Ivo Welch, 2021

Style Guide Accepted Papers

Reader Attention is scarce. Our prime objective is to make your article more eye-catching.

Use recent issues of the CFR for guidance on how articles should look like.

Use common sense. For example, how many columns can we reasonably typeset in a readable font? What is a good figure aspect ratio?

The two most important style instructions are:

  1. Interpretation: One unusual aspect of the CFR (compared to the JF or RFS) is that we want most exhibits (tables and figures) to have a "description" section followed by one "interpretation" sentence (or two). What should an interested reader looking primarily (only) at this exhibit learn from and remember about it? Ideally, description and interpretations alone are enough to understand your paper without having to read text.
  2. Paper Title: Use a title that makes it obvious what your paper finds. Do not use a title like "The Effects of Dividend Taxes on Profits." Instead, use "Dividend Taxes Have No Effect on Profits." Or "The Welch (2004) Capital Structure and Stock Returns Effect Disappeared in 2010." Or "The Welch (2004) Effect Disappears With Control for Firm Size." Or "The Welch (2004) Capital Structure Effect is Spurious." Your best weapon in the fight for attention is to be specific and to burn immediately what you find into the brains of anyone who is googling for the subject matter."

The remaining style suggestions are more about simpler details.

  1. Papers must include a full date (and abstract, of course).
  2. We use standard U.S. English. Neither British English nor Australian English, mate. Most importantly, this means that "end quotes" follow punctuation marks, and do not precede it. We do not follow the University of Chicago dictionary or style slavishly, but it usually tells us nicely what a reasonable approach is.
  3. The first page of your article (once typeset) should fit title, authors, abstracts, keywords, JEL codes, a thank-you note, and some publication/copyright info. If your abstract is too long, please rewrite it. (At beyond 150 words, it is getting ridiculous, anyway. A 50-word abstract is often more memorable than a 100-word abstract.) We want your names, email addresses, and institutions on the front page, one author each on one line. If an author does not fit, should we really remove him/her?
  4. Articles start without a section title called "Introduction."
  5. A negative number uses an en-dash, which is longer than a hyphen and shorter than an emdash. Make sure that negative numbers in your tables use endashes and not hyphens.
    • Depending on your browser font, you may be able to see the difference here: OK: −1 vs. BAD: -1 or -1. It's especially jarring when you also use +1, too, because you do want the length of the sign to be the same for positive and negative prefixes. An emdash like —1 is also not right.
    • If you use LaTeX, know that LaTeX assumes that a minus inside math-mode means negative (and thus it automatically uses an en-dash). If you use text-mode, you can either switch your negative number into math mode or use two dashes: not '-5' is negative, but '$-5$' is negative or '--5' is negative.
  6. Before equations, you do not need to write something like: "Our comparative statics are as follows: ...". Instead, omit the empty "as follows:" and write "Our comparative statics are ...".
  7. Most equations are not preceded by colons. Usually colons can be omitted.
  8. The phrase "is given by" can almost always be omitted.
  9. Use "because" instead of "since" if you mean logic. Use "since" if you mean time-precedence.
  10. Trivially short footnotes (such as, "see Fama-French:1993 and Black-Jensen-Scholes:1972") should go into the text. Super-long footnotes should be avoided. Normal footnotes are between 10 and 50 words long, and there should not be more than one per page on average.
  11. Spell out the word "to" instead of using ranges, when possible. What does "the coefficient is &negative;0.23&negative;0.05" mean? Does it mean &negative;0.18, or &negative;0.23 to 0.05, or even &negative;0.23 to &negative;0.05?
  12. Please think about how many digits after the decimal point you need the reader to know. When do you need 5.71249%, when is 5.7% enough? T-statistics should never have more than two digits after the decimal point. R2 rarely need any digits after the decimal point. etc.
  13. Let's try to use capital roman "T" for the statistical T distribution. That is, let's not lowercase or italicize "t" in Student-T distribution.
  14. If there are standard abbreviations in the literature already, please try not to innovate but to follow them. Readers really don't need to be confused any further. For example, t is usually the best variable to use for a time index, w a good name for a weight, and so on. If you have many variables, please provide a glossary table. If they are empirical variables, this table can also list data source and availability in a consistent format. If they are model variable, provide the domain.
  15. Every data table needs to have clear numbers of observations (in panels, firm-years), sample period, and (if it is a regression) an R2.
  16. We use commas for visual separation in thousands, millions, etc.
  17. Don't use & as an abbreviation for "and", unless customary or necessary (e.g., S&P500).
  18. Equations cannot contain footnotes. If you need to remark something, do so just before the equation.
  19. If you use emdashes—such as here—there is no space surrounding the quasi-parenthetical expression. However, you can often rewrite such sentences to be better without quasi-interruption.
  20. It is rare that you need a dash (—) to continue a sentence. Often, a period or colon is better.

Use sensible tenses. Many decades ago, some finance journals requested writing in the present tense, but the instructions were ambiguous. What they really meant was to write not about history in the present tense but about the analysis in the present tense: "The regression of x on y yields blah blah blah." They never meant to write about the historical events themselves in the present tense: "The impact of x on y was positive from 1980-1990" is correct, not "The impact of x on y is positive from 1980-1990." The JF now writes "Text must be inliterary present tense throughout. For example, "we predict the dependent variable" rather than"we predicted the dependent variable." Use past tense when describing historical events. For example, "investors sold shares in our sample" rather than "investors sell shares in our sample." Hallelujah!

For the CFR, we do not even care whether you use present (literary) or past tense for the analysis ("Our regression of x on y yielded ..." is just fine with us). And, of course, if you describe the past, then you should always use past tense.

(When you are at this stage, you should also put your author names on the draft you submit. The CFR allows blind or non-blind submissions in general, but because I will want to post accepted papers on the editorial site, unblinding at later stages takes care of one more checkpoint.)

CFR (LaTeX) Typesetting Instructions

The LaTeX style files are now supported at NOW Publishers.

In style-old.html, you can also find some older notes on how to typeset. They can be useful, but they are no longer supported and should not be blindly believed.


We expect you to provide the (basic) computer programs you have used, as well as a sample data set. The computer programs should make it possible to replicate your key results, not all results. (We do not encourage (or expect) you to offer support to everyone from the web who does not understand your computer programs. You are not obliged to help them further.) If the data is proprietary, please provide a random-number data set that makes it possible to run your programs. For purchased data (e.g., CRSP and Compustat), please provide at least a few sample observations (e.g., 5 stocks), so that readers can purchase the data and then replicate your and check their own calculations. Providing a few data points is not only permitted by copyright law, it is also not against commercial data vendor policies.


Authors must share copyright to their published article with the CFR. That is, unlike say Elsevier or Springer, both the authors and the CFR journal own non-exclusive copyrights to the published articles, and can either republish (or sell) the articles.

Authors who do not want to agree to these terms are hereby asked not to submit to the CFR.

/submit/style.html Last modified: