Updates, commentary, training and advice on immigration and asylum law

Analysis: the Borders Bill and the Refugee Convention


Older content is locked

A great deal of time and effort goes into producing the information on Free Movement, become a member of Free Movement to get unlimited access to all articles, and much, much more


By becoming a member of Free Movement, you not only support the hard-work that goes into maintaining the website, but get access to premium features;

  • Single login for personal use
  • FREE downloads of Free Movement ebooks
  • Access to all Free Movement blog content
  • Access to all our online training materials
  • Access to our busy forums
  • Downloadable CPD certificates

Under the sub-heading “Interpretation of Refugee Convention“, clauses 27-35 of the Nationality and Borders Bill 2021 seek to accomplish four main tasks:

  1. Translate some EU asylum law, currently residing in secondary legislation, into primary legislation.
  2. Turn back the clock on core principles of asylum law in relation to the identification of a “particular social group”, and the test and standard of proof for asylum claimants.
  3. Remove the protection against penalising refugees in Article 31(1) of the Refugee Convention from some asylum claimants.
  4. Lower the threshold for the definition of a “particularly serious crime” for the purposes of the Article 33(2) of the Refugee Convention.

This article analyses the Refugee Convention clauses of the Bill in some detail. For a summary of the Bill’s other provisions on asylum, see Free Movement briefing The Nationality and Borders Bill, Part 2 (asylum).

Clause 27: introductory

Clause 27 introduces this part of the Bill. It provides that the principles in subsequent clauses 28-34 apply “for the purposes of the determination by any person, court or tribunal” of whether a person is a Convention refugee, is excluded from Convention protection or is entitled to the immunity from penalties set out in Article 31(1) of the Convention. It is not explicitly stated that the Secretary of State and her employees are “persons” for these purposes, but there is no reason to suppose that they are not, so they are equally caught by the provisions.

Clause 27(4) repeals the Protection Regulations (Refugee or Person in Need of International Protection (Qualification) Regulations 2006 (S.I. 2006/2525). These regulations were made to implement, in part, the 2004 EU Qualification Directive. Clauses 28-33 then seek to translate many of the repealed regulations into primary legislation.

Clause 27(5)-(6) provides that clauses 28-33 do not have retrospective effect, but apply only to claims that were made on or after the day they come into force.  This exclusion from retrospectivity is not specifically extended to clause 34, which deals with the protection from penalties. This is presumably because the Secretary of State wishes to employ it against people who have already entered the country or made claims in circumstances which she deplores (one suspects it is primarily aimed at those crossing the Channel in small boats and victims of trafficking).

Clauses 28 and 30-33: translation of EU law principles

I use the word “translation” rather than “transfer” because some of the proposed changes involve a bit more tinkering than just swapping them from one legislative environment to another:

  1. Clause 28 replaces sections 3 and 5 of the 2006 regulations, which define “actors of persecution” and “act of persecution”.
  2. Clause 30 replaces Regulation 6, on “reasons for persecution”, but with amendments (see next section).
  3. Clause 31 replaces Regulation 4, defining “actors of protection”.
  4. Clause 32 implements Article 8 of the Qualification Directive on “internal relocation”.
  5. Clause 33 replaces Regulation 7, construing “exclusion” (i.e. the exclusion clauses in the Refugee Convention).

Minor changes made in the process of translation, which may or may not excite black-letter lawyers, include:

  1. Inserting the word “reasonable” before “protection” when translating Regulation 3(c) into clause 28(1)(c).
  2. Changing the words “shall” in Regulation 6(1)(a)-(b) and “shall for example” in Regulation 6(1)(c) to “may” in clause 30(1)(a)-(c) and “shall include” in Regulation 6(1)(f) to “includes” in clause 30(1)(a)-(c).
  3. Changing the words “cannot be understood to include” in Regulation 6(1)(e) to “does not include” in clause 30(5).
  4. Changing the words “Protection shall be regarded as generally provided when the actors [of protection]… take  reasonable  steps  to  prevent” in Regulation 4(2) to “An asylum seeker is to be taken to be able to avail themselves of protection… if …the [list of actors] takes reasonable steps to prevent” in clause 31(2)(a).
  5. Changing the words “and the person mentioned in paragraph (1) has access to such protection” in Regulation 4(2) to “the asylum seeker is able to access the protection” in clause 31(2)(b).

