Writing and L
A
T
E
X Tips for Ph.D. Students
Alexis Akira Toda
∗
First version: 2016
Updated: August 9, 2020
Abstract
This note provides small tips for using L
A
T
E
X and writing in general
so that Ph.D. students can improve their papers.
1 L
A
T
E
X
Nowadays almost all working papers as well as published papers in all serious
economics journals are typeset using L
A
T
E
X. For people below a certain age,
writing an academic paper without using L
A
T
E
X will signal that your research is
not serious. If you do not know how to use this system, you have to learn it now.
A very good source of information is the L
A
T
E
X Wikibook.
1
If your operating
system is Windows, then MikT
E
X
2
will give you a no-brainer installation. I
don’t know about other operating systems. To use L
A
T
E
X, you need a text
editor. There are many of them, and you can compare them here.
3
Personally,
my favorite is TeXstudio.
4
What I like about this editor is that
1. there is text highlighting and auto completion (though the latter is some-
times annoying),
2. you can compile a document by just hitting the F5 key,
3. after compiling, the PDF displays the location you were editing (which is a
huge advantage because you don’t need to navigate through your lengthy
document to look for where you were editing!).
1.1 Small L
A
T
E
X tips
1. You have noticed that this document has footnotes and URLs with hy-
perlinks. You can do so by loading the hyperref package.
2. Always load the amsmath and amssymb packages. Otherwise you can’t
type symbols like % (\succsim).
∗
1
https://en.wikibooks.org/wiki/LaTeX
2
http://miktex.org/
3
https://en.wikipedia.org/wiki/Comparison_of_TeX_editors
4
http://texstudio.sourceforge.net/
1
3. Define macros for symbols that you often use. For example, instead of
writing R (\mathbb{R}) each time, define your command like
\newcommand{\R}{\mathbb{R}}
and write \R throughout.
4. When you accumulate many macros of your own, it’s simpler to create a
style file (.sty file). Mine is here.
5
5. As you read papers, create your own bibliography (.bib) file and use
BibT
E
X to cite papers using the \cite command, for example Arrow
and Debreu (1954). \citep will output (Arrow and Debreu, 1954), and
\citealp will output Arrow and Debreu, 1954. (Notice the differences in
the parentheses.)
6. Loading the doi package lets you insert a hyperlink with a digital object
identifier (DOI). (See the reference list below for an example.) Doing so
will benefit the readers since they only need to click the link to get the
reference.
7. You can customize the style of the reference list by loading an appropriate
bibliography style file (.bst file). Personally I just use plainnat, but you
need to load the natbib package to use this format. Furthermore, to cite
papers with names and years with parentheses, write
\usepackage[authoryear,round]{natbib}
in the preamble.
8. Some journals require specific margins. An easy way to accommodate is
to load the geometry package.
9. Sometimes you want to include NBER recession bars in your Matlab fig-
ures. This happened to me once, and I spent many hours figuring out how
to do properly. The problem is that, usually you save Matlab figures in
.eps format and include them in the L
A
T
E
X document, but EPS figures
are not transparent so the recession bars become quite ugly. On the other
hand, if you save the figure in .png format, the resolution of the figure
becomes low and it looks very unprofessional.
Here is how you can create a high quality figure with transparent recession
bars without blank spaces around the figure. First, use the recessionplot
command in Matlab to create the figure. Next, execute the following lines
after the figure:
fig = gcf;
fig.PaperPositionMode = ’auto’;
fig_pos = fig.PaperPosition;
fig.PaperSize = [fig_pos(3) fig_pos(4)];
print(fig,’filename’,’-dpdf’)
5
https://github.com/alexisakira/alexisakira.github.io/blob/master/files/test.
sty
2
This way a figure called filename.pdf will be saved. See Figure 1 for an
example.
10. Sometimes you want to provide detailed explanations under tables and
figures. To do this load the caption package and use the \caption*
command. To make the text font size smaller, you can use \footnotesize.
See Figure 1 for an example.
Year
1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Cyclical Component of Top Income Shares (%)
-2
-1
0
1
2
3
4
Top 1%
Top 1-10%
Figure 1: Cyclical components of top income shares.
Note: the figure shows the cyclical components of the U.S. top income shares obtained by
the HP filter with smoothing parameter 100. The shaded bars indicate the NBER recessions.
