United Kingdom House of Lords Decisions

You are here: BAILII >> Databases >> United Kingdom House of Lords Decisions >> Islam v. Secretary of State for the Home Department Immigration Appeal Tribunal and Another, Ex Parte Shah, R v. [1999] UKHL 20; [1999] 2 AC 629; [1999] 2 All ER 545 (25th March, 1999)
URL: http://www.bailii.org/uk/cases/UKHL/1999/20.html
Cite as: [1999] INLR 144, [1999] 2 AC 629, [1999] 2 WLR 1015, [1999] Imm AR 283, [1999] UKHL 20, 6 BHRC 356, [1999] 2 All ER 545

[New search] [Buy ICLR report: [1999] 2 AC 629] [Buy ICLR report: [1999] 2 WLR 1015] [Help]

Islam v. Secretary of State for the Home Department Immigration Appeal Tribunal and Another, Ex Parte Shah, R v. [1999] UKHL 20; [1999] 2 AC 629; [1999] 2 All ER 545 (25th March, 1999)

HOUSE OF LORDS

Lord Steyn Lord Hoffmann Lord Hope of Craighead
Lord Hutton Lord Millett

OPINIONS OF THE LORDS OF APPEAL FOR JUDGMENT IN THE CAUSE

ISLAM (A.P.)
(APPELLANT)

SECRETARY OF STATE FOR THE HOME DEPARTMENT
(RESPONDENT)

REGINA

IMMIGRATION APPEAL TRIBUNAL AND ANOTHER
(RESPONDENTS)
EX PARTE SHAH (A.P.)
(CONJOINED APPEALS)

ON 25 MARCH 1999

LORD STEYN

My Lords,

The two appeals before the House raise important questions about the interpretation of article 1A(2) of the Convention Relating to the Status of Refugees, 1951, and in particular the meaning of the words "membership of a particular social group." Section 8(2) of the Asylum and Immigration Appeals Act 1993, provides that a person who has limited leave to enter the United Kingdom may appeal to a special adjudicator against a refusal to vary leave "on the ground that it would be contrary to the United Kingdom's obligations under the Convention for him to be required to leave the United Kingdom after the time limited by the leave." The common features of the two appeals are as follows. Both appeals involve married Pakistani women, who were forced by their husbands to leave their homes. They are at risk of being falsely accused of adultery in Pakistan. They are presently in England. They seek asylum in this country as refugees. They contend that, if they are forced to return to Pakistan, they would be unprotected by the state and would be subject to a risk of criminal proceedings for sexual immorality. If found guilty the punishment may be flogging or stoning to death. In these circumstances both women claim refugee status on a ground specified in article 1A(2) of the Convention, namely that they have a well founded fear of being persecuted for reasons of "membership of a particular social group." The Court of Appeal rejected these claims: Reg. v. Immigration Appeal Tribunal, Ex-parte Shah [1998] 1 W.L.R. 74. Both women have been granted exceptional leave to remain in the United Kingdom. But both women still seek refugee status. The principal question of law is whether the appellants are members of a particular social group within the meaning of article 1A(2) of the Convention. This question can only be considered against a close and particular focus on the facts of the case.

Women in the Islamic Republic of Pakistan

Generalisations about the position of women in particular countries are out of place in regard to issues of refugee status. Everything depends on the evidence and findings of fact in the particular case. On the findings of fact and unchallenged evidence in the present case, the position of women in Pakistan is as follows. Notwithstanding a constitutional guarantee against discrimination on the grounds of sex a woman's place in society in Pakistan is low. Domestic abuse of women and violence towards women is prevalent in Pakistan. That is also true of many other countries and by itself it does not give rise to a claim to refugee status. The distinctive feature of this case is that in Pakistan women are unprotected by the state: discrimination against women in Pakistan is partly tolerated by the state and partly sanctioned by the state. Married women are subordinate to the will of their husbands. There is strong discrimination against married women, who have been forced to leave the matrimonial home or have simply decided to leave. Husbands and others frequently bring charges of adultery against such wives. Faced with such a charge the woman is in a perilous position. Similarly, a woman who makes an accusation of rape is at great risk. Even Pakistan statute law discriminate against such women. The position is described in a report of Amnesty International dated 6 December 1995 on Women in Pakistan. The report states, at pp. 5-7:

Zina

Hudood

Zina

hadd

"In a rape case the onus of proof falls on the victim. If a woman fails to prove that she did not give her consent to intercourse, the court may convict her of illicit sexual intercourse. . . .

