Draft economics papers with proper structure and academic style
若当前会话由 Coase 研究工作流触发（/idea-discovery / /experiment-bridge / /paper-writing），本 skill 的输出必须按以下规则落入阶段文件，不得自行新建目录或脱离工作流上下文：
executor/ 目录下已有主回归和稳健性结果，生成 Main Results Paragraph / Explanation Paragraph / Robustness Paragraph / Limitation Paragraph / Presentation Summary，输出到 writer/stage_4_writing_blocks.md。不得夸大结论、不得补跑回归。文字必须与 Table Package 表格结论完全一致。若用户未指定工作流（直接提问使用本方法），忽略本节，按下方正文自由执行。
This skill helps economists draft, structure, and polish academic papers with proper conventions for economics journals. It provides templates for different paper types and guidance on academic writing style.
Ask the user:
For empirical papers, use:
\section{Introduction}
% Hook - Why does this matter?
[TOPIC] is a fundamental question in economics, with implications for
[POLICY AREA] and [BROADER RELEVANCE]. Despite extensive research,
we still lack clear evidence on [SPECIFIC GAP].
% Research question
This paper asks: [RESEARCH QUESTION IN PLAIN LANGUAGE]?
Specifically, we examine whether [PRECISE FORMULATION OF THE QUESTION].
% Preview of answer
We find that [MAIN RESULT IN ONE SENTENCE]. This effect is
[economically significant / modest / heterogeneous], with
[QUANTITATIVE SUMMARY: e.g., "a one standard deviation increase
in X associated with a Y percent increase in Z"].
% Methodology (brief)
To identify this effect, we exploit [IDENTIFICATION STRATEGY:
natural experiment / RCT / instrumental variable / RDD].
Our data come from [DATA SOURCE], covering [TIME PERIOD]
and [SAMPLE SIZE] observations.
% Contribution / Related literature
Our paper contributes to several strands of literature.
First, we extend the work of \citet{Author2020} by [EXTENSION].
Second, we provide new evidence on [MECHANISM/CHANNEL] that
complements \citet{OtherAuthor2019}. Finally, our findings
have implications for [POLICY/FUTURE RESEARCH].
% Roadmap
The remainder of the paper is organized as follows.
Section~\ref{sec:background} provides background and reviews
related literature. Section~\ref{sec:data} describes our data
and empirical strategy. Section~\ref{sec:results} presents our
main findings. Section~\ref{sec:robustness} discusses robustness
checks. Section~\ref{sec:conclusion} concludes.
\section{Literature Review}
\label{sec:literature}
% Thematic organization: group papers by research strand, not chronologically
This paper builds on three strands of literature.
% Strand 1: Core literature
The first strand examines [BROAD TOPIC]. Early contributions by
\citet{Seminal1} and \citet{Seminal2} established [FOUNDATIONAL FINDING].
More recent work has extended these insights to [EXTENSION],
with \citet{Recent1} showing [KEY FINDING] and \citet{Recent2}
documenting [COMPLEMENTARY FINDING] using [DATA/METHOD].
% Strand 2: Related mechanism or context
The second strand studies [RELATED TOPIC / MECHANISM].
\citet{Author3} provide evidence that [MECHANISM FINDING],
a pattern confirmed by \citet{Author4} in the context of
[DIFFERENT SETTING]. This literature suggests that [IMPLICATION FOR OUR PAPER].
% Strand 3: Methodological or empirical context
The third strand concerns [EMPIRICAL CONTEXT / IDENTIFICATION].
Several papers exploit [SIMILAR IDENTIFICATION STRATEGY]:
\citet{Author5} use [INSTRUMENT/EVENT] to identify [EFFECT],
finding [RESULT]. Our approach is closest to \citet{Author6},
who [BRIEF DESCRIPTION]. We extend their work by [HOW WE DIFFER].
% Positioning: be explicit about the contribution
This paper contributes to these literatures in two ways.
First, [CONTRIBUTION 1: what new evidence you provide].
Second, [CONTRIBUTION 2: new method, context, or mechanism].
Unlike [CLOSEST PAPER], we [KEY DISTINCTION].
\section{Data and Empirical Strategy}
\label{sec:data}
\subsection{Data Sources and Sample Construction}
Our primary data come from [DATA SOURCE], which covers
[UNIT OF OBSERVATION] over the period [START YEAR]--[END YEAR].
[BRIEF DESCRIPTION OF WHAT THE DATA CONTAINS AND HOW IT WAS COLLECTED].
We construct our sample by [SAMPLE SELECTION STEPS].
We [INCLUDE/EXCLUDE] [CRITERIA], resulting in a final sample of
[N] [UNITS] ([OBS] observations). [NOTE any important restrictions
or potential selection concerns].
Our main outcome variable is [OUTCOME], measured as [DEFINITION].
The treatment variable is [TREATMENT], which takes value one if
[CONDITION] and zero otherwise.
Table~\ref{tab:sumstats} reports summary statistics.