Clauses 29 and 30(2)-(4): Turning back the clock on core principles

Clause 29 aims to change the test for evidence and standards of proof in asylum claims. Clause 30 seeks in subclauses (2)-(4) to change the test for the definition of “particular social group”. These changes are clear attempts to reinstate approaches which have been rejected by the courts.

The test for evidence and standards of proof

The tests in asylum claims for considering past and present evidence and the applicable standard(s) of proof have been settled law in the UK since the Court of Appeal decided Karanakaran v Secretary of State for the Home Department [2000] EWCA Civ 11 in January 2000. The court in Karanakaran moved away from asylum seekers having to prove certain facts about their claim on the balance of probabilities (also known as the civil standard). Instead it preferred a single, lower test for risk — reasonable degree of likelihood — informed by an administrative rather than judicial mode of fact-finding.

As Lord Justice Sedley pointed out:

The civil standard of proof, which treats anything which probably happened as having definitely happened, is part of a pragmatic legal fiction. It has no logical bearing on the assessment of the likelihood of future events or… the quality of past ones.

The test approved in Karanakaran is consistent with a retreat from and rejection of the civil standard for the assessment of asylum claims in other common law jurisdictions. It has been an established cornerstone of asylum practice in this country for over 20 years.

Changing the test to “support improved decision-making”

A change to the Karanakaran approach was heralded in the government’s New Plan for Immigration in the following terms:

… we will take forward reforms to… support improved decision-making by setting a clearer and higher standard for testing whether an individual has a well-founded fear of persecution, consistent with the Refugee Convention.

In fact, the “clearer and higher standard” mandated by clause 29 of the Bill looks very like that rejected in Karanakaran. Clause 29(2) of the Bill would oblige a decision-maker to determine on the balance of probabilities:

(a) whether the asylum seeker has a characteristic which could cause them to fear persecution for reasons of race, religion, nationality, membership of a particular social group or political opinion (or has such a characteristic attributed to them by an actor of persecution), and

(b) whether the asylum seeker does in fact fear such persecution in their country of nationality (or in a case where they do not have a nationality, the country of their former habitual residence) as a result of that characteristic.

There are a number of problems with this.

Problem 1: Drafting and coherence

Clause 29(2)(a) is poorly drafted. It is not clear what “… a characteristic which could cause them to fear persecution for reasons of…” might refer to. It could simply be a rather garbled way of saying “whether, on the balance of probabilities, the facts of the claim engage a Convention reason“. But while that subjects the question of whether the claim engages the Convention to a balance of probabilities test, it does not necessarily mandate a balance of probabilities test to other findings of fact that might be relevant to the risk. For example, in a case where the Convention reason is ethnicity, clause 29(2)(a) would mandate the civil standard for the assessment of the appellant’s ethnicity, but no more.

Clause 29(2)(b) applies the civil standard to either or both of the claimant’s fear of a given risk and the reasons for that fear (probably both), while clause 29(4) mandates that the risk itself is only to be assessed to the lower standard of “reasonable likelihood”. While this is more or less intelligible as a matter of English syntax, it is by no means clear how a decision-maker is supposed to ensure that the two standards are not conflated or confused.

Furthermore, as clause 29 does not expressly reject the administrative approach to decision-making mandated by the Court of Appeal in Karanakaran, to what extent does it remain applicable? An incremental approach would be to give clause 29 a very narrow construction and say only those matters specifically and directly identified in the Bill need to be assessed at Kaja level 2 (“evidence they think is probably true”), whereas all other factual elements of the claim are to be given some weight if they pass Kaja level 3 (“evidence to which they are willing to attach some credence, even if they could not go so far as to say it is probably true”). A wider construction might be to find that all factual elements of the claim are relevant to the question of what the claimant fears and why, so they must all be assessed on the civil standard, before moving on to assess risk on the lower standard.

