Does ‘holding the pen’ give WTO staff too much power over disputes?

WTO REFORM: The light shed by recent papers on the role of the Secretariat is welcome, but there are unanswered questions

Image with writing saying “Be creative”, a drawing of a lightbulb, and a hand-held pen

By Peter Ungphakorn

Does the WTO Secretariat have too much influence over WTO dispute settlement rulings? Two experts argue controversially in recent papers that it does, backing their claims with sophisticated analysis of writing styles to detect who might have authored the rulings.

The analysis by Joost Pauwelyn (Graduate Institute of International and Development Studies, Geneva) and Krzysztof Pelc (McGill University, Montreal) also digs down in detail into the reasons behind the role that the Secretariat has been given.

That leads to a discussion covering a wide range of issues, some of them central to the debate about reforming the WTO and its dispute settlement system — including the controversy over whether case history and precedent should have a bearing on new rulings, and to what extent that should be guided by the institutional knowledge of Secretariat staff.

The analysis is useful in many ways. One is the explanation of why WTO dispute settlement is organised in a particular way — where ad hoc panels hear the case, often handing over to a fixed body of judges to hear appeals, in both cases with supporting secretariat work.


“Who guards the ‘guardians of the system’? The role of the secretariat in WTO dispute settlement”, American Journal of International Law, Volume 116, Issue 3, July 2022, pp. 534–566
“WTO Rulings and the Veil of Anonymity”, European Journal of International Law, chac027, August 2022 (paywalled, but an earlier version is here)

See also the blog posts (European Journal of International Law) “The WTO Secretariat’s ‘Open Secret’: Unpacking the Controversy” and “Does It Matter Who Writes Legal Rulings? It Depends”


● Giorgio Sacerdoti (Appellate Body judge 2001–09): “A critical reaction to Joost Pauwelyn and Krzysztof Pelc’s “The WTO Secretariat’s ‘Open Secret’: Unpacking the Controversy””, EJIL:Talk! blog, September 5, 2022

Another is the light it sheds on what the Secretariat does in dispute settlement cases. Pauwelyn and Pelc call this an “open secret”. Much of the information is available, but the Secretariat does keep a low profile in a process that’s supposed to be driven by the membership, not the bureaucracy. After all, the rulings are issued by the panels, with WTO members taking ultimate responsibility.

That means any contribution from the Secretariat is supposed to be institutional, which may be why the names of officials working with panels are no longer published.

(There has been a wider debate inside the WTO Secretariat about naming authors. It’s about output being institutional, versus allowing staff to be named so their work can go on their CVs. With some exceptions, the institutional line has prevailed.)

For Pauwelyn and Pelc, identifying the authors matters. This is why they also discuss at some length the separate issue of identifying any minority dissenting opinion from panellists and Appellate Body judges.

But some of the claims they make have been challenged by other experts, including former Appellate Body judge Giorgio Sacerdoti. The approach also raises several questions. The most fundamental is whether identifying the author of a text from the writing style is proof that the author has influenced a dispute settlement panel’s ruling.

These are my thoughts, as a non-lawyer who has only observed dispute settlement from outside the process.

Continue reading or jump to
There are two secretariats, not one | ‘Holding the pen’ | Other challenged points | The authors reply | See also

There are two secretariats, not oneBack to top

Pauwelyn’s and Pelc’s argument is weakened by conflating rulings from the (first stage) dispute settlement panels with those from the (second stage) Appellate Body.

The two are fundamentally different. Panellists are ad hoc. They are picked to work on a specific case, often from standing “indicative” lists of experts nominated by WTO member governments without any particular dispute in mind. (See this simple explanation and the accompanying sidebar box, or this comprehensive technical version.)

As Pauwelyn and Pelc explain, because the panellists are not full-time adjudicators, they need help from the WTO Secretariat: for secretarial and administrative tasks, and for research and drafting. Panellists are experts but they may not be familiar with all the legal details and history, including the legal arguments in previous disputes on similar issues.

The Appellate Body is different. In normal times there are seven permanent judges (officially Appellate Body “members”). Three of them work on each appealed case. (See this simple explanation, or this comprehensive technical version.)

In normal times the Appellate Body has its own secretariat — officially not even a secretariat but “support staff”. There are reasons why even full-time adjudicators need support from professional officials, and Pauwelyn and Pelc compare the WTO practice with those of other national and international courts or arbitrators.

These are not normal times. The US has blocked the reappointment of appeals judges and now none are left. The Appellate Body is suspended. We can overlook that here.

Controversially, Pauwelyn and Pelc treat the roles of the two secretariats as the same. They are not.

They do both report to the director-general and have the same or similar staff rules.

17. The Appellate Body and its support staff should be independent from the Secretariat

— Establishment of the Appellate Body WT/DSB/1

But the Appellate Body Secretariat’s focus is narrow. It only works on appeals in disputes. It is legally distinct from the WTO Secretariat, which covers all other WTO activities.

Supporting dispute panels is only one of the WTO Secretariat’s many functions, and for that task it often supplies subject specialists from other areas of work (such as on product standards or economic analysis) alongside legal experts.

