich Plaza, Greenwich, Conn. 06830 (http://www.factset.com) 1996. The data stored in the front end The CPLEX optimization engine of the preferred embodiment is a linear optimizer for solving linear programming problems encountered in a wide variety of resource allocation programs. CPLEX provides several solvers for different problem environments. See http:\\www.cplex.com. The CPLEX Linear Optimizer Base System provides a basic linear programming environment using continuous variables and employing algorithms mainly based on a well-known Simplex method. It also supports a variety of input/output formats such as MPS files, known in the art. This system can handle problems with millions of constraints and variables. The CPLEX Mixed Integer Solver (MIP) is an addition to the CPLEX Linear Optimizer Base system. It employs various heuristic algorithms such as a branch-and-bound technique to handle the difficult optimization problems involving integers. The CPLEX Barrier/QP Solver is an optimizer for solving linear and quadratic problems. CPLEX can be run on various computer platforms. The CPLEX programs are also available as parallel versions so that they can be run on multiple-CPU systems for increased performance. The optimization engine The optimization engine It should be noted that the computers of the disclosed embodiment are, in general, known devices that include a central processing unit, primary and secondary memory, and network interfaces, as well as other well known hardware components. As discussed, these computers are configured for the special purpose of providing substantial optimization of multi-party trades using software discussed herein. In another embodiment, the system of Tax law allows corporations to apply losses realized in a given year against gains incurred within the previous three years to receive tax rebate for previous taxes paid. See 1996 A tax swap is beneficial if tax refunds received today have positive economic value considering the present values of the bonds swapped to achieve the refund. If two firms own underwater bonds (i.e., bonds which values have dropped in comparison to their original values), swapping such bonds for bonds owned by others may enable the firms to take advantage of the tax refund. Tax-related advantages can, for example, result from swapping an underwater bond with a par bond and with a discount bond as discussed below. Swapping an underwater bond with a par bond that would produce the same yield as the underwater bond requires that the par bond necessarily has a higher coupon. Accordingly, the tax refund received today as a result of a swap is offset by higher future taxes paid on the greater coupons of the par bond. Also, the principal par amount invested in the underwater bond is necessarily larger than the principal amount of the par bond purchased as a result of the swap. Accordingly, some protected principal is lost due to the swap. The net of these effects depends on the discount factor, so that for reasonable discount factors, as illustrated below, the net effect favors doing the swap. For example, consider a swap of a $100 MM par amount of a 6.750% coupon underwater bond, having a current market price of 97.411% yielding 7.750%, for par bonds of equal yield, i.e., 7.750% coupon. Thus the owner of the underwater bonds obtains after the swap $97.411 MM par amount of the new bonds with 7.750% coupon. For the purposes of this illustration, it is assumed that all coupons are paid annually, that both bonds mature in three years from the day of the swap, and the tax rate is 35%. The net economic benefit of swapping the bonds is determined as follows. The seller of the underwater bonds receives $97.411 MM plus a tax refund of 35%×(100%-97.411%)×$100 MM=$0.906 MM. The same entity then uses the $97.411 MM to buy new par bonds, netting the tax refund. On three successive years, it receives, after taxes, a coupon of (100%−35%)×7.750%×$97.411 MM=$4.907 MM. On the third year, it also receives the return of the $97.411 MM principal. The opportunity cost of foregoing owning the underwater bond includes its coupons and return principal. The after-tax coupons would have been (100%−35%)×6.750%×$100 MM=$4.388 MM. The return principal would have been $100 MM. This analysis is summarized in Table 1, which uses a discount factor of 65%×7.75%=5.0375%.