Hudood

ta'zir

hadd

Zina;

Zina

"Most women remain in jail for two to three years before their cases are decided, often on the basis of no evidence of any offence."

For what may be a small minority, who are convicted of sexual immorality, there is the spectre of 100 lashes in public or stoning to death in public. This brief description of the discrimination against women, which is tolerated and sanctioned by the state in Pakistan, is the defining factual framework of this case.

The Shah case

The appellant is 43. Her husband turned her out of the marital home in Pakistan. She arrived in the United Kingdom in 1992 and gave birth to a child shortly thereafter. In June 1993 she claimed asylum. She is afraid that her violent husband may accuse her of adultery and may assault her or denounce her under Sharia law for the offence of sexual immorality. In her case the evidence of state toleration and sanctioning of discrimination against women was sketchy.

This claim was rejected on the ground that the appellant does not come within "a particular social group" under article 1A(2). The appellant appealed to the special adjudicator. On 25 July 1995 the special adjudicator found that the appellant's fear of persecution was well founded. But she concluded that the appellant does not fall within "a particular social group", being the only conceivable ground for her Convention claim. On 7 August 1995 the Immigration Appeal Tribunal refused leave to appeal on the ground "that the adjudicator gave clear adverse findings of fact, after giving to each element on the evidence the weight she considered appropriate." The appellant sought judicial review of the refusal of leave. The substantive hearing took place before Sedley J. (now Sedley L.J.). The Secretary of State conceded that the I.A.T. had misdirected itself but contended that relief should be denied because the claim to refugee status is as a matter of law unsustainable. Sedley J. held that the appellant's case is arguable. He granted an order directing the I.A.T. to grant leave and to hear and determine the appeal: Reg. v. Immigration Appeal Tribunal and Secretary of State for the Home Department, Ex parte Shah [1997] Imm.A.R. 145.

The Islam case

The appellant is 45 and has two children. She arrived with her children in the United Kingdom in 1991. In the same year she claimed asylum. She is a teacher. She married her husband in 1971. He was often violent towards her. But the marriage endured. In 1990 a fight broke out in the school where she was teaching. The fight was between young supporters of two rival political factions. She intervened. One faction became hostile towards her. They made allegations of infidelity against her. These allegations were made, inter alia, to her husband who was a supporter of the same faction. Her husband assaulted her and she was twice admitted to hospital. She left her husband. She stayed briefly at her brother's house. Unknown men threatened her brother. She could not remain with him. After a brief stay in a temporary refuge she came to the United Kingdom. In claiming asylum she relied on two Convention grounds under article 1A(2), namely a well founded fear of persecution for reasons of (1) membership of a particular social group and (2) political opinion.

The claim was rejected. By a determination dated 7 December 1995 the special adjudicator accepted the evidence of the appellant. She found that the appellant had been persecuted in Pakistan. She also found that the authorities in Pakistan are both unable and unwilling to protect the appellant. But she held that as a matter of law the appellant was not a member of a "particular social group" because the group could not exist independently of the feared persecution. Moreover, she found that on the facts neither the particular political faction nor the appellant's husband persecuted her because of an actual or perceived political opinion. By a determination of 2 October 1996 the I.A.T. dismissed the appeal. The I.A.T. found that the appellant cannot be said to belong to a particular social group because the "sub-group does not . . . have any innate or unchangeable characteristic, nor is it a cohesive homogeneous group whose members are in close voluntary association." The I.A.T. further concluded that on the facts the appellant's persecution was motivated neither by an actual nor attributed political opinion.

