In my last post, I raised questions about the 2012 Jakarta Statement on Principles for Anticorruption Agencies (ACAs). My main concerns were (1) that many of the principles were framed in such open-ended and flexible terms that they failed to really commit to anything in particular; (2) that a number of the principles that do endorse concrete criteria are questionable on substantive grounds; and (3) the statement failed to acknowledge or address a key tension between its calls for more mechanisms to promote ACA independence and its call (in more general terms) for mechanisms to preserve accountability and prevent ACAs from abusing their power. Here I want to follow up on the first concern, and highlight recent research on the effect of the 1991 Paris Principles on the Design of National Human Rights Institutions (NHRIs).

The comparison between the Jakarta Statement and the Paris Principles is apt. Samuel De Jaegere, a UN official and one of the main proponents and advocates of the Jakarta Statement, published a 2012 article outlining principles for anticorruption agencies, which the Jakarta Statement itself tracks closely, though not precisely. (I am not sure whether De Jaegere’s paper formed the basis of the Jakarta Statement, or whether both the paper and the Statement were the result of ongoing behind-the-scenes discussion and dialogue, but they are clearly related.) De Jaegare’s article specifically references the Paris Principles for NHRIs as a model that ACAs could follow, and goes so far as to suggest that the voluntary “accreditation” system that the International Coordinating Committee of National Institutions for the Promotion and Protection of Human Rights (ICC) has created for NHRIs (in which letter grades are assigned based on the degree of compliance with the Paris Principles) could be emulated for ACAs. As De Jaegere puts it, “The main lesson learnt from NHRIs for ACAs is: a set of principles and a monitoring mechanism may strengthen the independence, effectiveness and success of ACAs.”

That sounds appealing, but recent empirical research on the actual impact of the Paris Principles should give us pause. That research, by Katerina Linos of Berkeley Law School and Tom Pegram of University College London, suggests that while the Paris Principles appears to have succeeded in promoting adherence to the Principles’ fixed terms (in both democratic and authoritarian states), the results were quite different for those Principles framed in more flexible, open-ended terms: For those provisions, implementation generally did not improve, and in authoritarian states adherence to those Principles articulated in flexible language sometimes actually worsened. Linos and Pegram speculate, plausibly in my view, that states responded strategically to the Paris Principles, redirecting their efforts toward tasks that were specified in firm, precise, unconditional language (where there would be less room for dispute about compliance), and away from the more amorphous, open-ended tasks.

Linos and Pegram’s paper is clever, in part because they have found (or believe they have found) a clever way around an inferential problem that ordinarily bedevils efforts to assess the impact of international agreements on state behavior. Their results, insofar as we believe that they are valid, may have important implications for how we think about attempts to emulate the approach to international assessment of NHRIs for ACAs. Let me say a few words about each.

First, on methodology: The problem with trying to figure out how international agreements, or differences in the language of international agreements, might influence state behavior is that the same underlying factors may influence both the international agreements (or lack thereof) and the state behavior itself. (This is what statisticians would call an “omitted variable bias” problem.) Suppose, for example, we observe that countries that sign an international human rights agreement have a worse human rights record than countries that don’t. This might mean that the agreement doesn’t work, but it might also mean simply that countries that don’t take their international human rights obligations seriously are more likely to sign such an agreement. Alternatively, if countries that sign an international human rights agreement see an improvement in their human rights records relative to countries that don’t sign, it might be because the agreement has an impact, or it might be because countries that are experiencing (or are about to experience) an improvement in their human rights record are more likely to sign. The same problem occurs when trying to evaluate the impact of firm vs. flexible language in international agreements: Countries may opt for firm language because they already know they will be able to comply (and choose flexible language where they anticipate compliance problems). Alternatively, countries may choose firm language where concerns about compliance are greatest, precisely because they think an international agreement is most needed for such issues (but where compliance may nonetheless remain a problem).

Linos and Pegram believe that the unusual history of the drafting of the Paris Principles means that it is a special exception to these usual concerns. Long story short: Based on close analysis of the actual history of these principles, it appears that they were drafted somewhat haphazardly, in a three-day period, without any input from the official representatives of the States Parties who eventually signed on. In other words, Linos and Pegram argue, the selection of principles to be included, and the decisions as to which ones would be framed as mandatory and which ones would be framed as optional (or framed in open-ended, flexible language) was, for all intents and purposes, random–not literally random, but uncorrelated with any other factors which might affect states’ incentives to comply.

This is super-clever, and a nice example of how quantitative analysis and qualitative expertise can be combined to generate insights that reliance on one or the other approach alone would miss. I leave it to interested readers to take a look at Linos & Pegram’s paper to see if they are persuaded. The one caveat/concern I’ll mention here is that the fact that the states allowed these principles to be negotiated in this fashion, and then were willing to sign on without further discussion or amendment, is itself likely not a random choice. It may be, for example, that the negotiations could proceed in this way precisely because the states involved did not take this agreement seriously enough. So, the results of the study, even if valid in this domain, might not generalize.

That said, the Jakarta Statement (and related proposals, such as De Jaegare’s call for a similar assessment and certification mechanism) is probably as close an analogue to the Paris Principles as one will find. Suppose that we temporarily put aside questions about the validity or generalizability of Linos & Pegram’s findings, and ask what the implications of their findings would be for the Jakarta Statement, particularly if this Statement got more traction and a certification system were established. I think it might go something like this:

• We’re likely to see greater adherence to those Jakarta Statement principles that are framed in concrete terms, and where there’s less room for debate about compliance: Countries would be more likely to establish permanent ACAs via a constitutional provision or special law; ACAs would have full management and control over their budgets and human resources (outside of the normal civil service system); ACA heads would be insulated from removal, except by an impeachment procedure similar to that used for the removal of the Chief Justice or equivalent; ACA employees would have immunity from any civil or criminal proceedings for acts committed within the performance of their duties; and ACAs would likely file some sort of annual public report.
• On the other hand, we would be unlikely to see any significant improvement — and in less democratic countries, we might actually see worsening — with respect to principles that the Jakarta Statement frames in more open-ended, flexible terms. Those would include: the content of the ACA’s mandate; the nature of ACA relations with other agencies and stakeholders; the process for appointing ACA heads; ethical codes, and both internal and external accountability mechanisms.

The question is whether this world is better or worse than a world in which there are no explicit ACA standards articulated in an international agreement, but instead there is informal domestic and international pressure to improve ACA performance. Again, the risk that Linos & Pegram outline is that if something like the Jakarta Statement is coupled with an evaluation/certification system modeled on the ICC, then: (A) the evaluation system may, naturally and understandably, focus on those criteria that are easily observed; (B) that will give countries incentives to do well on those dimensions, thus earning a good grade from the evaluators; but (C) the very fact of having a good grade will help deflect or defray domestic or international criticism of the ACA’s performance, because the country will be able to say, “Hey, our ACA is fine — look, we got an ‘A’ grade from the international monitors for our compliance with the Jakarta Statement principles.”

I’m not sure whether I think that risk is a significant one, or whether — even if it is — it might nonetheless be a risk worth taking given the potential upside of improving ACA performance through a set of agreed-upon international standards. But I think that those who are interested in promoting such standards, and using the Paris Principles for NHRIs as a model, need to take Linos & Pegram’s research very seriously, and develop a cogent response to the critique that their research implies.