Doing Economics

participate?” The more clearly this is stated, the better, because
fewer are the occasions for the reader to be disappointed. This
should be one paragraph long.
• Antecedents. After stating your research question, it is time to
relate it and what you are doing to the existing literature. Here,
relate your work to the five to ten closest studies (the closer to

---

five, the better) in the literature. What the relevant literature—the
antecedents—is will obviously depend on the question at hand. If
you are lucky enough to work in a literature that has seen a lot of
activity, you may have a hard time narrowing it down, and you
will need to judiciously pick the five to ten closest studies. If you
are working on a problem that no one has really looked at, or that
no one has looked at in a long time, you might have to go back in
time a bit further or expand your parameters for what counts as
antecedents. Here, what counts is to tell a bit of a story; no one
wants to read a bland enumeration of studies: “Johnson (2011)
found this. Wang (2012) found that. Kim (2013) found something
else. Patel (2015) found something else altogether.” For every
topic, the intellectual history of that topic can be told in an
interesting way.
• Value Added. This is where you need to shine. What is your
contribution? How does your paper change people’s priors about
your topic? Ideally, your paper will have three contributions. For
instance, you may be improving on the internal validity front for
the question you are looking at by having a better identification
strategy. You may also be improving on the external validity front
by having data that cover a broader swath of the real world, or
you may be performing a mediation analysis that allows
identifying what mechanism m the treatment variable D operates
through in causing changes in y. Finally, you may also be
bringing a small methodological improvement to the table. This is
not necessary, as even papers with fewer than three contributions
deserve to be published, provided at least one of their
contributions is important enough.
• Roadmap. Finally, you should provide your reader with a roadmap
to your paper. This section usually starts with “The remainder of
this article is organized as follows,” and it lists sections and what
they do in order. So for a typical paper, it would go: “The
remainder of this paper is organized as follows. Section 2 presents
the theoretical framework used to study the research question and
derives this paper’s core testable prediction. In section 3, the
empirical framework is presented, first by discussing the
estimation strategy, and then by discussing the identification

---

strategy. Section 4 presents the data and discusses some summary
statistics. In section 5, the empirical results are presented and
discussed, followed by a battery of robustness checks and a
discussion of the limitations of the results. Section 6 concludes
with policy recommendations and suggestions for future
research.” I have seen some economists on social media state that
they have had papers rejected for many reasons, but never for
want of a roadmap section. Fair enough. In most cases, however,
it is simply easier to include such a roadmap section and delete it
at a reviewer’s request than to not have one and have to write one
when asked to revise and resubmit a paper, not to mention the fact
that some readers will simply expect there to be a roadmap, since
the majority of applied economics articles include them. Anything
that signals that you know what the unspoken rules and norms of
the profession are is a good thing for your article’s chances of
getting published.
It is best to start writing a paper’s introduction as soon as there are some
empirical results. After the title and the abstract, the introduction is where
most people will decide whether (i) they think your work is interesting
enough to keep on reading, and (ii) whether they think your work is of a
good enough quality for them to believe your findings. I would guess that
the fate of at least 75 percent of articles—whether they get sent out for
review, or whether a revision is solicited by the journal when they do get
sent out for review—is driven by the introduction. As such, the introduction
should be rewritten every time the file is worked on by any of the authors. I
would guess that, for most of my papers, I have gone over the introduction
at least a few hundred times.
A good introduction works because it sets your readers’ expectations just
right. If there is one thing that will make a reviewer recommend a rejection,
it is a bait-and-switch (i.e., when an introduction overpromises and the rest
of the paper underdelivers), or when an introduction is unclear as to what
the paper does and how it does it.
As mentioned earlier in this chapter, a busy reader will typically read: (i)
your title, (ii) your abstract, (iii) your introduction, then skip to (iv) your
tables of results, then read (v) your conclusion, going back to the other

---

sections if and only if they have questions about what you are doing, or how
you do it. A good introduction minimizes (or eliminates altogether) a
reader’s need to flip through the paper in search of answers to her questions.
In addition to Keith Head, Claudia Sahm and David Evans both provide
advice on how to structure the introduction of an economics article. In her
advice to job-market candidates, Sahm (2019) suggests the following order
of sections for papers in macroeconomics:
1. Motivation
2. Research question
3. Main contribution
4. Method
5. Findings
6. Robustness checks
7. Roadmap
Though her suggested structure is close to Keith Head’s, it is worth
noting that Sahm suggests a more precise content for Head’s “value added”
section, namely, you should tell your reader what empirical methodology
you use to answer your research question as well as how you make sure
your results are robust.
Similarly, after analyzing the introduction of “the most recent empirical
microeconomic development papers from a range of top journals,” Evans
(2020) suggests the following structure, as well as a suggested length for
each “section” of the introduction:
1. Motivations (1–2 paragraphs)
2. Research question (1 paragraph)
3. Empirical approach (1 paragraph)
4. Results (3–4 paragraphs)
5. Value added (1–3 paragraphs)
6. Robustness checks, policy relevance, limitations (optional)
7. Roadmap (1 paragraph)

---

Although Evans’ advice stems from analyzing empirical development
articles, it broadly overlaps with the advice given by both Keith Head and
Claudia Sahm, and there is a payoff to adopting a common structure for
your introduction: McCannon (2019) analyzed the papers published in the
American Economic Review from 2000 to 2009 by looking at their
readability score, and found that the papers that were hardest to read
suffered a statistically significant decrease in their citation count of 0.20
standard deviation.