Empirical Finance Paper Template

Finance manuscript: title page and abstract, disclosure page, hypotheses, data and attrition, identification, results, robustness, Internet Appendix, data and AI statement

शैक्षणिकपूर्णJournal of Finance submission guidelines; JF Data and Code Sharing Policy (2024-04-01)empirical finance paper mathअपडेट किया गया 24 अक्टू॰ 2026

एडिटर में खोलें.md डाउनलोड करें

उपयोग कैसे करें: Use it when drafting an empirical corporate finance, asset pricing or banking paper for a finance journal. Fill the question and answer, the sample attrition table and the baseline regression first, then write the introduction and abstract last. The main pitfall is reporting significance without economic magnitude or an explicit identification argument.

प्रीव्यू

Empirical Finance Paper: [Working title stating the question or finding]

Manuscript template for an empirical corporate finance, asset pricing, banking or household finance paper written for a peer-reviewed finance journal. It follows the structure most finance journals expect (introduction with contribution, hypothesis development, data and sample construction, identification, results, robustness, conclusion and an Internet Appendix) rather than the IMRaD format used in the health sciences. Submission facts cited for the Journal of Finance and the Journal of Financial Economics were checked on the journals' official pages in October 2026. Replace every [bracketed] value and every row marked Example. This is a document format, not investment, financial or legal advice. Always follow the current submission guidelines of your target journal, which take precedence over this template.

Document control

Field	Value
Paper ID	[PAPER-2026-001]
Working title	[Working title]
Field	[Corporate finance / Asset pricing / Banking / Household finance / Market microstructure]
Target journal	[Example: Journal of Finance]
Abstract limit	[Example: 100 words for the Journal of Finance; check the journal page]
Page limit	[Example: 60 pages including appendices, references, tables and figures for the Journal of Finance; Internet Appendix excluded; check the journal page]
Review model	[Single-blind / Double-blind]
Code and data policy	[Example: JF Data and Code Sharing Policy (1 April 2024): disclosure plan at first submission]
Version	[0.1]
Status	[Draft / Internal review / Working paper / Submitted / Revise and resubmit / Accepted]
Owner (corresponding author)	[@corresponding-author]
Approvers / reviewers	[@co-authors], [@friendly-reader]
Last updated	YYYY-MM-DD
Classification	[Confidential draft / Public working paper]
Working paper series	[SSRN / NBER / institution series, or "not posted"]

Approvals

Role	Name	Decision	Date
Corresponding author	[@name]	[Ready to submit]	YYYY-MM-DD
Co-author	[@name]	[Approved final version and author order]	YYYY-MM-DD
Co-author	[@name]	[Approved final version and disclosure statement]	YYYY-MM-DD

Revision history

Version	Date	Author	Change summary
0.1	YYYY-MM-DD	[@author]	First full draft
0.5	YYYY-MM-DD	[@author]	Example: Conference version; added staggered difference-in-differences estimator
1.0	YYYY-MM-DD	[@author]	Example: First journal submission
1.1	YYYY-MM-DD	[@author]	Example: Revise and resubmit round 1; new placebo tests in section 6.2

Title page and abstract
Disclosure statement
Introduction
Related literature and hypothesis development
Data and sample construction
Empirical methodology
Main results
Robustness and alternative explanations
Conclusion
References
Tables and figures
Internet Appendix
Data, code and AI-use statement
Pre-submission checklist
Glossary
Standard and licence note

1. Title page and abstract

Field	Value
Title	[Example: Losing the Analyst: Coverage Terminations and Corporate Investment]
Authors and affiliations	[Author A (Institution); Author B (Institution)]
Corresponding author	[Name, postal address, email]
Date of this version	YYYY-MM-DD
JEL codes	[Example: G24, G31, G14]
Keywords	[3 to 6 terms not repeated from the title]
Abstract word count	[Example: 76 of 100]

Acknowledgements footnote: [Thank discussants, seminar audiences and research assistants by name. List funding sources with grant numbers. State that errors are the authors' own.]

