Skip to content
The failures of a ‘model’ system: RSD in Canada

The drafters of the 1951 Refugee Convention must have been exhausted after negotiating the details of the refugee protection doctrine. They seem to have had no energy left to sort out how the refugee status determination (RSD) process should operate, declaring simply that signatories should design it according to their own legal traditions.

Canada’s answer to this challenge is routinely held up as a model for the world. Indeed, the Canadian refugee system has many noteworthy strengths. Claimants tell their stories at a full oral hearing to a professional adjudicator, not to a bureaucrat or a border officer. The adjudicator is not answerable to the government and has no competing priorities such as protecting the country’s political alliances or conserving its resources. Canadian adjudicators develop a good familiarity with the country of origin information and are instructed to be sensitive to claimants’ vulnerabilities. Legal representatives play an important role in most Canadian refugee hearings and the system provides trained interpreters. When claimants lose their cases, the majority have the right to appeal. Because of these and other progressive aspects of its design, the Canadian system recognises many refugees and mistakenly rejects many fewer than it otherwise would.

Yet this ‘model’ system regularly produces rejections that are as unreasonable as they are unfair, and its output is inconsistent to the point of arbitrariness. The reasons for this include: the architects of the Canadian system long ago lost sight of its fundamental purpose; they have never been committed to evidence-based reasoning; and they cannot agree on how to answer the key question that lies at the heart of this kind of legal decision-making.

Assessing risk

RSD is a risk assessment. The decision-maker has one job: to evaluate the danger that the claimant faces if sent home. This is where the Canadian model runs into its first major difficulty. In Canada’s common law legal tradition, as in many similar jurisdictions, administrative decision-making is a two-step process. First, adjudicators judge each allegation and accept as ‘fact’ all those – and only those – that they decide on a balance of probabilities are ‘probably’ true. Then they make a legal ruling based on these accepted ‘facts’.

Imagine if you were to use this kind of approach in deciding whether to eat a wild mushroom. You think that it is probably a chanterelle, so it is a chanterelle. That is now a fact. And since you are quite certain that chanterelles are edible, eating it would pose very little risk. In real life, your level of confidence in the proposition that the mushroom is ‘probably’ a chanterelle – and any remaining doubts that you might have about this – would be crucial to how safe you would feel eating it. ‘Probably’ covers a wide spectrum from ‘as likely as not’ to ‘almost definitely’. It makes a world of difference where within that range your ‘probably’ lies. When we assess risk, we must weigh up uncertainty. But in a Canadian refugee hearing, uncertainty disappears. Anything that the adjudicator thinks is probably true is certainly true, even if there is still a good chance that it is false. And anything that they think is probably false is certainly false, even if there is still a good chance that it is true. What is more, the chance that Canadian adjudicators are mistaken in their assumptions – that, for instance, the mushroom is not a chanterelle but is actually poisonous – is exacerbated by the system’s utter failure to promote evidence-based reasoning.

Evidence and plausibility

Canadian adjudicators consider evidence, of course: they consider the claimant’s statements and documents, the country of origin information, and sometimes a government dossier or the testimony of third parties. But in deciding what conclusions to draw from this evidence, the adjudicators are guided entirely by their own common sense, which is often at odds with the best available social scientific research.

Canadian adjudicators’ common sense regularly tells them, for example, that we form clear, stable and consistent memories of our experiences that we can play back in our minds like a video recording. According to this theory, if a claimant cannot remember clearly the dates or times or frequency or order of the events that they are describing, or if their testimony contains other kinds of minor errors, gaps or inconsistencies, it is fair to infer that they must have invented their story. Yet for many decades a major thrust of the study of cognitive psychology has been to document extensively how incomplete, how fallible and how changeable our memories are, even our everyday autobiographical memories – to say nothing of traumatic memories and the memories of those who have been affected by trauma.[1]

