Asset Prices Under Asymmetric Information

Christian Haefke - University of California, San Diego and Leopold Soegner - Vienna University of Economics and Business Administration

Ever since the CAPM financial economists have tried to come up with models that could explain the facts observed on capital markets. Single period asset pricing originated in the work of Lintner on the basis of Markovitz, and Tobin. Extensions to intertemporal settings are both in discrete-time and in continuous-time settings (Samuelson and Merton ). Merton linked asset pricing to agents maximizing their lifetime utility and found that "this behavior implies a type of intertemporal consumption smoothing [which] reflects an attempt to minimize the (unanticipated) variability in consumption over time".

The multi-beta asset pricing model can be rewritten so that the full model can be expressed as a market price for risk times the sensitivity of return to percentage movements in aggregate real consumption (consumption beta). Lucas investigates asset prices in an exchange economy in discrete time, Brock extends the scope of the analysis to a production economy. These models were the first of many real business cycle (RBC) models which take stochastic fluctuations in productivity as the predominant source of fluctuations in economic activity. Usually RBC models assume a representative agent and firm, thus avoiding aggregation issues (Stadler). These agents optimize explicit objective functions, i.e. utility and profit functions, respectively. Agents have rational expectations, markets are complete and always clear. In order to generate cycles from the stochastic productivity shock, different propagation mechanisms are applied. Agents seek to smooth consumption over time, individuals tend to substitute leisure intertemporally in response to transitory wage changes, and lagged investment influences investment in the future. Mehra and Prescott documented the failure of standard business cycle models to explain the risk premium and since then uncountably many solutions have been proposed. The key point of interest is to decouple consumption from asset returns.

Abel and Constantinides consider habit formation, propose agent heterogeneity via the consumption of stockholders and non-stockholders. Finally He and Modest find that a combination of trading frictions yields asset returns in accordance with the Hansen-Jagannathan bounds and hence their approach may be regarded as a potential solution to the equity premium puzzle. However, all these approaches are only partial equilibria as they completely leave the production sector aside.

Building on the works of Brock and Danthine and Donaldson introduce risk sharing between workers and shareholders and secondly adjustment costs which influence the substitution between capital and labor. In addition to these changes Rouwenhorst calls for the use of richer models with multiple sources of uncertainty or imperfect information.

In this paper we extend the standard real business cycle approach to asset pricing by introducing informational asymmetries between two different types of agents. To this end we propose a framework of asymmetric information where obtaining exact knowledge about a firm's fundamentals is only possible for one group of agents. The other group receives a much noisier signal. The price for the risky asset in this economy can be expressed as linear combination of both investors' expectations and collapses to the standard asset price of the Lucas model if both agents have the same information set. The same is true for different information sets if agents are risk-neutral. In general the presence of the noise in one agent's information increases the variance of the stock-price.

A simulation of the economy indicates that this is sufficient to match most empirical stylized facts. Especially the risk premium increases considerably.

Scheduled for Session 5.2 Asset Markets