Palgrave Handbook of Econometrics: Applied Econometrics

12 Methodology of Empirical Econometric Modeling

(2008) emphasizes that the outcomes reported would change substantively if data
had been more carefully evaluated prior to the econometric analysis. To quote:

my concern (is) with the status, within economics, of economic statistics.

By “economic statistics,” I mean the study of how we create, use and assess

economic data – what one might call “data appreciation.”...It is true that

economists are using empirical data to an unprecedented extent, and applying

tools of great sophistication. Economics is a much more data-driven subject than

it was in the past. But, I shall argue, economists have too often come to take data

for granted, without critical examination of their strengths and weaknesses.

With that caveat about data firmly in mind, let us turn to methodology: mea-
surement is reconsidered in section 1.4.3, and an illustration of some effects of
substantial revisions in section 1.7.1.
The route ahead views all models as arising from reductions of whatever pro-
cess generated the data, which is a combination of the economic outcome and
the measurement system. We discuss these reductions in relation to their impact
on the parameters that actually governed the economic decisions of the relevant
agents. Most reductions occur implicitly, as investigators usually approach mod-
eling from the opposite perspective, namely what to include in their analysis,
although its success or failure will depend on whether the sub-set of variables
considered allows a model to capture the salient and constant characteristics of
the data-generating process (DGP). What to include and how to include it cer-
tainly depends on the economics behind the analysis; but what is found depends
on the unknown data-generating process and the losses of information from the
reductions that were necessary to derive the postulated model.

1.4 A theory of Applied Econometrics

“Why, sometimes I’ve believed as many as six impossible things before

breakfast.” (Quote from the White Queen in Lewis Carroll, 1899)

If only it were just six! To believe that he or she has ascertained the “truth,”
an applied econometrician would have to believe at least the following dozen
impossible (composite) assumptions:

1. a correct, complete, and immutable underlying economic theory derivation


2. a correct, comprehensive choice of all relevant variables, including all dynamic
   specifications


3. exact data measurements on every variable


4. the absence of any hidden dependencies, including collinearity and
   simultaneity


5. the validity and relevance of all conditioning variables (including instruments)


6. the precise functional forms for every variable


7. that all parameters of interest are identified in the resulting model specification