The decision of the Court of Appeal

Both women appealed to the Court of Appeal. The appeals were held together. In separate and careful judgments the Court of Appeal dismissed both appeals: Regina v. Immigration Appeal Tribunal and Another, Ex parte Shah [1998] 1.W.L.R. 74. Given that those judgments are reported, it will be sufficient to state in outline effect the judgments. All three members of the Court of Appeal found that the appeal in the case of Islam, so far as it was based on persecution on the ground of political opinion, failed on the facts. The principal issue revolved round the question of law whether appellants could claim to be members of the "particular social group" under article 1A(2). Waite L.J. based his decision on the ground that independently of the feared persecution there was no common uniting attribute which could entitle the appellants to the status of "membership of a particular social group" under article 1A(2): pp. 86A-87B. Staughton L.J. went further. He held that that what is required is a number of people "joined together with some degree of cohesiveness, co-operation and interdependence": at p. 93D. And this requirement was not satisfied. Henry L.J. agreed with the ground on which Waite L.J. decided the matter. It is not clear whether there was a second ground for his decision. Henry L.J. agreed with Waite L.J. that "cohesion" was not "not necessary in every case": p. 91H. Henry L.J. added that "it is not necessary where the particular social group is recognised as such by the public, though is not organized . . ." It would seem that Henry L.J. contemplated that cohesiveness is sometimes a requirement.

Article óKóóK1A(2) in the scheme of the Convention

The critical and operative provision of the Convention is article 1A(2). It provides as follows:

"For the purposes of the present Convention, the term 'refugee' shall apply to any person who: . . . (2) . . . owing to well founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable or, owing to such fear, is unwilling to avail himself of the protection of that country; . . ."

In order to qualify as a refugee the asylum seeker (assumed to be a woman) must therefore prove: (1) That she has a well founded fear of persecution. (2) That the persecution would be for reasons of race, religion, nationality, membership of a particular social group, or political opinion. (3) That she is outside the country of her nationality. (4) That she is unable, or owing to fear, unwilling to avail herself of the protection of that country.

Article 1(F) provides for a number of circumstances in which the Convention does not apply including the case where the asylum seeker has previously committed a serious non-political crime outside the country of refuge: see paragraph (b). Article 1(C) provides for the termination of the application of the Convention due to a change of the circumstances in which she was recognised as a refugee, e.g. a material change of government. Article 32 provides for a right of expulsion of a refugee on the grounds of national security or public order: see also article 33.

In the search for the correct interpretation of the words "membership of a particular social group" the travaux preparatoires of the Convention are uninformative. The words in question were introduced at a late stage of the process leading to the finalization of the Convention. That fact tells one nothing about their contextual meaning. But the preambles to the Convention are significant. I set out the relevant preambles:

Considering

The relevance of the preambles is twofold. First, they expressly show that a premise of the Convention was that all human beings shall enjoy fundamental rights and freedoms. Secondly, and more pertinently, they show that counteracting discrimination, which is referred to in the first preamble, was a fundamental purpose of the Convention. That is reinforced by the reference in the first preamble to the Universal Declaration of Human Rights, 1948, which proclaimed the principle of the equality of all human beings and specifically provided that the entitlement to equality means equality "without distinction of any kind, such as race, colour, sex, language, religion, political or other opinion, national or social origin, property, birth or other status.": see articles 1 and 2.

Narrowing the issue

Putting to one side the separate question whether the appellant in the Islam case can rely on the Convention ground of political opinion, the principal issue before the House is the meaning and application of the words "membership of a particular social group." It is accepted that each appellant has a well founded fear of persecution in Pakistan if she is returned to that country. The appellants are outside the country of their nationality. And they are unable to avail themselves of the protection of Pakistan. On the contrary, it is an unchallenged fact that the authorities in Pakistan are unwilling to afford protection for women circumstanced as the appellants are. Except for the requirements inherent in the words "persecution for reasons of . . . membership of a particular social group" in article 1A(2) all the conditions of that provision are satisfied. Two issues remain: (1) Do the women satisfy the requirement of "membership of a particular social group?" (2) If so, a question of causation arises, namely whether their fear of persecution is "for reasons of" membership of a particular social group. I will now concentrate on the first question. It is common ground that there is a general principle that there can only be a "particular social group" if the group exists independently of the persecution. In A. v. Minister for Immigration and Ethnic Affairs and Another (1997) 142 A.L.R. 331, 358 McHugh J. neatly explained the point as follows:

". . . If it were otherwise, Art. 1(A)(2) would be rendered illogical and nonsensical. It would mean that persons who had a well founded fear of persecution were members of a particular social group because they feared persecution. The only persecution that is relevant is persecution for reasons of membership of a group which means that the group must exist independently of, and not be defined by, the persecution . . ."