Problem 2. Lack of a litmus test to ensure consistency

The civil standard is notoriously imprecise. For all the judicial assertions that there is only a single “balance of probabilities”, in some inherently probable cases proving that something probably occurred is fairly easy (such as, for example, proving an allegation of abuse in care proceedings against a parent who has a long track record of it); in other, inherently improbable cases, it may be very difficult (e.g. proving fraud against a person who has no record of dishonesty).

The risk is that different decision-makers may apply inconsistent standards of probability without making that clear. The administrative approach approved in Karanakaran has the great virtue of transparency. Decision-makers are obliged to set out their evaluations of different aspects of the evidence. Assessments against the civil standard are inherently vaguer, because it posits an absolute standard of probability which does not, in fact, exist.

Problem 3. Conflicting tests in Article 3 cases

In addition to the above, this change would complicate the law on international protection by creating different modes and standards of assessment between Refugee Convention claims and claims based on Article 3 of the European Convention on Human Rights. At the moment, both are subject to the same modes and standards. This proposed change would only affect Refugee Convention claims.

“Improved decision-making”, my eye

There is in fact no good reason to think that the proposed change will serve the asserted aim of “improved decision-making” by setting a “clearer” standard. The standard proposed is more confusing and distinctly lower in quality than the Karanakaran model. While it is almost certainly intended to impose a higher hurdle for asylum claimants to get over, there is no evidence whatsoever it will lead to better decisions. In fact, it creates a real risk that more people who genuinely need protection will fail to establish their claims. It is difficult to avoid the conclusion that this is precisely the aim.

The short-term consequence is likely to be a great deal of judicial time and effort being taken up addressing some of the issues described above. As with the ghastly history of Article 8 consideration since 2012, the courts will have to work out a way of giving effect to the statutory provisions while balancing a number of competing interests: the nebulous will of Parliament as expressed in potentially unclear legislation; the policy aims of an executive with democratic legitimacy; the human rights of the claimants; and the need to give effect to what the House of Lords in Adan referred to as “the true autonomous and international meaning” of the Refugee Convention), which must be “untrammelled by notions of… national legal culture”.

The test for establishing a particular social group

The proper test for identifying a “particular social group” has been a particularly vexing issue in Refugee Convention law.  Although some elements of the test are widely accepted – for example, that fear of persecution cannot be the defining characteristic, as that would be circular; or that there is no requirement for group cohesion – the core question of how to recognise a particular social group remains contested.

The controversy the Bill seeks to stoke relates to whether the core test has one limb or two. The two limbs are referred to by Hathaway & Foster as “the ejusdem generis approach” and “the social perception approach”:

The ejusdem generis approach identifies a group in society which shares a characteristic that either cannot be changed because it is immutable or should not be changed because it relates to a fundamental human right;

The ‘social perception’ approach identifies a group in society with a common characteristic (apart from fear of persecution), which distinguishes it from society at large and/or by reference to which it is perceived by the surrounding society as constituting a distinct social group.

Different jurisdictions and legal commentators have adopted different positions about which approach is preferable, or perhaps whether both tests need to be satisfied cumulatively. The courts in the US and Canada, for example, have historically applied the ejusdem generis approach, whereas courts in Australia, France and Germany have tended to favour a form of the social perception test. UNHCR favours recognising a particular social group where the posited group passes either test.

In the case of K and Fornah [2006] UKHL 46, Lord Bingham found for the majority that the UNHCR interpretation was preferable and that imposing both tests cumulatively “propounds a test more stringent than is warranted by international authority”. Two years later the immigration tribunal in SB (PSG – Protection Regulations – Reg 6) Moldova CG [2008] UKAIT 2 thought that Lord Bingham’s opinion was obiter in this regard and ruled that the tests should be applied cumulatively. (It was probably wrong, and in any case its finding was arguably not binding.)

In 2020, the ghost of SB (at least, in this respect) was laid to rest with the judgment of DH (Particular Social Group: Mental Health) Afghanistan [2020] UKUT 223 (IAC). It repudiated SB and adopted the UNHCR approach: the criteria for establishing a particular social group must be read as alternative not cumulative tests.