Similarly, Canadian adjudicators routinely assume that when danger arises, people will quickly take effective measures to protect themselves. If the claimant persevered for a while before deciding to flee, if they hesitated to claim asylum when they finally reached safety, if they ever dared to return home, then surely their story must be a lie. They would have acted otherwise – ‘more sensibly’ – if the danger were real. I recently analysed 300 rejections written by Canadian adjudicators. In nearly two thirds of the decisions in which the adjudicator concluded that the claimant was lying, this finding rested at least in part on the adjudicator’s impression that the claimant’s response to an alleged danger was too unreasonable to be believed.[2]

The Canadian refugee system provides its adjudicators with hundreds of thousands of pages of country of origin information to help them to do their job well, yet it provides not one single page of social scientific evidence about how people think and act. There is no excuse for this failure. Adjudicators need this kind of evidence to make fair decisions about where to draw the line between plausible and implausible memory failures, for example, or between plausible and implausible responses to risk.

Which mistake is worse?

Perhaps most fundamentally, Canadian refugee law – and indeed international refugee law – has failed to answer the most important question at the core of this kind of legal decision-making: that is, which is the wrong kind of mistake in an RSD decision. Two potential errors hang in the balance any time a decision-maker has to decide whether to accept an allegation under conditions of uncertainty. They might reject a true allegation, or they might accept a false one. Which kind of mistake would be worse?

Blackstone’s ratio is one of the most famous maxims in Anglo-American common law: “it is better that ten guilty persons escape, than that one innocent suffer.”[3] Throughout the ages, the architects of this body of law have felt strongly that convicting the innocent is the wrong kind of mistake, and as a consequence Anglo-American common law is uniquely hard on the prosecution: the State bears the burden of proof and it must meet a very high standard of proof. As a result, in theory and in keeping with Blackstone’s ratio, the prosecution should pay the price for judges’ and jurors’ uncertainty.

International refugee law should recognise an imperative under the Convention to resolve doubt in the claimant’s favour for a variety of legal and ethical reasons.[4] It should loudly proclaim that it is a worse mistake to deny protection to someone who needs it than to give it to someone who does not. But in the absence of a sufficiently clear statement in the Convention to this effect, the creators of refugee law in Canada – the judges of the Canadian Federal Courts – are divided on this question. Some are more worried about sending refugees home to persecution. Others are more worried about giving people a benefit that they do not deserve. As a result, over time their judgments have constructed two parallel legal landscapes, one that resolves doubt in the claimant’s favour and the other at the claimant’s expense. Canadian adjudicators are free to choose, in any case and for any reason, which of these bodies of law to use. Under such circumstances, it not surprising that there are “vast disparities” in the grant rates of Canadian adjudicators.[5] And when a legal system’s decision-makers have the discretion to make whichever decision they want for whatever reason they want, the human beings who depend on it will be vulnerable to abuse.

Canada has a world-leading refugee system and its decision-making model is a very good place to start the conversation about what good RSD looks like. It gets many things right and grants protection to very many people who need it. Yet the Canadian system too often denies claims for the wrong reasons. Anyone looking to emulate it should think hard about why this is and should do better.


Hilary Evans Cameron
Assistant Professor, Faculty of Law, Ryerson University


[1] Hilary Evans Cameron, (2010) ‘Refugee Status Determinations and the Limits of Memory’, International Journal of Refugee Law 22

[2] ‘Credibility assessment in refugee hearings: A quantitative study and a way forward’: author’s research project, funded by the Social Sciences and Humanities Research Council of Canada, results forthcoming.

[3] Volokh A (1997) ‘n Guilty Men’, University of Pennsylvania Law Review 146

[4] Hilary Evans Cameron (2018) Refugee Law’s Fact-finding Crisis: Truth, Risk, and the Wrong Mistake, Cambridge University Press

[5] Rehaag S (2019) ‘2018 Refugee Claim Data and IRB Member Recognition Rates’

This site is registered on as a development site. Switch to a production site key to remove this banner.