Abstract:
Previous research on inferential expectations (IE) (Menzies and Zizzo, 2009) has only
considered a test statistic that is exogenous, based on time. This thesis examines the theory of
IE for a test statistic that is endogenously determined, and incorporates IE into the standard
cobweb model. Three applications are developed; an IE cobweb model nested in adaptive
expectations, IE employed to estimate the value of a new parameter, and an IE model which
generalises econometric learning. Under the latter, it is shown that belief conservatism results
in greater forecast errors, even in a model where equilibrium outcomes are dependent on
expectations.
