Abstract: This paper contains an explanation for the non-structural comovement of asset prices using a model for insider trading. In this model, the price comovement appears solely through an informational channel. When rational investors select portfolios accessing noisy inside information, the market prices are a noisy spread of the inside information. When an investor faces a limited information-processing capacity, the expected profits derived from an information structure that focuses on reducing the aggregate external noise may be higher than those generated by an information structure aiming to deliver separate information about each market. There is a trade-off between statistical independence in information and aggregate precision. As a result, when the insider chooses an information structure that delivers correlated information, his portfolio will be negatively correlated. The negative correlation appears as a form of hedging in response to the correlated information. Such a correlation vanishes as the learning capacity increases and its magnitude is scaled down when the insiders exhibit higher degrees of risk aversion.
JEL codes: D81, D83, G11, G14
Key words: market microstructure, insider trading, rational inattention, entropy learning, price comovement.
This version: September 2020
Abstract: In this paper, I study how artificial price correlation across financial markets arises as a rational response to limited capacity for information processing. The model I use for the analysis involves three types of agents: investors with private information, uninformed investors, and market makers, trading two generic assets under two possible protocols. In the first protocol, the market makers act as specialists, a a specialist can only trade one type of asset; in the second one, the market makers trade both assets. The market makers compete à la Bertrand to set each asset's price. In equilibrium, the market makers set the prices equal to the assets' conditional expected payoffs, given the observed aggregate trades (order flow). By doing so, a market maker faces an adverse selection of informed traders, since she cannot screen investor profiles. Investors can either buy or sell a single unit of each asset. Uninformed investors, or noise traders, place random orders. The informed-type investors, or insiders, are allowed to build their information structure in the form of signals, provided their limited processing capacity. Insiders facing a low capacity, optimally choose a unidimensional information structure—only one signal containing relevant information for predicting both asset payoffs. As a result, the market orders are functions of the same signal. Since the market orders are the only source of relevant information on each market, the information available to the market makers has a common source across markets. Thus, the information received by the market makers through the order flow has a common origin, causing a structural relation between asset prices. I set the asset payoffs to be statistically independent to avoid other potential sources of correlation. Moreover, when the market makers can trade both assets, there is an additional source of comovement. The market makers cross-subsidize between markets when they set prices, to compensate for the adverse selection of insiders.
JEL codes: D81, D83, G11, G14
Key words: bid-ask spread, market microstructure, insider trading, rational inattention, entropy learning, price comovement.
This version: September 2020
Draft available upon request.
Finance Research Letters Vol. 52. March 2023. (click here)
IB DANE Vol 3, No. 1. With Jaime Tenjo, Martha Misas and Alejandro Gaviria. (click here)