The breakeven discount rate that makes the swap beneficial is 2.8%. The profitability of the swap increases with increasing maturity of the bonds, decreasing price of the underwater bonds, and increasing discount rate. Alternatively, one may swap for market-discount bonds, i.e., bonds currently trading at a discount. Normally, securities are taxed on an effective-yield basis; however, market-discount securities have different taxation. If the income from the bond exceeds the financing cost for the bond (which is assumed to be true in this example), the investor may elect to pay tax on cash flow rather than yield. For a discount bond, tax on cash flow is always lower than tax on yield. If the investor makes this election, there is an additional tax due on excess of sale or redemption proceeds over cost. This election may be made on a bond by bond basis. Swapping to a market-discount bond achieves greater economic benefit than swapping to a par bond, as illustrated in the example below. For simplicity, consider that the underwater bond, discussed in the previous example, is swapped with a bond of identical attributes (but different issuer to avoid a wash sale). The only modifications to the previous analysis are the cash flows of the new discount bonds. As a result of the swap, the new bonds are bought for $97.411 MM, netting the tax refund. On three successive years, the investor receives, after taxes, a coupon equal to the coupon foregone. On the third year, the investor also receives the return principal of $100 MM, however, we are required to pay tax on the accrual from the discount price. Thus, we receive $100 MM plus $4.388 MM minus 35% of ($100 MM−$97.411 MM)=$103.48 MM. This analysis is summarized in Table 2 below, which illustrates that the resulting profit is greater than that of the previous scenario.
Software Implementation In general, a multi-party book loss optimization problem of the exemplary application described above is well-suited to linear programming, a known optimization technique. Book loss is defined as the par sold multiplied by the difference in book price and market price for the securities available in the secondary market at the time of the transaction. Table 3 below defines variables used in the following discussion, where indexes i, j, k correspond to the set of all bonds, firms, and sectors, respectively.
In the following discussion it is assumed that PRICE The objective function representing total book loss, optimized by the system, is expressed as follows:
Bond conservation: for a given bond, the par amounts bought and sold over all participating firms must net to zero, i.e., there is a closed universe of bonds.
Present value neutrality: for every firm, the total of all trades must be present-value neutral.
Duration neutrality: the total of all trades must leave the dollar-duration within a reasonable tolerance. This is a relaxed form of dollar-duration-neutral trading. The constraints are applied on a per party (j) basis.
Convexity neutrality: These constraints are similar to the above constraint, except that $DUR Other market-value weighted attributes: Yield and rating are constrained in an identical manner as duration and convexity. In other embodiments, other portfolio characteristics can be defined in a manner similar to duration and convexity. Par-value weighted attributes: Maturity and coupon are constrained in a manner similar to duration and convexity; however, par-value rather than market-value is used for weighing. As noted, in other embodiments, other characteristics can be similarly defined. Proceeds bounding within sectors: The total of all trades must leave the present value (within every sector) between reasonable (predefined) bounds. These constraints can enforce present-value-neutral trading, possibly weakened to provide additional flexibility. Alternately, the use of these constraints may provide an opportunity to employ the transaction in order to reallocate the portfolio. These constraints, expressed below, are applied on a per party (j) basis.
The sectors include an industry sector type, such as Financials, Utilities, Industrials and Sovereign/Agencies, as well as other types of sectors including rating, broad maturity, fine maturity, duration, convexity, EJV sector, EJV subsector, EJV subsubsector, holdings, issuer, SIC code, and other sectors customized to specific firms. Another category of sectors is a specification of bonds that cannot be sold to a given firm. Non-negativity and boundedness: the amount bought and sold must be non-negative, and the amount sold must not be greater than the original par amount owned. Additionally, the amount bought must not exceed the total amount owned by all other firms.
If the right-hand-side of the buy equation is zero, then no variable BUY In the model defined by the above objective function and constraints, churning and wash sales may occur when more than one party owns the same bond. Churning refers to buying and selling of the same security to generate spurious book loss. Churning involves swapping bonds with the same CUSIP (Committee on Uniform Security Identification Procedure) code. A wash sale involves bonds that satisfy the following three similarity conditions: 1) the same issuer, 2) maturities within five years of each other, and 3) coupons within 25 bp. To eliminate churning and wash sales, the results obtained by employing the above continuous model may then be modified by computing net sales of each bond for each firm. The net sales would then be presented as a resultant portfolio produced by the transaction. However, it is unlikely that the resultant allocation of bonds would be substantially optimal with respect to the goal of book loss maximization, and therefore this is not the preferred approach. For example, if each of two firms hold two bonds A and B, co-members of a given sector, the objective function may be maximized by a wash sale of bonds A and B: each sells the other both A and B. If only net sales are taken into account these sales would net zero for each bond, and, therefore, no trades and book losses would be produced. The optimal solution, however, is for one firm to sell A and the other to sell B, allowing each to achieve a book loss. The formulation of the objective function, provided above, maximizes achieved book loss. In an alternative embodiment, this function can be generalized as follows to include the economic value of tax deferral:
