Doing Economics

2.2 Theoretical Framework
Since the goal of empirical economics is generally to answer questions of
the form “Does x cause y?” or “If x increases by one unit, how many units
does y change by?,” most applied work in economics begins with a
theoretical hypothesis about the relationship between x and y.
The best research articles tend to focus on a single question (e.g., “What
is the impact of having a land title on agricultural productivity?”) or on the
mechanisms behind a given question (e.g., “If land titles improve
agricultural productivity, do they do so because land titles allow landowners
to use their land as collateral?”) Thus, a first decision has to be made about
what empirical relationship of interest a given article will focus on. In other
words, the best empirical articles tend to be rather narrowly focused on a
single question, and so you will almost inevitably have to leave some
material on the cutting room floor.
2
The question, then, is how to take your hypothesis and convert it into a
proper theoretical framework for an empirical economics article. Here,
there are two possible scenarios: (i) you are investigating a question which
has already been studied by theorists, or (ii) you are investigating a question
which has not already been studied by theorists.

---

In the first scenario, there are two options. The first option is to include a
theoretical framework in the article by incorporating or adapting somebody
else’s theoretical framework lock, stock, and barrel. Though it often feels
like a research paper has to innovate on all fronts, that is not the case. For
applied papers in particular, what matters is that the research question, the
empirical strategy, or both be novel. In most cases, it is fine to use someone
else’s theoretical framework—provided that you clearly state that you are
doing so and cite the source of your theoretical framework.
A closely related option is to adapt somebody else’s theoretical
framework to suit your needs—say, by incorporating an additional variable,
or by relaxing or making additional assumptions to suit the needs of your
application. In the second scenario, when you are investigating a question
for which the theory of change has not already been studied by theorists,
you have to clearly state the theoretical reasons behind your hypothesis. In
some cases, this may require a formal theoretical model. In other cases, it is
enough to merely present a verbal conceptual framework.
3
 In all cases, your
theoretical framework—be it mathematical or verbal—should start from the
primitives and make the necessary assumptions to generate the result “x
causes y through mechanism m”; no more and no less.
Generally when writing papers that combine theory and empirics, you
should make sure that your empirics actually test the testable predictions of
your theory. A prediction that is not tested empirically should not be
included in your theoretical framework, and your hypothesis tests should be
grounded in your theoretical framework. In other words, avoid any
disconnect between your theory and your empirics.
One could write an entire book on how to write economic theory (and
some have; see Thomson 2011), so nothing more will be said on this topic
save for the following: writing theoretical models in economics is an art
form, and if you have not learned how to do it in graduate school, it is
perhaps best to work with someone who has as a coauthor. When it comes
to publishing an applied economics article, better an informal, chatty
conceptual framework than a bad formal theoretical model.
That said, even if your working paper includes an elegant theoretical
model, it will sometimes happen that you will be asked by reviewers or by
an editor to get rid of your theoretical model before your paper can be
published, or to put said theoretical model in an appendix. If that happens,

---

know that this is not uncommon. In that sense, having a theoretical
framework in your paper often only serves as a signal that you know what
you are doing. This is especially true for job-market papers, which are used
to show the breadth of their author’s skills in addition to making a
contribution to research.