856 Macroeconometric Modeling for Policy
(2003) we have shown, by extensive testing of a previous version of our model,
that the Lucas critique has little force for our system of equations. This finding is
consistent with the international evidence presented in Ericsson and Irons (1995)
and Stanley (2000). On the basis of these results, our model is more consistent with
agents adopting robust forecasting rules, in line with the analysis and suggestions of
Hendry and Mizon (2000). In that case the Lucas critique does not apply, although
the degree of autonomy remains an issue that needs to be evaluated as fully as
possible, given the information available to us.
This chapter documents the approach we use to dynamic macroeconometric
model-building and policy analysis. To make the analysis applied, the approach
is illustrated through a model of the Norwegian economy. Our approach is, of
course, applicable to other economies, but we know more about market character-
istics, policy changes and institutional development in Norway than in any other
country or economic area. And since such factual knowledge is an indispensable
and complementary aid to formal econometrics in the building of an empirical
model, we prefer to work with the economy we have most knowledge about.
The rest of the chapter consists of three main sections. Section 17.2 sets out a
coherent approach to dynamic macroeconometric model building; section 17.3
demonstrates the approach through building and evaluating a small econometric
model; section 17.4 demonstrates various tools for policy analysis using the model.
Section 17.2 involves three steps in going from general to specific. The first step is
theoretical and establishes a framework for linearizing and discretizing an approx-
imation to a general theory model with constant steady-state values. The second
step is to estimate, and solve, the steady-state model in the form of overidentifying
cointegrating relationships and common trends. The third step is to identify and
estimate the dynamic structure of the model.
Section 17.3 illustrates the approach set out in section 17.2 by the construction
and evaluation of a small open-economy model.
Section 17.4 demonstrates five tools for policy using the model: tractability, sim-
ulations of policy responses, optimal policy considerations, theory evaluation, and
forecasting. The first use is illustrated by introducing a method to construct stylized
versions of complex models. The second use is illustrated by evaluating responses
to monetary policy shocks. The third use shows how important model specification
is for the derivation of optimal monetary policy. The fourth use is illustrated by
testing the New Keynesian Phillips curve. The final use evaluates possible sources
affecting forecast performance.
17.2 A modeling framework
As the values of all major economic variables are announced regularly, it is easy to
believe that a local approximation to a DGP can exist. It is an interesting philo-
sophical question whether the true generating mechanism can (ever) be completely
described, but the usefulness of the concept does not hinge on the answer to that
question. The main point is that once the real economic world, in its enormous,
ever-changing complexity, is accepted as a premise for macroeconomic modeling,