The implementation of the preferred embodiment for the exemplary application considered here, enhances the continuous model discussed above by employing mixed-integer techniques. The enhancement of the preferred embodiment effectively addresses the issue of churning and wash sales (including taking into account bonds owned by the subsidiaries of the same parent). In the preferred embodiment, SELL Two or more affiliated parties (e.g., subsidiaries of the same parent firm) cannot trade with each other, yet may require different constraints in order to not be treated as a single entity. In the preferred embodiment this requirement is modeled in the following manner. If a bond i is originally held by at least one of the affiliated parties θ, two cases are possible: (1) i is not held by any party outside of θ; or (2) at least one party outside of θ holds i. Accordingly, in the first case, the constraint BUY Although mixed-integer programs such as presented before are difficult to solve optimally for large data sets, sufficiently satisfactory solutions can be obtained using the method of the preferred embodiment as described herein. Results that are not strictly optimal, but are sufficiently optimized to be acceptable, may also be referred to as optimal in this discussion. The front end User and System constraints can be specified and stored in the front end Constraints on sectors specify (1) which sectors are constrained; (2) over what statistic the constraint is defined; and (3) the bounds of the constraint. The sectors within a constraint are defined either as an individual identifier or any number of identifiers connected with logical operators. The following grammar for name expressions is used to specify sectors.
An identifier is either a full CUSIP, a name (a six character CUSIP), or an alpha-numeric string previously defined as a sector of a certain bond. For example “˜(AAA|AA)” specifies all bonds rated lower than AA. The “|” operator is logical OR; “&” is logical AND; and “˜” means NOT. Parenthesis are used in a conventional manner. A full CUSIP specifies an exact bond issue, whereas a name specifies an issuer. For example “912827T6 & 312911” specifies a single Treasury bond and a group of mortgages. A client firm, for example, may specify names it refuses to buy, e.g., “˜369856”. The following grammar is used for constraint specification:
In the above grammar, “print-name” is an optional string that provides textual representation of the constraint for summary purposes. The “Or-applies” expressions specify a group of firms in which a given constraint applies individually to each firm. The “And-applies” expressions specify a group of firms to which the constraint applies collectively. In general, a constraint is of the form:
The base statistic is defined by both variable and value specifications. For example, if a firm is interested in constraining the market-value-weighted dollar duration of all bonds it buys, the numerator is set to #PV#DUR#BUY. The variable #BUY specifies that the set of bonds bought should be considered. The values #PV#DUR specify that the desired statistic is present value times duration times par amount. Other variables that can be used are #SELL (bonds sold), #NET (buys minus sells), #SECTOR (pay attention to the sectors specified in the constraint), #ALL (ignore sectors), #FINAL (original plus buys minus sells), and #AVG (buys plus sells divided by two). These variables can also be combined as in the example above. The values include #CONV (convexity), #MAT (maturity), #COUPON (coupon), #RATING (rating) and #LOSS (book price minus price), as well as other values defined by the user, as will be understood by one skilled in the art. As mentioned, the denominator is used to optionally normalize the base statistic. For example the previous numerator #PV#DUR#BUY needs to be normalized by the denominator #PV to compute a valid duration. All the variables specified above can be used in the denominator. In addition the variables used in the denominator include #CURR (current portfolio) and #NONE (denominator equals one). Commonly used constraints may also be specified as macros. Constraints can be bound with respect to #ABS (absolute value of bounds), #REL (a value relative to a base value, i.e., base value±percentage points), and #PROP (proportional values, i.e., base value multiplied by percentages; the base value is always computed from the incoming portfolios). For example, suppose firms
Here the zero lower bound guarantees that the original convexity cannot be lower than the resulting convexity. The large upper bound indicates that convexity is allowed to increase up to 1000% of the original value (essentially unlimited). At step Next at The optimization engine at step After the optimizer has completed its processing, a transaction proposal is generated at step If all clients agree on the proposed transactions, as illustrated in First the yields of currently traded US Treasuries are determined as known in the art. Instead of using all US Treasury prices, only the on-the-run prices are used. First, the closing prices of every UST and the market prices of all the on-the-runs are collected. Second, a butterfly portfolio for each UST is constructed using the two on-the-runs with the closest durations as barbells. Third, the change in the current present value of each UST is determined by that of the two ends of the barbell, taking into account the butterfly weights. Subsequently, the prices of the bonds used in a transaction are easily computed based on the spreads quoted by the traders. The yield of a bond is the yield of the benchmark plus the spread. The spread quoted may be based on yield to maturity, yield to call, yield to put, or yield to average life. The date corresponding to settlement of the final transaction has to be used when converting the bond yield to the final bond price. Upon receiving the actual trading prices from the traders, the optimization is repeated at step Alternatively in another preferred embodiment, the actual prices provided by traders may be entered into the system before the complete agreement of the parties on the final transaction has been reached. Specifically, in such an embodiment, the actual prices are introduced when the parties are in substantial but not complete agreement with respect to the proposed swap, so that several final iterations involving optimization are performed with the actual prices obtained from the traders. This embodiment modifies flowcharts of The Optimization interface module At step At step If the numerator variable is #ALL then the program does not check for sector inclusion: this bond will have non-zero coefficients. If the numerator variable is #SECTOR then the interface program If the bond is constrained, the program determines the proper coefficient a The program also accounts for an optional denominator. To save MPS file preparation time, the program generates the denominator only once for both the upper and lower bounds. This is done by generating a new linear programming variable and creating an equality constraint for the denominator.