Or so we thought.

Changing the particular social group test

How naive we were. For clause 30(2) of the Bill seeks to reimpose the cumulative test:

A group forms a particular social group for the purposes of Article 1(A)(2) of the Refugee Convention only if it meets both of the following conditions [emphasis added].

There is simply no good reason for this. Of the two tests, the ejusdem generis approach is the one that relies most surely on clearly identifiable principles. The “social perception” principle is beset with uncertainty: is it actually necessary for society to perceive a difference? How is one to evidence that?  Or is it simply that the group can be distinguished in some conceivable way?

Nevertheless, there may be occasions when the second test is genuinely useful. The catch-all approach approved last year in DH at least has the advantage that someone who is being persecuted for being a member of a group that is perceived as being distinct but may struggle to bring itself within the ejusdem generis approach (Stalin’s persecution of doctors, for example, or Pol Pot’s persecution of “intellectuals”) will have access to international protection.

The only possible utility to the arbitrary insistence on meeting both tests is that it might reduce the number of people who are able to access international protection under the Refugee Convention, regardless of whether they are genuinely at risk of persecution. That in no way serves the stated aim of “support[ing] improved decision-making” unless by “improved decision-making” the government actually means “refusing more claims”.

Clause 34: removing protection against penalising refugees

Clause 34 of the Bill aims to remove the protection against penalising refugees in Article 31(1) of the Refugee Convention from the following groups of claimants:

  • Clause 34(1): People who “stopped in another country outside the United Kingdom, unless they can show that they could not reasonably be expected to have sought protection under the Refugee Convention in that country”;
  • Clause 34(2)(a): People who entered the UK as a refugee if they failed to make a claim “as soon as reasonably practicable after their arrival in the United Kingdom”;
  • Clause 34(2)(b)(i): People with a sur place claim that arose when they were lawfully present in the UK, unless they make the claim before the expiry of their permission;
  • Clause 34(2)(b)(ii): People with a sur place claim that arose when they were unlawfully present in the UK unless they make the claim “as soon as reasonably practicable after they became aware of their need for protection under the Refugee Convention”; and
  • Clause 34(4): People who incur a penalty in the course of trying to leave the UK.

There will also be two amendments to section 31 of the Immigration and Asylum Act 1999, which is the section that provides refugees with defences to criminal charges.

Section 31(2) of the 1999 Act will now read:

If, in coming from the country where his life or freedom was threatened, the refugee stopped in another country outside the United Kingdom, [the defence] applies only if he shows that he could not reasonably have expected to be given be expected to have sought protection under the Refugee Convention in that other country.

This appears to be an attempt to restrict asylum seekers from saying that they were advised not to seek asylum in a third country because they would not have got it there anyway.

Clause 34(5)(b) of the Bill then says that the section 31 defence “does not apply to an offence committed by a refugee  in an attempt to leave the United Kingdom”.

The function of Article 31(1) is to protect people with a well-founded fear of persecution in their home countries from being penalised for the consequences of that fear. Clause 34 of the Bill seeks to exempt some refugees from that protection. In effect, it is a bully’s charter which enables the punishment of frightened people for some of the consequences of their fear.

Human nature being what it is, it is of course conceivable that some asylum claimants are not actually refugees. Decision-makers are obviously entitled to ask why a claimant has delayed making a claim, or not claimed asylum in any country they might have passed through on the way to this one.

But being a victim of persecution is not a crime. No human rights norm or principle of international law requires that people forced to flee their homes be punished by being obliged to stay in countries with which they have no connection. Nor is there any positive international law requirement that frightened people have to make protection claims as soon as they discover their predicament. There are lots of perfectly rational reasons why genuinely scared people might be reluctant to bring themselves to the attention of the very authorities with the power to send them home to face persecution.