Abstract: [State the question, the setting and the identification source, the main finding with its economic magnitude, and one sentence on the mechanism. Example: We study how losing sell-side analyst coverage affects corporate investment. Using broker closures and mergers as exogenous shocks to coverage, we find that treated firms reduce capital expenditure by 7.75 percent of the sample mean relative to matched controls. The effect is concentrated in firms with high information asymmetry and is absent when the departing analyst covered the firm for less than one year. The evidence suggests that analysts relax financing constraints by reducing information frictions.]

A strong abstract answers four questions in under 100 words: what you ask, how you identify it, what you find and how large the effect is, and why it matters. Avoid adjectives such as "novel" or "important"; let the magnitude carry the claim.

2. Disclosure statement

Author	Funding and paid relationships	Data provider relationships	Other potential conflicts	Statement
[Author A]	[Grant name and number; consulting for named firms in the past three years]	[Example: None]	[Example: None]	[I have nothing to disclose / Disclosures listed]
[Author B]	[Example: Paid speaker for an industry association, YYYY]	[Example: Data obtained under a non-disclosure agreement with a data vendor]	[None]	[Disclosures listed]

Write one entry per author even when nothing is disclosed. Include relationships with any party that has a financial interest in the paper's findings: data vendors, regulators, consulting clients, asset managers, and expert witness work in related litigation.

3. Introduction

3.1 Question and answer

[State the research question in one sentence and the answer in one sentence. Example: Does sell-side analyst coverage affect how much firms invest? Yes: firms that lose coverage for reasons unrelated to their prospects cut investment, and the cut is largest where information asymmetry is highest.]

3.2 Motivation

[Explain why the question matters for theory, for practice or for policy. Give one fact that shows the economic stakes, with its source.]

3.3 Empirical approach in brief

[Describe the setting, the source of variation and why it is plausibly exogenous, in plain language. Name the main threat to identification and say how you address it.]

3.4 Main findings and magnitudes

[Give the headline estimate, its standard error or confidence interval, and its economic magnitude relative to the sample mean or standard deviation. Then list the mechanism and heterogeneity results.]

3.5 Contribution

Closest paper	What it shows	What this paper adds
[Author (Year)]	[Their finding in your words]	[Example: Uses a different source of exogenous variation and studies real investment rather than liquidity]
[Author (Year)]	[Their finding]	[Example: Separates the information channel from the monitoring channel]

3.6 Roadmap

[One short paragraph: "Section 4 develops the hypotheses. Section 5 describes the data..." Many journals now consider this optional; keep it short.]

4. Related literature and hypothesis development

4.1 Literature streams

[Group the literature into two to four streams and state the open question each leaves. Use the Literature Review template (academic/literature-review) to organise the search if the field is new to you.]

4.2 Hypotheses

ID	Hypothesis	Economic argument	Predicted sign	Test in section
H1 (Example)	Loss of analyst coverage reduces corporate investment	Coverage lowers information asymmetry and the cost of external finance	Negative	7.1
H2 (Example)	The effect is larger for firms with high information asymmetry	The marginal information value of an analyst is greater when other signals are scarce	Negative, larger in magnitude	7.2
H3 (Example)	The effect is weaker when the departing analyst had short tenure	A new analyst produces little firm-specific information	Closer to zero	7.3
[H4]	[Hypothesis]	[Argument]	[Sign]	[Section]

Distinguish confirmatory hypotheses fixed before looking at the outcome data from exploratory analyses added later. If you pre-registered any part of the design, cite the registration and link the Pre-Analysis Plan (academic/pre-analysis-plan).