The new linear programming variable is then appended to the end of upper and lower bounds inequalities. The coefficient of the new linear programming variable is the negative upper or lower bound, respectively, as illustrated below.
Next, at step As noted at step The previously described preferred embodiment is neutral with respect to multiple firms, i.e., no firm is given an advantage over another. However, the resultant trades may distribute gains among the firms not completely evenly. Although, completely fair distribution of gains is difficult, the fairness of the distribution can be improved by utilizing one of the techniques discussed below, or other techniques known in the art. Although the solution which does not attempt to achieve a fair distribution is sufficient for the implementation of the preferred embodiment, alternative embodiments may include additional processing that addresses fairness as discussed below. One such approach to achieving fairness that may be used in an alternative embodiment is to employ a method developed by Shapley for constructing a “fair” solution to the classic coalition problem in game theory. See H. Raiffa, In formulating a tax swap as a coalition problem, the majority of a subgroup's utility is attributed by its tax loss, which can be evaluated with the optimizer for each subgroup. Two additional factors contributing to utility include: 1) a consideration that discount securities (priced below par), purchased in the swap, have a smaller future tax burden than par or premium securities, so that all players wish to swap in discount securities; and 2) by swapping among themselves, the firms have less total transaction costs than the market would charge, especially considering premiums due to the inelasticity of supply of discount bonds. Once these considerations are factored into the subgroup utilities, Shapley values can be computed, to determine a fair division of proceeds. In some alternative embodiments, it may also be desirable to tilt the objective function. Since the objective function thus far is to maximize total loss, it may be achieved through one firm receiving a disproportionate share of the tax loss relative to other firms. One method of rectifying the immediate book-loss and concomitant tax advantage bias is with the following objective function:
To negotiate an actual deal it is important for the entity acting as an intermediary to standardize security prices in the resulting trades in accordance to the market prices of the corresponding investments. As discussed above, the benchmark pricing module manages such a pricing. The standardized pricing gives the multiple parties to the swap confidence in the impartiality of the intermediary entity. Payment for the services of the intermediary may for example, come from a fixed percentage of realized tax deduction, or using another compensation scheme. Individual parties must be prevented or at worst dissuaded from “cherry picking” prices or securities, i.e., viewing the optimized trades and selectively committing to only certain trades. For example, a party which avoids an assigned buy trade that is perceived as too expensive is hoping to engage in a form of arbitrage. That party wants to buy at no worse than fair value, but of course does not identify the bonds it is selling above fair value. The intermediary entity must tightly control the timing of the swap, not allowing individual parties to stretch the target trade date. With time slippage comes the risk that the market will rally. If the market rallies, there will be fewer underwater securities in the pool and less losses embedded in each security. One technique of controlling timing is to limit participation and plan a series of swaps. The present invention is not to be limited in scope by the specific embodiments described herein. Indeed, modifications of the invention in addition to those described herein will become apparent to those skilled in the art from the foregoing description and accompanying figures. Doubtless, numerous other embodiments can be conceived that would not depart from the teaching of the present invention, which scope is defined by the following claims. Referenced by
Classifications
Rotate |