4 Methodology of Empirical Econometric Modeling
1.4.3 Data exactitude 30
1.4.4 Hidden dependencies 30
1.4.5 Conditioning variables 31
1.4.5.1 Weak exogeneity 32
1.4.5.2 Super exogeneity and structural breaks 34
1.4.5.3 Weak exogeneity and economic theory 35
1.4.6 Functional form 36
1.4.7 Identification 36
1.4.8 Parameter constancy 37
1.4.9 “Independent” homoskedastic errors 38
1.4.10 Expectations formation 38
1.4.11 Estimation 40
1.5 Model selection 40
1.5.1 Automatic model selection 41
1.5.2 Costs of inference and costs of search 44
1.6 Teaching “Applied Econometrics” 45
1.7 Revisiting the “experiment in applied econometrics” 47
1.7.1 An update 50
1.8 Automatic modeling of a VAR 4 ( 25 ) 54
1.9 Conclusion 56
1.1 Introduction
“Now, here, you see, it takes all the running you can do, to keep in the
same place. If you want to get somewhere else, you must run at least
twice as fast as that!” (Quote from the Red Queen inThrough the Looking-
Glass and What Alice Found There, Lewis Carroll, Macmillan & Co., 1899,
[henceforth cited as “Lewis Carroll, 1899”])Most econometricians feel a bit like Alice did at having to run fast even to stand still.
Handbooks are an attempt to alleviate the problem that our discipline moves for-
ward rapidly, andinfoglutcan overwhelm, albeit that one has to run even faster for
a short period to also find time to read and digest their contents. That will require
some sprinting here, given that the contents of thisHandbook of Econometricspro-
vide up-to-date coverage of a vast range of material: time series, cross-sections,
panels, and spatial; methodology and philosophy; estimation – parametric and
nonparametric – testing, modeling, forecasting and policy; macro, micro, finance,
growth and development; and computing – although I do not seeteaching. Such
general headings cross-categorize “Applied Econometrics” by types of data and
their problems on the one hand – time series, cross-sections, panels, high frequency
(see, e.g., Barndorff-Nielsen and Shephard, 2007), limited dependent variables (see,
e.g., Heckman, 1976), or count data (excellently surveyed by Cameron and Trivedi,
1998), etc. – and by activities on the other (modeling, theory calibration, theory
testing, policy analysis, forecasting, etc.). The editors considered that I had written
on sufficiently many of these topics during my career to “overview” the volume,
without also noting how markedly all of them had changed over that time. The