Source: a very old working paper version of Toda and Walsh (2020).
1.2 Common mistakes among novices
1. Don’t quote text like ”this”. The correct way is “this”, which is typed
‘‘this’’. Some text editors like TeXstudio will automatically replace
double quotes by the correct quotes in each language (English, French,
German, etc.). Click here for more details.
2. Be careful with the dots. a
1
× a
2
× . . . × a
N
(\ldots) and x
1
, x
2
, · · · , x
N
(\cdots) are incorrect. (a
1
×a
2
×· · · ×a
N
and x
1
, x
2
, . . . , x
N
are correct.)
Most of the time, using the \dots command will automatically choose the
correct dots style. Click here for more details.
3. When writing texts within an equation, use the \text command. Compare
“subject to” ($subject~to$) and “subject to” ($\text{subject to}$).
In L
A
T
E
X, MRS (marginal rate of substitution, $MRS$) is interpreted as
M × R × S. Write instead MRS ($\mathrm{MRS}$).
4. Do you see the difference between Prof. Arrow (Prof. Arrow) and Prof. Ar-
row (Prof.~Arrow)? In the former case, the dot is interpreted as a period
and the spacing is wider. In the latter case, the dot is interpreted as
abbreviation and also ~ prevents a line break between Prof and Arrow
3
because it is interpreted as one word (if you do not need to prevent line
breaks, write Prof.\ Arrow). Be careful when you spell U.S., U.K., Dr.,
etc.
2 Writing
Once you master how to write professional documents using L
A
T
E
X, the next
thing you need to learn is how to write well. Writing well is very important,
because referees get annoyed when the paper is poorly written, and even if the
paper is published, readers will be frustrated and you will not get as many
citations as you deserve. To learn how to write well, start from reading Strunk
and White’s “The Element of Style”,
6
Halmos (1970), and John Cochrane’s
“Writing Tips for Ph. D. Students”.
7
A paper consists of the title, abstract, introduction, and the rest. The title
is the shortest summary of the paper, the abstract is a short summary of the
paper, and the introduction should contain the longest summary of the paper
as well as a literature review.
2.1 Body
Start writing the paper from the body.
1. Think hard about what the main contributions are. You can have many
small results in the paper, but they should be connected by one big theme.
2. Since readers are impatient, you should get to the main results as soon
as possible. Write the paper in a logically linear order (i.e., stuff A must
precede stuff B if and only if A is necessary to understand B) as much as
possible. If this is not possible (say A is too long or technical), you can
either relegate A to the appendix or make the model less general to say
the same thing under stronger assumptions.
3. Whenever you have a proposition, try to explain the intuition.
4. Don’t call a model “general equilibrium” unless all endogenous variables
are endogenously pinned down. (Once a researcher from Minnesota pre-
sented a “general equilibrium” model with an exogenously fixed interest
rate, and Michael McGill politely pointed out that such a model is called
partial equilibrium.)
5. A proposition is a statement that can be proved to be true. A theorem is
an important proposition. A lemma is a small proposition that is used to
prove a theorem. A corollary is a proposition that is easily derived from a
theorem. (There are some exceptions: Zorn’s lemma and Ito’s lemma are
important theorems.) Some people claim “propositions” without actually
stating the precise assumptions or rigorously proving them (such things
are called “observations”, “claims”, “conjectures”, etc.): never do this.
6
The book is no longer copyrighted: you can find a free copy by Googling.
7
The link is here: https://faculty.chicagobooth.edu/john.cochrane/research/papers/
phd_paper_writing.pdf. Note that “Ph. D.” should be either “Ph.D.” or “Ph. D.”, right?
4
6. When you label objects such as x
1
, x
2
, . . . , label them as {x
n
}
N
n=1
, not as
{x
i
}
n
i=1
. That is, use a lower case letter to denote the general element,
and an upper case letter to denote the cardinality of the set. That way
you only need to introduce one alphabet, which makes easier to remember
what is what.
2.2 Introduction
The introduction is extremely important. Usually readers look at the title and
the abstract, and if interested, skim through the introduction. If the introduc-
tion is boring, readers stop reading. You need to write an introduction that
compels the reader to keep reading. Here are a few tips for writing the intro-
duction.