The Appellate Body Secretariat defends the distinction. At one time it was rumoured to be seeking a building of its own. More recently its corridor has been locked, restricting access from other parts of the WTO building.

For appeals, the need for professional assistance with drafting is more to do with saving time and resources. Appeals judges can be expected to be more familiar with WTO jurisprudence than ad hoc panellists, but they don’t want to spend time drafting the nitty-gritty.

Pauwelyn and Pelc barely address those distinctions. As a result, the question of the Secretariat’s role in citing precedent and aiming for consistency in panel rulings, is lumped together with the US objections to precedent in appeals rulings.

Yes, the Appellate Body Secretariat has also come under fire from the US for over-stepping its mandate, but the suggestion that precedent cited in appeals is at the behest of the Appellate Body Secretariat doesn’t hold water. Appellate Body judges can think for themselves and are pretty vocal.

Photo of Bhatia with a quote about dealing with ambiguity from his speech
Unlikely to be influenced: Appellate Body judges are pretty forthright with their own views | Photo: WTO

The same can be said for many panellists. They are unlikely to be swayed by a Secretariat official without thinking it through for themselves. People like Thomas Cottier, Crawford Falconer, Tim Groser, Christian Häberli, Stuart Harbinson, Valerie Hughes, Debra Steger, to name just a few of those I happen to know or know of. (A full list of panellists can be found here.)

It is possible that some panellists might be more swayed than others by the drafts and other information presented to them by Secretariat staff.

But Pauwelyn and Pelc do not look at how independently-minded the panellists are. Instead they use computerised stylometry and text comparisons to detect who the authors of key parts of a ruling might be.

‘Holding the pen’Back to top

The result is a fascinating piece of detective work, particularly if we can guess the names represented by initials in some of the results.

When the technique is used to break the anonymity of a dissenting opinion in one Appellate Body ruling, Pauwelyn and Pelc say the American judge Thomas Graham “can be determined to be the author with near certainty”.

The case was a dispute where China challenged various aspects of US countervailing actions against alleged Chinese subsidies (case DS437).

The result is unsurprising. The dissenting opinion took the US line and was quoted by the US delegation in meetings of the Dispute Settlement Body (where the WTO membership oversees disputes). Pauwelyn and Pelc do this to show that hiding the identity of the dissenting adjudicator is pointless.

Still life photo, includes an object with “Write without fear. Edit without mercy” written on it
Correlation is not causation: identifying the author does not prove influence | Hannah Grace Unsplash licence

The problem with their stylometric analysis is that the result is used without reservation to answer the question: “How much actual influence does the WTO Secretariat ultimately exercise over WTO rulings?

The assumption is: if a writing style can be identified as likely to have come from a particular author, then that author must have influenced the content — the ruling.

Their conclusion is nuanced by recognising possible errors in stylometry and correlation:

“Authorship attribution remains a probabilistic exercise, and no test is definitive. Yet the preponderance of the evidence, using two distinct methods*, strongly suggests that the Secretariat plays a greater role in the drafting of WTO rulings than the panellists themselves. […] If anything, the gap in experience between the Secretariat staff and the average panellist has only grown over time, just as the volume of jurisprudence to master and complexity of cases has increased.” — EJIL article page 27

* “comparing reports to one another, and then comparing reports to external texts written by our candidate authors” — earlier version page 29

But it still misses other important reasons why the correlations might be less significant than they claim.

Here’s an extreme and personal one. In my professional life, I’ve spent almost as much time editing, rewriting and paraphrasing others as I have writing my own texts.

If Pauwelyn and Pelc were to analyse the writing style of Understanding the WTO (even in its current version), they would be hard-pressed to spot the authorship of “JK”, “GM” and other initials that appear in their output charts (Figures 1–4 on pages 26–28 of this version of the paper).

Why? Because I wrote it. Stylometry would identify me as “the author”. But the input came from dozens of experts — including “JK” and “GM” — who also checked early drafts and asked me to correct parts I had got wrong. We spent hours discussing those parts. Almost none of the input was my own. But the style was.

Or look at some of the pieces here on Trade β Blog. Some are written jointly with Bob (Robert) Wolfe, such as this one. Bob has also authored his own article.

In some cases, Bob gave me an idea and I wrote it. In others he sent me text. Sometimes I copied and pasted. Sometimes I rewrote it to fit the style of the joint piece. Even with his own article, he generously allowed me to edited it so it is closer to the style of the blog.

Bob’s a professor and a prolific writer. He has his own writing style. It would be fascinating to see what kind of correlation with the styles of “PU” and “RW” would emerge if those pieces on the blog were analysed for syntax. Authorship is complex.

I know panel reports are not edited for consistent style. (Sometimes I wish they were, and in a style that would help us all understand better.) And I’m not denying that if a Secretariat expert submits a draft, that might influence the panellists.

But there are three possibilities that also count.

One is that the panellists agree with the staff — independently, not because they have been influenced by a draft. Many of them have been in and around the WTO as negotiators who helped write the agreements in the first place, and as ambassadors and delegates involved in implementation and disputes. It should be no surprise if panellists and Secretariat officials think alike.