Clause 35: lowering the “particularly serious crime” threshold

The second part of Article 33(2) of the Refugee Convention permits signatory states to refoule a refugee”‘who, having been convicted by a final judgment of a particularly serious crime, constitutes a danger to the community of that country”. The UK enacted section 72 of the Nationality, Immigration and Asylum Act 2002 to set some parameters on what might constitute a “particularly serious crime”. It creates presumptions that apply where someone has been convicted either inside or outside the UK of a crime:

  1. for which a prison sentence of at least two years was imposed (or could have been imposed if convicted in the UK); or
  2. is specified as a particularly serious crime for these purposes on an order made by the Secretary of State, or (if convicted abroad) which has been certified by the Secretary of State as similar to a crime specified on such an order.

Each of these conditions is prefaced in the 2002 Act with the words “A person shall be presumed to have been convicted by a final judgment of a particularly serious crime and to constitute a danger to the community of the United Kingdom if… “. Under section 72(6) of the Act, both the presumption that the conviction represents a particularly serious crime and the presumption that the person constitutes a danger to the community of are made rebuttable. This interpretation was authoritatively fixed in 2009 by the immigration tribunal in IH (s.72; ‘Particularly Serious Crime’) Eritrea [2009] UKAIT 12 and the Court of Appeal in EN (Serbia) [2009] EWCA Civ 630, in part in order to ensure that the provision remained consistent with the EU Qualification Directive.

Clause 35 of the Bill will make three changes, to:

  1. Clarify that section 72 relates only to the question of whether a refugee may be expelled or returned and not to the person’s status as a refugee;
  2. Lower the threshold length of prison sentence from two years to 12 months; and
  3. Make the presumption that the conviction represents a particularly serious crime irrebuttable.

The third point in particular is asking for trouble. In EN (Serbia) the Court of Appeal specifically found (paragraphs 68-69) that it would be incompatible with the Refugee Convention to make the presumption that a two-year sentence indicates a particularly serious crime an irrebuttable one. Stanley Burnton LJ wrote:

I do not think that every crime that is punished with a sentence of 2 years imprisonment is particularly serious. One only has to appreciate that determinate sentences may be many times longer than 2 years for it to be obvious that a sentence of 2 years’ imprisonment is not necessarily indicative of a particularly serious crime.

The suggestion that every crime which attracts a 12-month prison sentence is without exception “a particularly serious crime” is patently absurd.  But although this change would be incompatible with the Refugee Convention, the Bill would be primary legislation, which, in this country, cannot be struck down even where it is patently absurd and/or conflicts with the UK’s international obligations. It remains to be seen how the courts will deal with this problem.


It is hard to avoid the conclusion that the government is spoiling for a fight.

In principle it makes sense to translate secondary legislation that was created to implement EU law that the UK wishes to preserve into primary legislation, so there is not likely to be much controversy over most of the provisions described in the first section above (apart perhaps from some imaginative litigation inspired by creative black-letter lawyers and judges).  The only exception to that is likely to be the fact that the changes to the exemptions from penalties in clause 34 may have retrospective effect.

The rest of the changes are very likely to produce years of litigation, much as the government’s attempts to fix Article 8 claims to the detriment of applicants have done. To borrow Colin’s assessment, while this is good for lawyers, it is bad for everyone else.

Download a more detailed version of this article in Word format here.

Interested in refugee law? You might like Colin's book, imaginatively called "Refugee Law" and published by Bristol University Press.

Communicating important legal concepts in an approachable way, this is an essential guide for students, lawyers and non-specialists alike.

Relevant articles chosen for you
Picture of Rudolph Spurling

Rudolph Spurling

Rudolph started law-school at the age of 41 and re-trained as a barrister, since when he has practised in immigration law, focussing mainly on asylum and human rights at One Pump Court in London. He is a regular fixture in the First-tier and Upper Tribunals and appeared in the current country guidance cases on the persecution risks for returnees to post-civil war Sri Lanka (2013) and Iranian Kurds (2018). Before coming to the Bar, he had a varied work history as a freelance lexicographer, research assistant on the Survey of English Usage at University College London (1989-90), Teacher of English and the methodology of teaching in China (1991-1994), and a local government officer at the London Boroughs of Camden (1995-6) and Hammersmith & Fulham (1996-2005).


4 Responses

  1. Thank you for this. Lots of questions I had are addressed (and can only be answered by your conclusions about the purpose of the legislation).