1. Don’t start the introduction from a literature review.
2. The reader should be able to grasp what the paper is about (i.e., intro-
ducing the problem you are going to address) by the end of the second
paragraph in the introduction.
3. Think about potential readers and referees and try to convince them to
keep reading.
4. Most of the time, the problem you are going to study is already studied by
previous researchers. Briefly explain what has been done in the literature,
what are unresolved questions, and why they are important.
5. After introducing the problem, explain the model and main results in some
detail.
6. After all of this, you can do a detailed literature review in a paragraph or
a subsection.
2.3 Title
Think hard about the title: it should be concise and informative. In general, the
shorter, the better. For example, “On the Existence of Equilibrium. . . ” can be
shortened to “Existence of Equilibrium. . . ”. “A Model of Income Distribution”
(Champernowne, 1953) is short but uninformative because it does not say any-
thing about what the paper does except that it’s about income distributions.
“Existence of an Equilibrium for a Competitive Economy” (Arrow and Debreu,
1954) is an excellent title, because it explains well what the paper is about and
it is as short as it can be. On the other hand, “On Equilibrium in Graham’s
Model of World Trade and Other Competitive Systems” (McKenzie, 1954) is a
poor title, because
1. we know that the paper is about some property of equilibrium but the title
does not say whether it’s about definition, existence, uniqueness, etc.,
2. nobody remembers what “Graham’s Model of World Trade” is (lesson:
use only well-established proper nouns like Pareto or Nash), and
3. the paper has actually little to do with world trade.
5
There is no wonder that Arrow and Debreu (1954) is cited about 7,000 times,
while McKenzie (1954) is cited only about 500 times, even though these two
papers are the first to prove the existence of equilibrium and published almost
simultaneously.
2.4 Conclusion
As Cochrane suggests, I don’t think a conclusion is necessary because you have
already explained what you do in the paper in the abstract, introduction, and
the body. When I write short papers, say 25 pages or less, I usually omit the
conclusion.
2.5 References
1. Use BibT
E
X to compile a professionally-looking reference list.
2. I hate to be strategic, but if your paper goes to the hands of the “wrong”
referees, the chance of acceptance will be quite low. Cite many papers
written by people who are likely to support you. If your paper is mostly
theoretical, don’t bother citing many quantitative papers (and vice versa).
Oftentimes theorists don’t care about calibrations, and quantitative people
don’t care about theorems. After your paper is accepted (or received
R&R), you can change the reference list to be scientifically honest.
3. People often find papers by searching for papers that cite important pa-
pers. To improve the discoverability of your work, cite important or recent
papers in your field.
2.6 Appendix
The appendix is quite useful. When you write papers, think of the appendix
as the temporary trash: if you find some part of the paper less important than
others (or can be deferred without spoiling the flow of the paper), make that
part an appendix. Examples are:
1. Unless you are a theorist, most people do not care so much about proofs
(I do care a lot, though). Put most proofs in the appendix. That way you
can make your paper appear shorter and satisfy the page limits imposed
by journals.
2. Referees always complain that you have not done enough robustness checks.
Do all robustness checks that you can think of, and dump them into the
appendix. After the paper is accepted, you can make a separate online
appendix.
3. If there are materials that is standard to you, but not for most people (like
the definition and properties of some power law distributions), instead of
just citing sources explain briefly in an appendix. The referees will find it
easier to understand the paper.
6
References
Kenneth J. Arrow and Gerard Debreu. Existence of an equilibrium for a compet-
itive economy. Econometrica, 22(3):265–290, July 1954. doi:10.2307/1907353.
David G. Champernowne. A model of income distribution. Economic Journal,
63(250):318–351, June 1953. doi:10.2307/2227127.
Paul R. Halmos. How to write mathematics. L’Enseignement Math´ematique,
16:123–152, 1970. doi:10.5169/seals-43857.
Lionel W. McKenzie. On equilibrium in Graham’s model of world trade
and other competitive systems. Econometrica, 22(2):147–161, April 1954.
doi:10.2307/1907539.
Alexis Akira Toda and Kieran James Walsh. The equity premium and the
one percent. Review of Financial Studies, 33(8):3583–3623, August 2020.
doi:10.1093/rfs/hhz121.
7