In other words relying on persecution to prove the existence of the group would involve circular reasoning. It is therefore unsurprising that counsel for the appellants and counsel for the United Nations High Commissioner for Refugees (UNHCR) accept the general principle that there can only be a "particular social group" if it exists independently of the persecution.

The first issue: is cohesiveness a requirement for the existence of a particular social group?

Before the Court of Appeal [1998] 1 W.L.R. 74 counsel for the Secretary of State submitted that "there is a need for the group to be homogeneous and cohesive." On that occasion counsel said that the adjective "social" refers to persons who are "interdependent or co-operative": at p. 85. This argument persuaded Staughton L.J. to rule that as a matter of law a particular social group can only exist if there is "some degree of cohesiveness, co-operation or interdependence": at 93D. If this ruling is right, the arguments of the appellants fail at the first hurdle. There is some authority for this view. The origin of the idea appears to be the decision of the United States Court of Appeals, Ninth Circuit, in Sanchez-Trujillo v. Immigration and Naturalization Service (1986) 801 F.2d 1571. This case involved young, working class Salvadoran males, who failed to do military service in El Salvador. They claimed that if they were repatriated to El Salvador they would be persecuted. They contended that they were members of a particular social group. The court held that "particular social group" implies a collection of people closely affiliated with each other: at 1576. The claimants to refugee status did not meet this standard and failed on this ground. But they also failed on the anterior ground that they were unable to demonstrate that the government of El Salvador had singled the alleged group out for persecution. On the contrary, the court found that the risk of persecution related to the existence of actual or imputed political opinion, which was found to turn on individual circumstances. This decision has been followed on the same Circuit in De Valle v. Immigration and Naturalization Service (1990) 901 F.2d 787, 793. Counsel for the Secretary of State suggested that the Court of Appeals for the Second Circuit followed Sanchez-Trujillo in Gomez v. Immigration and Naturalization Service (1990) 947 F.2d 660, 664. For my part I found the passage relied on equivocal.

In any event, on circuits other than the Ninth Circuit, a less restrictive interpretation of the words "particular social group" has been adopted. The foundation of the contrary view is the earlier decision of the Board of Immigration Appeals in In re Acosta (1985) 19 I. & N. 211. This decision was not mentioned in Sanchez-Trujillo. In Acosta the Board dismissed the claim of a collection of Salvadoran taxi drivers who allegedly feared persecution from an organised group of taxi drivers in El Salvador. The reasoning is important. The Board observed:

the Act

"Applying the doctrine of ejusdem generis, we interpret the phrase 'persecution on account of membership in a particular social group' to mean persecution that is directed toward an individual who is a member of a group of persons all of whom share a common, immutable characteristic. The shared characteristic might be an innate one such as sex, color, or kinship ties, or in some circumstances it might be a shared past experience such as former military leadership or land ownership. The particular kind of group characteristic that will qualify under this construction remains to be determined on a case-by-case basis. . . . By construing 'persecution on account of membership in a particular social group' in this manner, we preserve the concept that refuge is restricted to individuals who are either unable by their own actions, or as a matter of conscience should not be required, to avoid persecution."

Support for this approach is to be found in a number of United States decisions: see Matter of Toboso-Alfonso, (unreported) 12 March 1990, Interim Decision, cited in Re: G.J. [1998] 1 N.L.R. 387, 418; Bastanipour v. Immigration and Naturalization Service (1992) 980 F.2d 1129; Fatin v. Immigration and Naturalization Service (1993) 12 F.3d 1233; Matter of Kasinga 13 June 1996, Interim Decision 3278, reported in International Journal of Refugee Law 1997, at pp. 213-234. It is therefore clear that there are divergent streams of authority in the United States. And it may be right to say that the preponderance of U.S. case law does not support Sanchez-Trujillo.

