Date Posted: 2001
Full text (most recent) on SSRN
According to the general wisdom in the finance literature, the bid-ask spread on a security is viewed as a form of transaction cost faced by security traders. In both floor and screen markets, traders are willing to pay this transaction cost in order to obtain immediacy in the execution of their transaction. The dealer sets his ask and bid prices on the basis of the expected conditions of the market and the marketability of the specific security, thus providing immediate liquidity to potential buyers and sellers. Dealers generally set the ask and bid prices around the actual price, expecting profits from the random orders of buyers and sellers.
Given the specific function of the dealers, the theoretical finance literature has identified three main determinants of the bid-ask spread in a competitive market. Namely, the spread represents a measure of the price of the dealer's services given (a) the operational efficiency of the trading system (operational efficiency formulation); (b) the degree of a security's marketability (inventory hypothesis); and (c) the presence of private information of the traders (adverse information hypothesis).
This comment considers the impact of different methods of trading (and different levels of anonymity in trading) on the operating costs and adverse selection costs. The paper presents an interesting case of bilateral adverse selection, analyzing it from a theoretical perspective.
(A) Traders' Adverse Selection. The first form of adverse selection depends on the traders' use of superior private information. The effects of this form of adverse selection are not different from those of the lemons problem, a well-known asymmetric information problem considered by Akerlof (1970). Costs are imposed on dealers who react by adjusting the bid-ask spreads in the trading of the stock. This asymmetric information problem is related to the level of private information which is likely to be present. Likewise, the adverse selection problem increases with the percentage of traders that operate on the basis of superior private information.
By assumption, dealers have greater ability to identify and discriminate between different types of traders in the floor market. Anonymity enhances the ability of traders to take advantage of private information that they possess, since dealers cannot differentiate between the two types of traders and cannot penalize information-driven traders in subsequent dealings. Floor trading attenuates these problems and thus dealers can afford to keep lower spreads to the benefit of the mass of liquidity-motivated traders.
(B) Dealers' Adverse Selection. The presence of parallel screen and floor markets creates an opportunity for dealers to react to the traders' adverse information problem in ways that are different from the usual edging of potential losses with larger spreads.
Dealers can engage in adverse selection on their part by choosing different spread levels in the two markets and consequently changing the relative market share of each stock in the two trading systems. Dealers thus respond to the traders' adverse selection by engaging in selective fixing of spread levels and, as a result, induce different choices of trading systems.
In first analysis, the presence of bilateral adverse selection strategies would suggest the emergence of a separating equilibrium: information-motivated traders would generally prefer to trade in anonymous screen markets at higher spread costs, while liquidity-motivated traders would be less willing to pay the spread premium, since they draw no immediate financial benefit from the anonymous form of trading. This separating equilibrium, however, would likely encounter further adjustments. As a result of the initial matching of trader-type and market form, the density of information-driven traders in the two markets would change, rendering further increases of the bid-ask spread necessary. This will induce some of the information-driven traders to drop out of the screen market, since the higher spread costs are no longer justified by the limited private information that they possess. Only high private information traders would thus be willing to pay the increased spread costs in order to ensure anonymity. In a frictionless world, this bilateral adverse selection problem would unravel until all private information traders would drop out of the screen market entirely to avoid the yet higher spread costs imposed by dealers on anonymous traders.
In a nutshell, the static adjustments would yield a selection of traders on the basis of their "information level" (or trading motivation). The result is a separating equilibrium where different types of traders will choose different markets according to whether the expected profit from the use of private information justifies the higher premium (i.e., higher bid-ask spread). Alternatively, dynamic adjustments would yield an unraveling such that, in equilibrium, only one market would be utilized for each type of stock.
In a world of perfect mobility, however, the bilateral adverse selection dynamic described above would naturally lead to corner solutions where each market would ultimately dominate for the trade of a specific category of stock. Interestingly, the presence of bilateral adverse selection problems has partially offsetting effects, so as to minimize the total social cost of trading.