5. Data and sample construction

5.1 Data sources

ID	Source	Content used	Frequency and period	Access and version	Licence or restriction
DS-001 (Example)	CRSP (via WRDS)	Daily and monthly stock returns, prices, shares outstanding	Daily, 1995-2024	Subscription; downloaded YYYY-MM-DD	Licensed, not redistributable
DS-002 (Example)	Compustat North America (via WRDS)	Annual fundamentals: capital expenditure, total assets, cash flow	Annual, 1995-2024	Subscription; downloaded YYYY-MM-DD	Licensed, not redistributable
DS-003 (Example)	I/B/E/S detail history	Analyst identifiers, broker identifiers, estimate dates	Event level	Subscription	Licensed, not redistributable
DS-004 (Example)	Hand-collected broker closure and merger dates	Event dates from press releases and regulatory filings	Event level	Authors' collection	Released with the replication package
[DS-005]	[Source]	[Content]	[Period]	[Access]	[Restriction]

Name commercial databases by product name only. State which identifiers link the files (for example CUSIP, PERMNO, GVKEY, analyst code) and how you resolved many-to-one links.

5.2 Variable definitions

Put the full definitions in an appendix table (section 11.3) and summarise the key variables here: the outcome, the treatment, and the main controls. Define each variable with its database item names so that a reader can recompute it, state the timing (fiscal year end, lag structure) and say how you treated extreme values.

Example: Investment is capital expenditure in year t divided by total assets at the end of year t minus 1. All continuous variables are winsorised at the 1st and 99th percentiles by year.

5.3 Sample attrition

Step	Filter	Observations removed	Observations remaining
0 (Example)	Firm-years in Compustat with a CRSP link, 1995-2024		48,250
1 (Example)	Remove financial firms and utilities (SIC 6000-6999, 4900-4999)	9,830	38,420
2 (Example)	Remove firm-years with missing or non-positive total assets	2,115	36,305
3 (Example)	Remove firm-years without analyst coverage in year t minus 1	4,005	32,300
Final (Example)	Estimation sample		32,300 firm-years

5.4 Descriptive statistics

[Report N, mean, standard deviation, 25th percentile, median and 75th percentile for every variable used in the regressions, for the full sample and separately for treated and control groups. Add a column with the normalised difference or the t-statistic of the difference in means.]

6. Empirical methodology

6.1 Baseline specification

y_{i,t} = \alpha_i + \gamma_t + \beta \, \text{Treat}_{i} \times \text{Post}_{i,t} + X_{i,t-1}' \delta + \varepsilon_{i,t}

Symbol	Meaning
$y_{i,t}$	Outcome for firm $i$ in year $t$ (Example: capital expenditure over lagged assets)
$\alpha_i$ , $\gamma_t$	Firm and year fixed effects
$\text{Treat}_i \times \text{Post}_{i,t}$	Equals one for treated firm $i$ in years after its own coverage shock (treatment dates are staggered, so Post is firm-specific); zero for control firms
$X_{i,t-1}$	Lagged controls (Example: size, Tobin's q, cash flow, leverage)
$\beta$	Coefficient of interest
$\varepsilon_{i,t}$	Error term; standard errors clustered by [firm / firm and year]

6.2 Identification

[State the identifying assumption in one sentence, for example parallel trends in investment between treated and control firms absent the shock. Then list the threats and your response to each.]

Threat ID	Threat to identification	Response	Section
ID-001 (Example)	Brokers close coverage of firms with poor prospects	Use only terminations caused by broker closures and mergers, which drop whole coverage portfolios	6.2
ID-002 (Example)	Pre-existing differences in investment trends	Event-time plot of coefficients for years minus 3 to plus 3	7.1, Figure 1
ID-003 (Example)	Staggered treatment timing biases two-way fixed effects	Report a heterogeneity-robust staggered difference-in-differences estimator	8.1
ID-004 (Example)	Contemporaneous shocks to the same industry	Industry-by-year fixed effects	7.1, column (3)

6.3 Inference

State the clustering level and why it matches the level at which treatment varies or errors are correlated. Panel data in finance often have both firm and time correlation; Petersen (2009) discusses the choice. When the number of clusters is small, report wild cluster bootstrap p-values. For multiple outcomes or many subgroups, say how you control the false discovery rate.