\subsection{Empirical Strategy}
Our baseline specification is:
\begin{equation}
Y_{it} = \alpha + \beta \, D_{it} + \mathbf{X}_{it}'\gamma + \mu_i + \lambda_t + \varepsilon_{it}
\label{eq:baseline}
\end{equation}
\noindent where $Y_{it}$ is [OUTCOME], $D_{it}$ is [TREATMENT],
$\mathbf{X}_{it}$ is a vector of controls including [CONTROL LIST],
$\mu_i$ are [entity/firm/individual] fixed effects,
$\lambda_t$ are year fixed effects,
and $\varepsilon_{it}$ is the error term.
Standard errors are clustered at the [LEVEL] level to account for
[SERIAL CORRELATION / SPATIAL CORRELATION].
% Identification argument — be explicit
The key identification assumption is [ASSUMPTION IN PLAIN LANGUAGE].
This would be violated if [POTENTIAL THREAT]. We address this concern
by [STRATEGY: e.g., including controls, conducting placebo tests,
using an instrument]. [BRIEFLY STATE EVIDENCE FOR VALIDITY,
e.g., pre-trend tests, balance tables, first-stage F-statistic].
\section{Discussion}
\label{sec:discussion}
\subsection{Mechanisms}
Our results are consistent with [MAIN MECHANISM].
To provide evidence for this channel, we [TEST: e.g., split sample
by X, examine intermediate outcomes, test for mediators].
Table~\ref{tab:mechanisms} shows that the effect is [larger/present only]
among [SUBGROUP WHERE MECHANISM APPLIES], consistent with
[MECHANISM EXPLANATION].
An alternative explanation is [ALTERNATIVE MECHANISM].
However, this would predict [TESTABLE IMPLICATION OF ALTERNATIVE],
which we do not observe in [EVIDENCE]. We therefore conclude that
[MECHANISM FAVORED BY DATA].
\subsection{Heterogeneity}
[INTRODUCE HETEROGENEITY DIMENSION AND WHY IT IS THEORETICALLY INTERESTING]
Table~\ref{tab:hetero} reports treatment effects separately for
[SUBGROUP 1] and [SUBGROUP 2]. The effect is [X times larger /
present only] for [SUBGROUP], consistent with [EXPLANATION].
[SUBGROUP] may respond more because [ECONOMIC REASON].
\subsection{External Validity and Limitations}
Our estimates apply most directly to [POPULATION/CONTEXT].
Several factors may limit external validity.
First, [LIMITATION 1: e.g., our sample covers only X country/sector].
Whether these findings generalize to [OTHER CONTEXT] is an open question.
Second, [LIMITATION 2: e.g., our identification relies on a specific
policy change that may not be representative].
Third, [LIMITATION 3: data quality, measurement error, partial compliance].
\subsection{Policy Implications}
Our findings suggest that [POLICY INTERVENTION] could [EFFECT ON OUTCOME].
Back-of-the-envelope calculations suggest [QUANTITATIVE POLICY IMPLICATION].
However, policymakers should be cautious because [CAVEAT:
e.g., general equilibrium effects, targeting concerns, political economy].
\section{Results}
\label{sec:results}
% Lead with the main finding
Table~\ref{tab:main} presents our main results. Column (1) shows
the baseline OLS specification without controls. The coefficient
on [TREATMENT VARIABLE] is [POINT ESTIMATE] (s.e. = [SE]),
statistically significant at the [1/5/10] percent level.
% Add controls incrementally
In column (2), we add [CONTROL SET 1]. The point estimate
[increases/decreases slightly/remains stable] to [ESTIMATE].
Column (3) includes [CONTROL SET 2] and adds [FIXED EFFECTS].
Our preferred specification in column (4) includes [FULL CONTROLS]
and yields [FINAL ESTIMATE].
% Interpret magnitude
To gauge economic significance, note that [INTERPRETATION].
A one standard deviation increase in [X] is associated with
a [Y] percent [increase/decrease] in [OUTCOME], or roughly
[COMPARISON TO MEAN/OTHER BENCHMARK].
% Brief mention of mechanisms/heterogeneity if relevant
Table~\ref{tab:hetero} explores heterogeneity by [DIMENSION].
We find that the effect is [larger/concentrated among]
[SUBGROUP], suggesting that [INTERPRETATION].
\begin{table}[htbp]
\centering
\caption{Main Results: Effect of X on Y}
\label{tab:main}
\begin{tabular}{lcccc}
\hline\hline
& (1) & (2) & (3) & (4) \\
& OLS & + Controls & + FE & Preferred \\
\hline
Treatment & 0.052*** & 0.048*** & 0.041** & 0.039** \\
& (0.012) & (0.011) & (0.015) & (0.016) \\
\\
Controls & No & Yes & Yes & Yes \\
Fixed Effects & No & No & Yes & Yes \\
Cluster SE & No & No & No & Yes \\
\\
Observations & 10,000 & 9,850 & 9,850 & 9,850 \\
R-squared & 0.05 & 0.12 & 0.35 & 0.35 \\
\hline\hline
\multicolumn{5}{l}{\footnotesize Notes: * p<0.10, ** p<0.05, *** p<0.01.} \\
\multicolumn{5}{l}{\footnotesize Standard errors in parentheses.} \\
\end{tabular}
\end{table}
\section{Conclusion}
\label{sec:conclusion}
% Restate question and answer
This paper examined [RESEARCH QUESTION]. Using [METHOD/DATA],
we found that [MAIN FINDING]. This result is robust to
[ROBUSTNESS CHECKS].
