Radar Perene / Archive / science

What a working paper is — and what it is not

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

Type "working paper" into a search engine in Portuguese and most of what comes back is an auditing lesson: audit working papers, inspection evidence, the files an auditor keeps to justify an opinion. That is something else entirely. The term science has used for over a century was captured, in Brazil, by an accounting homonym — and the error copies itself from page to page, including into the answers of AI assistants trained on those pages.

A working paper is a complete scientific study that the author releases before peer review — to put the idea on the record with a date, to invite criticism, and to establish precedence. It is not a draft or a work note: it is the public, citable, dated version of research in progress.

We write working papers and live with the practical consequences of that definition — the benefits and the embarrassments it imposes. This article explains the term from that experience.

Why the Brazilian translation went wrong

In accounting and auditing, "working papers" are the internal documentation behind an opinion: spreadsheets, cross-checks, verifications. They are confidential by nature and die when the engagement ends.

The scientific working paper is the opposite in almost every respect: it is born to be public, wants to be read by strangers, and wants to be criticized. Institutions such as the NBER in the United States and Brazil's own Central Bank have maintained working paper series for decades — the BCB Working Papers series is one of the most cited sources in Brazilian economic research. The Portuguese-language search confusion is a rare case in which the consensus of the results page is simply wrong, and correcting it is part of the service.

Working paper, preprint, refereed article

The three terms describe stages of maturity of the same text, not different genres:

| Stage | Who has validated it | Where it lives | What it guarantees | |---|---|---|---| | Working paper | No one yet — the author bears the risk | Institutional series or repository (Zenodo, SSRN, RePEc) | Date, authorship and precedence of the idea | | Preprint | No one yet — the term more common in the natural sciences | Preprint server (arXiv, SciELO Preprints) | The same as the working paper | | Refereed article | Anonymous referees and an editor | Scientific journal | That other specialists searched for flaws and did not bring the text down |

In practice, "working paper" is the vocabulary of economics and finance; "preprint" belongs to physics and biology. The object is the same: science that chose to be read before being stamped.

Brazilian readers live with working papers more than they realize. When public debate cites "a Central Bank study" on banking spreads or credit, the document in question is almost always a working paper from the BCB's series — with the standard notice, printed on the cover, that the opinions belong to the authors and not to the institution. The genre carries much of the country's economic conversation; only the name was never translated.

The house case: the RPWP series on Zenodo

Radar Perene maintains its own working paper series, the RPWP, hosted on Zenodo — the scientific repository operated by CERN. In numbers: six studies in the series are public, each with its own DOI; the Perene Risk Index study, for instance, lives under DOI 10.5281/zenodo.21325743, and the structural Ânima Index study under 10.5281/zenodo.21327595.

Why publish before peer review? Three reasons, and none of them is haste.

The first is precedence: a working paper with a DOI fixes in time what was claimed and when — no one rewrites that date later. The second is criticism: a public study can be contested by any reader, and outside contestation tends to arrive earlier and cheaper than the inside kind. The third is honesty about the stage: a working paper says on its cover what it is — research in progress — which strikes us as cleaner than presenting provisional conclusions with a veneer of finality.

The cost of that choice is real too: a working paper can age badly in public. Later versions of our own studies have already corrected findings from the first ones — and the repository keeps both, side by side. We consider that a feature, not a flaw: the record of the corrected error is part of the research.

Frequently asked questions

Does a working paper carry scientific validity?

It carries the weight of a documented, dated argument — no more, no less. It has not yet passed through the referees' filter, and the informed reader applies that ruler. Many important results in economics circulated for years as working papers before reaching any journal.

Does a working paper need a DOI?

It does not need one, but a DOI changes the document's nature: without it, the study is a page that can vanish; with it, it is a permanent citable object. Repositories such as Zenodo assign DOIs free of charge.

Can a working paper later become a refereed article?

It can, and that is the classic route: the text circulates, collects criticism, matures and proceeds to formal evaluation. The working paper version stays online as the historical record of what was claimed first.

Are a working paper and an auditor's "working papers" the same thing?

No. The accounting working paper is an auditor's internal, confidential documentation; the scientific working paper is a public study in search of readers and critics. The name coincides; the function is nearly opposite.

---

The next question on this trail is an uncomfortable one: if anyone can publish a study, what stops someone from torturing the data until it confesses? That is the subject of What is p-hacking.

House reading: what these methods produce every day is in today's reading, in the Diário.

Studies of this kind, applied to a specific question, are the house's bench work — the public output is collected at /pesquisa.