6.4 Matching or weighting (if used)

[Describe the matching variables, the algorithm, the caliper and the balance diagnostics. Report balance before and after matching in a table.]

7. Main results

7.1 Baseline result

[Describe Table 3. Example: The coefficient on Treat x Post in column 3 is minus 0.0062 (standard error 0.0021). The mean investment rate in the sample is 0.0800, so the estimate implies a decline of 7.75 percent of the mean (0.0062 divided by 0.0800).]

Column	Fixed effects	Controls	$\hat{\beta}$	Standard error	N	Within $R^2$
(1) (Example)	Firm, year	No	-0.0071	0.0023	32,300	0.012
(2) (Example)	Firm, year	Yes	-0.0065	0.0022	32,300	0.041
(3) (Example)	Firm, industry by year	Yes	-0.0062	0.0021	32,300	0.048

7.2 Cross-sectional heterogeneity

[Report the test of H2: split by a pre-determined measure of information asymmetry, or interact the treatment with it. State the difference between groups and its standard error, not only the two separate estimates.]

7.3 Mechanism

[Show evidence for the channel and against rival channels. Example: treated firms see a rise in the cost of new debt and a drop in equity issuance, consistent with the financing-friction channel.]

7.4 Economic significance

Translate the estimate into units a reader cares about: currency amounts, percentage of the mean, or the effect of a typical shock. State the conversion explicitly so that the reader can check it.

8. Robustness and alternative explanations

8.1 Robustness tests

Test ID	Concern addressed	Change to the baseline	Result	Location
RB-001 (Example)	Staggered treatment bias	Heterogeneity-robust estimator	Similar magnitude	Table 6
RB-002 (Example)	Outliers	Winsorise at 5th and 95th percentiles	Similar	Internet Appendix Table IA.2
RB-003 (Example)	Alternative outcome definition	Investment including research and development	Larger in magnitude	Internet Appendix Table IA.3
RB-004 (Example)	Placebo timing	Shift event dates two years earlier	No effect	Table 7
[RB-005]	[Concern]	[Change]	[Result]	[Location]

Review the table before submission with three questions. First, does every test map to a threat named in 6.2 or to a rival story in 8.2? A test that answers no stated concern belongs in the Internet Appendix or nowhere. Second, is the sample the same as in the baseline? If a test loses observations, report the baseline on the reduced sample next to it so the reader can separate the effect of the specification from the effect of the sample. Third, do you report magnitudes, not only "similar" or "significant"? Give the coefficient and its standard error for every row in the table that appears in the paper. A common error is to describe a coefficient that halves as "robust" because it remains significant; say how much the estimate moves and why.

8.2 Alternative explanations

[For each rival story, state what it predicts that your story does not, and show the test.]

8.3 Limitations

[State what the design cannot rule out, external validity limits (period, country, firm size) and measurement limits. A frank limitations paragraph reduces referee friction.]

9. Conclusion

[Write the conclusion in three short paragraphs: findings, implications for theory and practice, and directions for future research.]

10. References

[Reference list in the journal's style.]

Method references commonly cited in empirical finance papers, listed here as a starting point:

Petersen, Mitchell A. (2009). Estimating standard errors in finance panel data sets: comparing approaches. Review of Financial Studies 22(1).
MacKinlay, A. Craig (1997). Event studies in economics and finance. Journal of Economic Literature 35(1), 13-39.
[Data citations: provider, product name, version or download date.]