Another is that the panel can tell the official what to say or how to redraft a text, in the same way that Bob presented me with ideas which I then wrote using my own writing style. Or how “JK” and “GM” got me to correct my drafts.

The third is that whoever drafts a ruling has to respond to submissions from the countries in the dispute, which might or might not conform with staff preconceptions about precedent. That could constrain the drafter’s influence.

So while the textual analysis is interesting and may indicate something is at play, it does not prove undue Secretariat influence. To be truly empirical, what we need is information about what happens inside the panels. That is missing.

Other challenged pointsBack to top

Pauwelyn and Pelc argue that the WTO Secretariat’s influence over panellists (and appeals judges) also comes from its power to appoint panellists and to determine how much they are paid by controlling how many hours they work.

A number of experts have challenged this on Twitter.

Desiree LeClercq of Cornell University kicked off by asking: “Can someone WTO-savvy explain how Secretariat officials are responsible for appointing justices & controlling remuneration @ the @wto? At other IOs [international organisations], it is the member states, not the IO civil servants, who vote on those matters.”

The appointments and payments for the Appellate Body are set by the members, tweeted Debra Steger, former head of the Appellate Body Secretariat: “The WTO members appointed Appellate Body members [ie, judges] in the Dispute Settlement Body.”

WTO members also control payments for both through relevant committees, she went on:

“The WTO members in the Budget Committee and General Council are also responsible for setting the annual budgets for the divisions of the WTO Secretariat and formerly, the Appellate Body.

“Thus, WTO members decide on remuneration. The last time panellist remuneration was decided was 2016, to my knowledge. Secretariat does not do that. Members control purse strings.”

Tweets from Debra Steger (her account is locked)

On appointing panellists, Rob Howse of New York University Law School suggested the influence of the Secretariat was overstated.

“In practice, parties to the dispute exercise considerable influence, an effective veto, leading to delay, deadlock & [the director-general] appointing [panellists],” he tweeted.

And even though the rules say countries in a dispute should only oppose panellists proposed by the Secretariat “for compelling reasons”, in practice they have not been restrained, he added.

Describing how levels of pay are set for appeals, former Appellate Body judge Jennifer Hillman tweeted: “From AB end, level of pay is set by WTO Members. ABMs [Appellate Body members or judges] were paid a modest stipend (set by [WTO] Members) to compensate for being on call all the time (which precludes certain other work), plus travel expenses, plus a fee for hours worked — all reviewed but not changed by secretariat.”

For panellists: “The ‘control of remuneration’ is simply that panellists (& presumably AB members) submit their timesheets to the Secretariat. The levels of compensation are not set by the Secretariat. I have never before heard any suggestion that this has been used to control panellists,” tweeted former WTO Secretariat staffer Jesse Kreier.

“Not quite,” replied Pelc. “The expected [number] of hours panellists devote to a case is formally set out, and it’s strikingly low. Any additional time above that must be justified and approved by the Director of the Division, who must sign the claim form.”

Kreier responded: “In 26 years in the Secretariat, I never heard even the least inkling that this approval power should somehow be used to control panellists. Further, many if not most panellists are govt officials who don’t receive the payments but only the per diem for expenses.”

And, “I don’t generally dispute your assessment about the division of labour between panel and Secretariat, although I can think of notable exceptions.  I just think that, at least at panel phase (don’t know the AB well enough) it had very little to do with compensation issues.”

Pauwelyn and Pelc also say the WTO Secretariat’s unusual role is signified by the fact that some staff sit on the podium in the meeting room alongside the panellists. But that is standard practice in the WTO in all councils and committees, including negotiations among the members.

And so the debate is likely to continue.

One final thought: Pauwelyn and Pelc say the extent of Secretariat involvement means WTO dispute settlement is more like an “administrative review” than a judicial process. Where would that leave the study of “WTO law”?

The authors replyBack to top

Joost Pauwelyn: “Thanks @CoppetainPU

“Two staff roles you forgot to mention & are key when all roles added up:

“1. writing of ‘issues paper’ with dispute solutions even before adjudicators meet = sets default, groupthink

“2. lead voice at internal deliberations = unheard of in other tribunal”

Krzysztof Pelc: “Great summary + continuation of a rich discussion”

See alsoBack to top

September 28, 2022 — adding the link to Armin Steinbach’s paper under “see also”
September 7, 2022 — adding Giorgio Sacerdoti’s rebuttal
August 22–23, 2022 — adding the authors’ replies and “see also”

Image credits:
• “Be creative” | Pexels licence, Pixabay licence
• Bhatia photo | WTO
• “Write without fear” | Hannah Grace, Unsplash licence


Author: Peter Ungphakorn

I used to work at the WTO Secretariat (1996–2015), and am now an occasional freelance journalist, focusing mainly on international trade rules, agreements and institutions. (Previously, analysis for AgraEurope.) Trade β Blog is for trialling ideas on trade and any other subject, hence “β”. You can respond by using the contact form on the blog or tweeting @CoppetainPU

%d bloggers like this: