1 NONLINEAR
REGRESSIONS,
MODELS OF
EXPECTATIONS, AND
NON NORMALITY
What is in this Chapter?
In this chapter we first discuss nonlinear estimation methods: the Newton-Raphson
method, and the Gauss-Newton method for nonlinear least squares. We then discuss
grid-search methods and illustrate them with respect to partial adjustment models with
adaptive expectations.
We next discuss expectations models, and tests for rationality. The estimation of rational
expectations models is illustrated with a simple demand and supply model under rational
expectations. The problems of serial correlation in the errors in these models are also
outlined.
Finally, we consider the nonnormality of errors. We discuss tests for normality and
estimation methods under the assumption of nonnormality. We discuss Huber's robust
methods, least absolute residual (LAR) method, and some other alternatives to least
squares.
The methods described in this chapter can all be estimated using the SHAZAM
computer package.

-391-