11. Tables and figures

11.1 Table conventions

Convention	Rule for this paper
Numbering	[Tables 1 to n in order of first mention; Internet Appendix tables IA.1 to IA.n]
Standard errors	[In parentheses below the coefficient; clustering level stated in the caption]
Significance markers	[Example: , , ** for 10, 5 and 1 percent, or none, as the journal prefers]
Decimal places	[Example: Coefficients to four decimal places; descriptive statistics to three]
Units	[State units for every variable in the appendix definitions table]

11.2 Planned tables and figures

Exhibit	Content	Status
Table 1 (Example)	Sample attrition	Done
Table 2 (Example)	Descriptive statistics, treated versus control	Done
Table 3 (Example)	Baseline regression	Done
Table 4 (Example)	Heterogeneity by information asymmetry	In progress
Figure 1 (Example)	Event-time coefficients, years minus 3 to plus 3	Done
[Exhibit]	[Content]	[Status]

11.3 Variable definitions appendix

Variable	Definition	Source items	Source
Investment (Example)	Capital expenditure in year t over total assets at end of year t minus 1	CAPX, AT	DS-002
Treat x Post (Example)	One for treated firms from the first fiscal year ending after the shock	Event dates	DS-004
[Variable]	[Definition]	[Items]	[Source]

12. Internet Appendix

Section	Content
IA.A (Example)	Details of event identification and sources for each broker closure
IA.B (Example)	Additional robustness tables IA.1 to IA.8
IA.C (Example)	Proofs or model derivations
[IA.D]	[Content]

13. Data, code and AI-use statement

13.1 Data and code availability

[State where the code will be deposited, which data are included, which are licensed and how a replicator obtains them, and whether a pseudo dataset will be provided for restricted data. Build the full README with the Replication Package README (academic/replication-package-readme).]

Journal facts checked in October 2026 (check the current policy before submission):

Journal	Requirement summarised in our words
Journal of Finance	Data and Code Sharing Policy of 1 April 2024: a disclosure plan at initial submission; code, computation details, log files and permitted data within about 30 days of conditional acceptance; pseudo data where real data cannot be shared
Journal of Financial Economics	Accepted empirical, simulation or experimental papers share data, programs and run instructions in Mendeley Data or a domain repository; a pseudo dataset where data are proprietary
[Other journal]	[Requirement]

13.2 AI-use statement

[State whether and how AI tools were used: language editing, code assistance, data classification or generation. Authors remain accountable for all content. If AI generated or classified data, record the model, version or date stamp and the prompts in the replication package. The Journal of Finance states that AI tools may assist but may not be listed as authors, and asks for open-source or clearly time-stamped models and documented prompts when AI produces data or classifications.]

14. Pre-submission checklist

15. Glossary

Term	Meaning in this template
Identification	The argument that the estimated coefficient reflects the causal effect of interest, not a confounding factor
Difference-in-differences	Comparison of the change in outcome for treated units with the change for control units over the same period
Fixed effects	Unit or time intercepts that absorb time-invariant or common shocks
Clustering	Allowing regression errors to be correlated within groups when computing standard errors
Winsorising	Replacing extreme values beyond chosen percentiles with the percentile value
Internet Appendix	Online supplementary material published with the paper
Pseudo data	Fictional data with the same structure as restricted data, used to test that code runs
Economic magnitude	The size of an effect in meaningful units, as opposed to its statistical significance

Related templates: Research Paper (IMRaD) (academic/research-paper-imrad) for experimental and health-science formats, Event Study (academic/event-study) for short-window market reaction studies, Dissertation Proposal (academic/dissertation-proposal) for the thesis stage, and Conference Abstract (academic/conference-abstract) for finance conference submissions.

16. Standard and licence note

Structure aligned with common practice for empirical finance manuscripts and with the public submission pages of the Journal of Finance (American Finance Association, afajof.org) and the data and code sharing policy of the Journal of Financial Economics, as checked in October 2026. Journal requirements are summarised in original words and are not the official wording; always read the current author guidelines of your target journal. All other text and the fictional example study are original; example numbers are illustrative. Template licence: MIT.