Counsel for the Secretary of State also tried to rely on dicta by two members of the majority in the High Court of Australia in A. v. Minister of Immigration and Ethnic Affairs (1997) 142 A.L.R. 331. This case involved a claim to refugee status by a husband and wife who had come from China to Australia. They said that they feared sterilization under the "one child policy" of China if they were returned. The majority (Dawson, McHugh and Gummow J.J.) rejected the argument that the appellants were members of a particular social group. Brennan C.J. and Kirby J. dissented. Contrary to counsel's submission I consider it clear that Dawson J. did not accept the Sanchez-Trujillo theory. On the contrary, he said that Sanchez-Trujillo is unpersuasive so far as it suggested that "the uniting particular must be voluntary": at p. 341. Gummow J. may have adopted the Sanchez-Trujillo principle. And McHugh J. and the dissenting judges took a broader view. In any event, in a case such as A. v. Minister of Immigration and Ethnic Affairs a significant difficulty in the way of claimants to refugee status is the fact that the one child policy is apparently applied uniformly in China. There is no obvious element of discrimination. That may be the true basis of the decision of the Australian High Court. Far from assisting the argument of the Secretary of State the trend of the dicta in A.'s case (except for the observations of Gummow J.) is against a requirement of cohesiveness. Moreover, in Canada the Supreme Court has adopted a broader approach which depends on the reasoning in Acosta and is inconsistent with Sanchez-Trujillo: see Attorney-General of Canada v. Ward (1993) 103 D.L.R. (4th) 1; see also Chan v. Minister of Employment and Immigration 128 D.L.R. (4th) 213, 247. This is made explicit by La Forest J. in Ward. He said, at p. 34 that "social group" could include individuals fearing persecution on "such bases as gender, linguistic background and sexual orientation."

Apart from the judgment of Staughton L.J. in the present case, there is no English authority for the view that cohesiveness is an indispensable requirement for the existence of a "particular social group." Counsel for the Secretary of State cited the decision of the Court of Appeal in Savchenko v. Secretary of State for the Home Department [1996] Imm AR 28. The judgments in that case contain references to Sanchez-Trujillo but no adoption of its reasoning on the element of cohesiveness. The ratio of Savchenko is that the alleged group (Russian security guards at a hotel who feared victimisation by the mafia) did not exist independently of the persecution. But MacCowan L.J. recorded, at p. 34:

"The Secretary of State submits, we were told by Mr. Pannick, that the concept of membership of a particular social group covers persecution in three types of case: (1) membership of a group defined by some innate or unchangeable characteristic of its members analogous to race, religion, nationality or political opinion, for example, their sex, linguistic background, tribe, family or class; (2) membership of a cohesive, homogeneous group whose members are in a close voluntary association for reasons which are fundamental to their rights, for example, a trade union activist; (3) former membership of a group covered by (2)."

Paragraph (1) is in line with Acosta's case (1985) 19 I. & N. 211 and inconsistent with Sanchez-Trujillo's case (1986) 801 F.2d. 1571. It was not explicitly adopted by the Court of Appeal but it was also not rejected. In these circumstances Savchenko's case [1996] ImmAR 28 cannot assist the argument of the Secretary of State. Counsel for the Secretary of State informed the House that the Secretary of State no longer supports his submission in paragraph (1). That is understandable: his earlier submission weakens his case on the present appeals.

