Information Aggregation, Currency Swaps, and the Design of Derivative Securities

A model of security design based on the principle of information aggregation and alignment is used to show that (i) ̄rms needing to ̄nance their operations should issue di®erent securities to di®erent groups of investors in order to aggregate their disparate information and (ii) each security should be highly correlated (closely aligned) with the private information signal of the investor to whom it is marketed. This alignment reduces the adverse selection penalty paid by a ̄rm with superior information. Adverse selection costs are often contingent on ex post publicly observable and contractible state variables such as exchange rates. In such cases, debt contracts are dominated by currency swaps and optimal securities, in general, are derivative contracts that are contingent on state variables that in°uence adverse selection costs. This is because the netting of cash °ows in these derivative contracts, in e®ect, alters the state-by-state seniority of di®erent claims in a desirable way. JEL classi ̄cation: G10 One of the long held tenets of ̄nancial economics is that in a perfect capital market, the precise packaging of securities is irrelevant. However, the practice of ̄nance in the 1980's and 1990's is largely noted for the proliferation of new contractual arrangements that package securities payo®s in di®erent ways. As Ross (1989) suggested in his presidential address to the American Finance Association, we still do not understand why ̄rms go through the trouble of creating such redundant assets and liabilities. Surely, such ̄nancial engineering is costly. The role of standard securities like debt and equity in the ̄nancing of real investments has been explored in a rapidly burgeoning literature on security design.1 However, only recently has research in ̄nancial economics begun to address why seemingly trivial packagings of securities are so popular. Allen and Gale (1988) show that when it is costly to issue securities and when di®erent groups of investors place di®erent values on the same security, optimal securities split up the ̄rm's state-contingent cash °ows, allocating all cash °ow in a given state to the investor who values it the most. Madan and Soubra (1991) introduce marketing costs into the Allen and Gale model and show that the sharing of cash °ow in several states may be optimal in the presence of marketing costs. Ross (1989) also explores the implications of marketing costs and shows that ̄nancial innovation can reduce the costs of marketing securities; Pesendorfer (1991) generalizes this result to a general equilibrium framework. Boot and Thakor (1993) argue that selling multiple ̄nancial claims partitions a ̄rm's total cash °ow into \informationally sensitive" and \informationally insensitive" components. This encourages the acquisition of information by investors, which enhances ̄rm revenue. DeMarzo and Du±e (1995) analyze the e®ect of adverse selection when issuing ̄rms possess superior information. In their model, the signal from the quantity issued generates a downward sloping convex demand curve for the security. They then show that the design of securities like Collaterized Mortgage Obligations allows intermediaries to retain the portion of the security's return for which adverse selection, due to private information, is greatest, thereby reducing the demand curve e®ect on revenues collected. We model a situation in which di®erent agents possess signals about di®erent components of a ̄rm's aggregate cash °ow and show that the joint pricing of the ̄rm's securities may reveal these signals to all of the agents.2 For example, suppose domestic banks receive See Allen and Winton (1994) for a recent survey of literature on security design. An example of this is also found in Kraus and Smith (1995).