% Implications
Our findings have several implications. For policy, they suggest
that [POLICY IMPLICATION]. For theory, they provide support for
[THEORETICAL MECHANISM] and challenge [ALTERNATIVE VIEW].
% Limitations (brief, honest)
Several limitations warrant mention. First, [LIMITATION 1:
e.g., external validity]. Second, [LIMITATION 2: e.g.,
data constraints]. Future research could address these by
[SUGGESTION].
% Future directions
This paper opens several avenues for future work.
[DIRECTION 1]. [DIRECTION 2]. We hope our findings
stimulate further research on [BROADER TOPIC].
When generating the full .tex file, use this structure. It incorporates the conventions that AER/QJE expect and avoids the layout issues that commonly arise when compiling locally.
\documentclass[12pt]{article}
\usepackage{amsmath,amssymb}
\usepackage[margin=1.25in]{geometry}
\usepackage{setspace}
\usepackage{booktabs,caption,threeparttable,pdflscape,makecell}
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage{natbib}
\usepackage{appendix}
\usepackage{subcaption}
% \usepackage{microtype} % enable if distribution supports it
\setstretch{1.5} % body text at 1.5x spacing (standard for submissions)
\captionsetup{labelfont=bf, labelsep=period, font=small}
\hypersetup{colorlinks=true, linkcolor=black, citecolor=black, urlcolor=blue}
\graphicspath{{../figures/}} % figures/ relative to the paper's directory
The global \setstretch{1.5} would push the abstract onto page 2 if applied to the title block. Wrap the front matter in \begin{spacing}{1.0} to keep it compact. Economics journals do not use a table of contents — omit it entirely.
\begin{document}
\thispagestyle{empty}
\begin{spacing}{1.0} % override 1.5x just for this page
\begin{center}
{\large\bfseries Your Paper Title Here:\\[0.2em]
Subtitle If Any}\par
\vspace{0.7em}
{\normalsize\bfseries Author Name}\par
\vspace{0.2em}
{\small \textit{Affiliation} \quad \href{mailto:[email protected]}{[email protected]}}\par
\vspace{0.2em}
{\small Month Year}\par
\end{center}
\vspace{0.5em}
\begin{abstract}
\noindent [Abstract text, 150-250 words]
\end{abstract}
\vspace{0.4em}
\noindent\textbf{JEL Codes:} J23, J31, O33 \quad
\textbf{Keywords:} keyword1, keyword2, keyword3
\vspace{0.3em}
\noindent\footnotesize\textit{I thank [seminar participants] for helpful comments.
All errors are my own.}
\normalsize
\end{spacing}
\clearpage
% NO \tableofcontents — economics journals don't use one
\section{Introduction}
...
\titlepage environment — it forces abstract to a separate page\tableofcontents — not standard in economics journal submissions\begin{spacing}{1.0} wrapping the title block — the global 1.5x stretch would otherwise inflate the title's line spacing\thispagestyle{empty} — suppresses page number on cover page\large\bfseries not \LARGE** — \LARGE combined with 1.5x stretch consumes too much vertical spaceAfter generating the .tex file, always compile twice (for cross-references) and inspect the log:
cd paper/
pdflatex -interaction=nonstopmode paper.tex
pdflatex -interaction=nonstopmode paper.tex # second pass for \ref, \cite
# Check for common problems:
grep -i "overfull\\|undefined\\|missing\\|error" paper.log
# Specifically: overfull tables
grep "Overfull .hbox" paper.log
# > 10pt: table or figure is visibly cut off at page margin — fix that table
# 1-5pt: minor, usually invisible — acceptable
# Missing packages
grep "File.*not found" paper.log
# If siunitx.sty: switch to c or dcolumn columns (see table skill)
# Unicode characters in .tex files
grep "Unicode character" paper.log
# Find the file and replace with LaTeX macro equivalents
Auto-fix checklist for the three most common compile errors:
| Error | Cause | Fix |
|---|---|---|
File 'siunitx.sty' not found | table used S columns | Switch to c or D{.}{.}{-1} |
Unicode character U+XXXX | Python wrote ≥, → directly into .tex | Replace with $\geq$, $\rightarrow$ |
Overfull \hbox (Xpt too wide) > 10pt | Table overflows margin | Add \begin{landscape}, \footnotesize, p{Xcm} column |
\Misplaced \omit / \Misplaced \span | \multicolumn used outside tabular | Body file is missing \begin{table}...\begin{tabular} wrapper |