Doing Economics

2. In your application, you cluster your standard errors at the
community level because that is the level at which households
were randomly selected. But following Abadie et al. (2017),
you should cluster your standard errors at the household level
since that is the level at which randomization took place.
Response: You are entirely right, and we now cluster our standard
errors at the household. This does not change our results.
3. Instead of interacting your treatment variable with the
gender of the individual who is getting treated, I would rather
you present two sets of estimation results, i.e., one for women,
and one for men.
Response: Done. You can find the results by gender for what used
to be Table 4 in what are now Tables 4a and 4b.
As with everything else, clarity of communication is key. This means you
should never assume that your reviewers will understand why you did
something, and for each comment, tell them how you addressed their
comment, and show them. Do not just write “Done” or “See page 14.” in
response to a reviewer’s comment. Write “Done. On page 14, we now have
added the following paragraph in response to this comment,” and quote the
full paragraph. That way (and in the spirit of what I said in chapter 2), you
are writing so as to not be read: writing your response, that is, so that the
reviewers hopefully do not feel the need to look at your paper too closely
this time around.
When preparing responses to the reviewers, it is not uncommon for each
such reviewer-response document to be ten pages or more if you reproduce
each reviewer comment and copy the relevant sentences, paragraphs, and
tables from your paper, or if you show additional results that the reviewer
wanted to see, but which did not go in the paper.
That said, there is an optimal amount of work to do in response to
reviewer comments. You want to do what they ask you to do, and not less,
because then it might make the reviewers suspicious that you are hiding

---

something, and thus more likely to recommend a rejection. You also want to
do no more than what they ask to avoid wasting their time having to read
things they did not ask for.
22
If any of the reviewers’ comments are unclear, you should still do your
best to address them. This obviously means reading and re-reading that
comment to try to understand it, but it also means having colleagues have a
look at those comments. Very often, you might be coming at a problem a
certain way, but a reviewer might be coming at it a different way, and the
same words (e.g., “identification,” or “exogenous”) might take on different
meaning. For instance, someone who is used to doing causal inference
using reduced-form methods is unlikely to have the same meaning of
“identification” as someone who does time-series econometrics.
If you disagree with any of the reviewers’ comments, it is still a good
idea to do what they are asking for, and to keep your disagreement for your
cover letter to the editor, which I discuss below. When it comes to refusing
to do what reviewers ask, a simple rule of thumb is this: if you must do so,
do so no more than once on any given paper, and make sure you have good
reasons behind your refusal. I recently handled a manuscript where the
reviewers, quite reasonably, asked the authors to present versions of their
regression results where the outcome variables were in logarithms instead
of in levels. The authors refused to do so, telling me that “papers are getting
too long these days.” After fighting my inclination to reject the paper
because the authors were not being reasonable (the paper did contain
interesting findings, after all), I sent the manuscript back to the authors,
explaining to them that this was not a negotiation, and that as long as they
did not address that comment, the paper would not move closer to
publication.
If reviewer comments are contradictory, once again, you should do your
best to address them. Very often, this will mean picking whichever one you
feel is the best of the two comments, and telling the reviewer whose
comments you did not incorporate about that contradiction (and really, do
not invoke this defense without it being true, because reviewers can almost
always see each other’s comments), but leaving open the possibility to
change the manuscript if they can find a way forward by writing something
like: “Reviewer 3 asked us to do something that is in direct contradiction
with this comment, and we felt their comment made for a stronger

---

manuscript. If you can provide a suggestion for how we might be able to
accommodate both your comment and theirs, we would be happy to do so
in the next version of the manuscript.” Likewise with comments that are
rather vague, like “The authors are missing some key citations to the
literature in their review of the relevant literature” (in my experience, this is
usually brought up by reviewers who are upset that you did not cite more of
their work). Here, it is perfectly fair game to tell the reviewer that you
would be delighted to cite the right papers—if they tell you which papers
they would like you to cite.
Once you have addressed all of the reviewers’ comments to the best of
your ability, it is time to write a cover letter to the editor explaining how
you have addressed those comments as well as the editor’s own comments,
if she made any. Your cover letter should start by telling the editor what you
have done in response to her comments. That part of your cover letter
should follow the structure of a response to a reviewer, as outlined above.
Then, you should discuss what you have done to address each of the
reviewers’ major comments in broad strokes. That is, you should tell the
editor how you have addressed each major comment, rearranged the
manuscript, whether and how your qualitative findings have changed, and
so forth, in one or two sentences. If the editor wants to know the details of
how you tackled a specific point, she can look at your responses to the
reviewers. Generally, that letter will not be shared with the reviewers, so if
you feel that a reviewer was particularly obtuse, this is the place to mention
it.
After submitting the revised version of your article, the wait for an
editorial decision begins once more. In the best-case scenario, the editor
chooses not to send your revised article back to the reviewers, and accepts it
after having had a look at it and at your responses to the reviewer
comments. The second-best scenario occurs when the editor accepts (or
conditionally accepts, i.e., accepts subject to minor revisions) your article
after sending it back to the reviewers. The third-best scenario occurs when
you get another R&R decision, as it is not uncommon for articles to go
through additional rounds of revision after the initial one before it gets
accepted.
23
Obviously, the worst-case scenario is for your article to be
rejected after R&R—a situation that is increasingly likely the closer you get
to the top of the journal rankings.

---