Premium
Testing Macro Models for Policy Use—An Insurrection in Applied Modelling
Author(s) -
Minford Patrick
Publication year - 2016
Publication title -
the manchester school
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.361
H-Index - 42
eISSN - 1467-9957
pISSN - 1463-6786
DOI - 10.1111/manc.12164
Subject(s) - indirect inference , inference , macro , robustness (evolution) , wald test , econometrics , monte carlo method , computer science , economics , statistical hypothesis testing , statistics , mathematics , artificial intelligence , estimator , biochemistry , chemistry , gene , programming language
I describe a new departure in classical testing methods based on Indirect Inference. I argue that it gives policymakers, anxious to know if their models give reliable policy conclusions, a way to find out. I discuss how using Monte Carlo experiments my co‐authors and I have found that in the small samples typically available in macroeconomic modelling, the Indirect Inference Wald, IIW, test has considerably more power than the popular direct inference test using the Likelihood Ratio, LR. This is both because the LR is applied after re‐estimation of the model error processes and because the IIW test uses the false model's own restricted distribution for the auxiliary model's coefficients. This greater power allows users to focus this test more narrowly on features of interest, trading off power against tractability. If they can find a model version that is not rejected by the test, they can then discover the robustness of their model results to the parameter variations that might also have passed the test.