ABSTRACT: This paper studies
adaptive learning with multiple models. An agent operating in a
self-referential
environment is award of potential model
misspecification, and tries to detect it, in real-time, using an
econometric
specification test.
If the current model passes the test, it is used to construct an optimal policy. If it fails the test, a new
model is selected from a fixed set of models. As the rate of
coefficient updating decreases, one model becomes
dominant, and is used almost always. Dominant models can be
characterized using the tools of large deviations theory.
The analysis is applied to Sargent's (1999) Phillips Curve model.
Heterogeneous Beliefs and Tests of Present Value Models
(with Todd Walker and Charles Whiteman)
ABSTRACT: This paper develops a
dynamic asset pricing model with persistent heterogeneous beliefs. The
model
features competitive traders who receive
idiosyncratic signals of an underlying fundamentals process. We adapt
Futia's (1981) frequency domain methods
to derive conditions on the fundamentals that guarantee noninvertibility
of the
mapping between market data and the
underlying shocks to agents' information sets. When these conditions are
satisfied,
agents must 'forecast the forecasts
of others'. The paper provides an explicit analytical characterization
of the resulting
higher-order belief dynamics. These
additional dynamics can account for observed violations of variance bounds, predictability
of excess returns, and rejections of cross-equation restrictions.
(slides)