Real expectations: replacing rational expectations with survey expectations in dynamic macro models

By Jeff Fuhrer

This paper examines the implications of changing the expectations assumption that is embedded in nearly all current macroeconomic models. The paper substitutes measured or “real” expectations for rational expectations in an array of standard macroeconomic relationships, as well as in a DSGE model. The author finds that the use of survey measures of expectations — for near-term inflation, long-term inflation, unemployment, and short-term interest rates — improves performance along a variety of dimensions. Survey expectations exhibit strong correlations to key macroeconomic variables. Those correlations may be given a structural interpretation in a DSGE context. Including survey expectations helps to identify key slope parameters in standard relationships, and eliminates the need for having lagged dependent variables in structural models that is often motivated by indexation for prices and habit formation for consumption. Including survey expectations also obviates the need for autocorrelated structural shocks in the key equations. In a head-to-head empirical test, the weight placed on the DSGE model’s rational expectations is essentially zero and the weight on survey expectations is one. The paper also discusses the modeling complications that arise once the rational expectations assumption is abandoned, and proposes methods for endogenizing survey expectations in a general equilibrium macro model.

Another recent paper that deviates from rational expectations with interesting results, here by taking actual survey expectations. The cost is that one cannot pinpoint expectations within the model, and inferring policy implications unconditionally on expectations becomes iffy. We would then need a model to determine the econometrician’s expectations of the agents’ expectations. Should they be rational?

One Response to Real expectations: replacing rational expectations with survey expectations in dynamic macro models

  1. Very interesting paper, taking steps in a very important direction.
    I wonder if the author tried performing a Granger Causality test to ensure that `measured expectations’ cause, but are not caused by, future state variables. It seems crucial that measured expectations pass such a test in order to be given a structural interpretation.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: