Doing Economics

résistance of any empirical talk, and so special care should be put
into making sure that the figures and slides are self-explanatory
and legible from the back of the average conference room. One
thing that many empirical economists do is to present empirical
results that omit control variables. Opinions are divided as to
whether that is a good thing or not. On the one hand, it makes it
easier to focus on the relationships of interest by not fitting
(almost) all of a regression’s coefficients and standard errors onto
one slide. On the other hand, the results for the control variables
can help diagnose specific problems with the results for the
relationships of interest. I suppose a middle-of-the-road approach
is to both (i) show slides that omit the control variables as part of
the main presentation and (ii) have detailed slides as part of the
appendix. The ideal scenario is one where there are so few
controls that everything can fit on one slide. Here, it is probably
best to show all of the core results as well as the main robustness
checks, but to keep the more offbeat robustness checks for the
appendix. This section is also where, before presenting any results
per se, the ancillary results necessary to make the case that the
identification strategy does what it purports to do are presented:

---

graphs of regression discontinuities, tests of the parallel trends
assumption, maps showing the rollout of a given program over
time, and so forth, should be featured early on in this section.
7. Summary and Concluding Remarks. It can seem like a waste of
time to have to summarize the paper you have just finished
presenting. And to some extent, it is tedious to do so, but the old
saw “tell ’em what you’re gonna tell ’em, tell ’em, and then tell
’em what you’ve told ’em,” however cliché, is nonetheless useful,
if only because some people may have gotten lost during your
presentation and this is a way to bring them back. Much like jazz
musicians playing a standard will typically play the melody, take
turns soloing over the chord progression, and then play the
melody once again to conclude, one should summarize the paper
—play the melody one last time—before offering some
concluding remarks.
8. Appendix. Not everyone realizes that, just like a paper can have
dozens of pages of appendix materials, a slide deck can also have
dozen of slides of appendix materials. Those materials tend to be
things that are included just for those curious audience members,
and are not meant to be presented in the course of a normal
seminar. Think of these appendix materials as you would the extra
features on a DVD: they consist of interesting bits of material
such as experimental protocols, tables of additional robustness
checks, extra test results, visual aids used during fieldwork, as
well as derivations and proofs of theoretical results. If you have
had the opportunity to give a talk often, you are likely to have
heard the same questions a number of times, so the appendix
slides can even contain answers to some of the most commonly
asked questions.
4
Once again, giving an invited seminar is not only a chance to publicize
one’s research findings, it is also a chance to engage in a dialogue with
other researchers about those findings and to get feedback which will
ultimately improve the paper. They should thus be approached as such
rather than as a monologue. You are there to talk with other researchers, and
not at them. Before moving on to job talks, I should discuss how to get

---

invited to give seminars. Obviously, reputation matters, and the more you
become known for your research, the more you will get to give invited
seminars. But reciprocity also matters, as does being entrepreneurial. On the
former, if you volunteer to organize seminars at your own institution, you
can invite people who may be in a position to invite you to give a seminar
at their institution later on. This obviously means that you should not only
invite superstar researchers from top departments, but also people from
comparable departments. On the latter, a colleague who is excellent at
networking suggested the following: if you know you will be in a given city
or region, you can get in touch with the departments in that city or region
and politely offer to give a talk. This will not always work, and your
success rate is likely to be strictly decreasing as the quality of the
department you offer this to increases, but it is a strategy whose expected
payoff is nonzero.