The American Economic Association’s Data and Code Availability Policy is one of the strictest reproducibility mandates in academic publishing. It applies to every AEA journal — the American Economic Review, AER: Insights, the Journal of Economic Literature, the Journal of Economic Perspectives, and the four AEJ titles (Applied Economics, Economic Policy, Macroeconomics, Microeconomics) — and, in an abbreviated form, to AEA Papers and Proceedings. Unlike a funder data management plan, which is mostly a promise made before the work happens, the AEA policy is enforced after acceptance: a paper does not reach final publication until its data and code package clears review by the AEA’s own verification service.
This guide sets out exactly what an economics researcher must deposit, where it goes, what can be legitimately withheld, and how the AEA’s verification process works, current as of the AEA’s February 2026 policy revision.
Which papers the policy applies to
The policy applies to any paper accepted for publication that contains empirical work, simulations, or experimental work, regardless of whether the underlying data are public, restricted, or collected by the authors themselves. Purely theoretical papers with no computational results are generally out of scope. The obligation attaches before final acceptance — authors are expected to have their replication package ready, not to assemble it retroactively once a paper is already typeset.
What you must deposit: data, code, and a README
A compliant replication package has three required components:
- Data. Any raw data the authors collected themselves, plus any secondary data that is not otherwise freely and easily accessible to another researcher. If the analysis dataset can be fully reconstructed from a cited, publicly accessible source using the deposited code, the raw source file itself does not need to be re-deposited — but the code that builds it does.
- Code. Every program used to clean, transform, merge, and analyze the data — enough to reproduce every table, figure, and in-text statistic in the paper from the deposited (or clearly sourced) data.
- A README. A PDF document describing the contents of the package, the software and version requirements to run it, the exact instructions to execute it end to end, and the required Data Availability Statement (see below). The AEA Data Editor’s office publishes a widely used README template that most authors adapt directly.
The Data Availability Statement
Every package must include a statement describing how, where, and under what conditions an independent researcher can access the data needed to replicate the paper’s results — including any restrictions, application processes, fees, or embargo periods that apply. This statement is not boilerplate: reviewers check it against the actual deposited materials, and a vague or inaccurate statement is one of the most common reasons a package is sent back for revision. See CASRAI’s embargo entry for how time-limited access restrictions are typically documented in a statement like this.
Where to deposit: the AEA Data and Code Repository
The AEA strongly encourages — and in practice expects — deposit in the AEA Data and Code Repository, a dedicated collection hosted on openICPSR, the University of Michigan’s ICPSR data archive. Deposit in the AEA repository is free to authors regardless of institutional openICPSR membership. Other trusted repositories that meet FAIR data principles are acceptable only with the Data Editor’s specific approval — authors should not assume an alternative repository (a personal GitHub account, a lab website, a general-purpose archive) satisfies the policy on its own without that sign-off.
Exceptions: confidential, proprietary, and restricted-access data
Not every dataset can be openly deposited — administrative microdata, proprietary firm data, and data covered by a data use agreement are common in empirical economics. The policy accommodates this without waiving the obligation entirely. Authors withholding non-public data must still:
- Preserve the original data and code for at least five years after publication;
- Commit to assisting good-faith replication requests from other researchers within that window;
- Deposit whatever code and non-restricted materials can be shared publicly; and
- Document precisely where the restricted data reside and how a legitimate researcher could obtain access (e.g., through a Federal Statistical Research Data Center, a data use agreement with the original provider, or a restricted-access enclave).
Full exemptions from the policy’s requirements are rare and require explicit approval from both the AEA Data Editor and the handling journal editor — they are not something an author can self-declare in a cover letter.
Verification: what the AEA Data Editor’s office actually checks
The AEA has run a dedicated Office of the Data Editor since 2018 (Lars Vilhuber, Cornell University) that operates a pre-publication verification service: before a paper is finalized, a reviewer independently attempts to run the deposited code against the deposited (or documented) data and confirms it reproduces the paper’s reported tables and figures. This is a substantively stronger check than most journal data-sharing policies, which typically require only that a statement or a link be present, not that the code actually runs. See CASRAI’s guide to the replication crisis for how this kind of pre-publication computational check fits into the broader shift toward verified reproducibility in empirical research, and the reproducibility audit entry for the general concept this process is an instance of.
Reviewers work through a structured checklist covering package completeness, whether the README’s instructions actually work, and whether reported results match. Packages that fail are returned to the authors for revision before publication can proceed — this is why building a clean, well-documented package well before final acceptance saves real time at the point where a paper is otherwise ready to go.
Papers and Proceedings: a lighter version of the same policy
AEA Papers and Proceedings, the association’s annual conference volume, follows a related but somewhat abbreviated version of the same requirements. Authors preparing a P&P submission should still expect a replication package review at the AEA Data and Code Repository if they deposit there, including a completed compliance checklist from the reviewing team, even though P&P articles are shorter and typically carry less extensive empirical apparatus than a full journal article.
Common compliance mistakes
- Treating the README as an afterthought. A README that doesn’t specify exact software versions, run order, or expected runtime is one of the most common reasons a package fails review.
- Depositing code that depends on an author’s local file paths or an unshared private dataset with no substitute or documentation. If the code can’t run outside the author’s own machine, it isn’t reproducible.
- Writing a Data Availability Statement that describes intentions rather than facts — e.g., “data available upon request” without specifying from whom, under what conditions, or at what cost.
- Assuming a public GitHub repository alone satisfies the policy. GitHub is a fine complement (and code-only deposits there are common), but it is not a substitute for the AEA-vetted repository deposit and Data Editor approval where required.
- Leaving package preparation until after acceptance. Verification happens before final publication, so packages assembled at the last minute are the ones most likely to contain errors reviewers catch.
Frequently asked questions
Does the AEA policy apply to purely theoretical papers?
Generally no. The policy is triggered by empirical work, simulations, or experimental work. A paper with no computational results to reproduce falls outside its scope, though authors should confirm this with the handling editor if there’s any ambiguity.
What repository does the AEA require for data and code deposits?
The AEA Data and Code Repository, hosted on openICPSR, is the default and strongly preferred option, and deposit there is free to authors. Other FAIR-compliant repositories are acceptable only with the Data Editor’s specific approval.
Can I withhold confidential or proprietary data entirely?
Not entirely, no. You can withhold the restricted data itself, but you must still document how a legitimate researcher could access it, share whatever code and non-restricted materials you can, preserve the full materials for at least five years, and commit to assisting replication requests within that period. A blanket exemption from all of this requires explicit Data Editor and journal editor approval.
Who actually checks that my code runs?
The AEA’s Office of the Data Editor runs a pre-publication verification service in which a reviewer independently executes the deposited code against the deposited or documented data and checks the results against what’s reported in the paper, before the article is finalized for publication.
Is the requirement different for AEA Papers and Proceedings versus a full journal article?
P&P submissions follow a related, somewhat lighter version of the same policy, but authors depositing at the AEA Data and Code Repository should still expect a package review and completed compliance checklist from the reviewing team.
When does the replication package need to be ready?
Before final acceptance, not after. Verification happens as part of the path to final publication, so a package prepared only once a paper is otherwise done tends to surface errors at the worst possible time.