The support in the case law for the Sanchez-Trujillo view is slender. In the literature on the subject there is no support: see the criticism in Hathaway, The Law of Refugee Status, (1991), at p. 161 (note 182.) Considering that view on its merits I am satisfied that for the reasons given in Acosta's case the restrictive interpretation of "particular social group" by reference to an element of cohesiveness is not justified. In 1951 the draftsmen of article 1A(2) of the Convention explicitly listed the most apparent forms of discrimination then known, namely the large groups covered by race, religion, and political opinion. It would have been remarkable if the draftsmen had overlooked other forms of discrimination. After all, in 1948 the Universal Declaration had condemned discrimination on the grounds of colour and sex. Accordingly, the draftsmen of the Convention provided that membership of a particular social group would be a further category. It is not "an all-encompassing residual category": Hathaway, The Law of Refugee Status, 1991, at p. 159. Loyalty to the text requires that one should take into account that there is a limitation involved in the words "particular social group." What is not justified is to introduce into that formulation an additional restriction of cohesiveness. To do so would be contrary to the ejusdem generis approach so cogently stated in Acosta. The potential reach of the Acosta reasoning may be illustrated by the case of homosexuals in countries where they are persecuted. In some countries homosexuals are subjected to severe punishments including the death sentence. In Re G.J. [1998] 1 N.L.R. 387 the New Zealand Refugee Status Authority faced this question. Drawing on the case law and practice in Germany, The Netherlands, Sweden, Denmark, Canada, Australia and the U.S.A., the Refugee Status Authority concluded in an impressive judgment that depending on the evidence homosexuals are capable of constituting a particular social group with the meaning of article 1A(2): see pp. 412-422. This view is consistent with the language and purpose of article 1A(2). Subject to the qualification that everything depends on the state of the evidence in regard to the position of homosexuals in a particular country I would in principle accept the reasoning in Re G.J. as correct. But homosexuals are, of course, not a cohesive group. This is a telling point against the restrictive view in Sanchez-Trujillo's case. Finally, the restrictive interpretation is at variance with the principle that a treaty ought to be construed in a purposive sense: see article 31 of the Vienna Convention on the Law of Treaties.

Given the unequivocal acceptance by Staughton L.J. of the restrictive theory of the interpretation of article 1A(2) I have thought it right to explain at some length why in my view his conclusion was not justified. In oral argument counsel for the Secretary of State expressly conceded that Staughton L.J. erred in ruling that cohesiveness is an indispensable requirement. Instead counsel submitted that "particular social group" normally requires cohesiveness. What the practical implications of this qualification are I do not know. For my part the position is as follows. Cohesiveness may prove the existence of a particular social group. But the meaning of "particular social group" should not be so limited: the phrase extends to what is fairly and contextually inherent in that phrase.

The second issue: The different theories of "particular social group."

In oral argument different foundations for treating the two appellants as members of a particular social group were explored and tested. First, counsel for the appellants argued that three characteristics set the appellants apart from the rest of society viz gender, the suspicion of adultery, and their unprotected status in Pakistan. He submitted that this combination of characteristic exists independently of persecution. Secondly, while counsel UNHCR made no submissions on the merits of the cases of the appellants he placed before the House general submissions as to the meaning of "particular social group." He submitted that individuals who believe in or are perceived to believe in values and standards which are at odds with the social mores of the society in which they live may, in principle, constitute "a particular social group" within the meaning of article 1A(2). Women who reject those mores--or are perceived to reject them--are capable of constituting "a particular social group." The third way of approaching the matter was suggested in argument by my noble and learned friend Lord Hoffmann. It involves the proposition that women in Pakistan are a particular social group. óKóóKCounsel for the Secretary of State pointed out that this way of approaching the case is a new development. But it was thoroughly explored and tested in oral argument in the House. In these circumstances it must be considered on its merits. Indeed as the wider theory it seems right and convenient to examine it first.

Women in Pakistan as a Group

The idea so incisively put forward by Lord Hoffmann is neither novel nor heterodox. It is simply a logical application of the seminal reasoning in Acosta's case 19 I. & N. 211. Relying on an ejusdem generis interpretation the Board interpreted the words 'persecution on account of membership in a particular social group' to mean persecution "that is directed toward an individual who is a member of a group of persons all of whom share a common immutable characteristic." The Board went on to say that the shared characteristic might be an innate one "such as sex, color, or kinship ties." This reasoning covers Pakistani women because they are discriminated against and as a group they are unprotected by the state. Indeed the state tolerates and sanctions the discrimination. The analogy of discrimination against homosexuals who may in some countries be a "particular social group" supports this reasoning. What is the answer to this reasoning? It avoids any objection based on the principle that the group must exist dehors the persecution. The objection based on a requirement of cohesiveness foists an impermissible restrictive requirement on the words of article 1A(2). What then is left by way of counter-argument? Counsel for the Secretary of State said that there is a clear answer to this line of reasoning. That turned out to be the fact that some Pakistani women are able to avoid the impact of persecution, e.g. because their circumstances enable them to receive protection. In such cases there will be no well founded fear of persecution and the claim to refugee status must fail. But this is no answer to treating women in Pakistan as a relevant social group. After all, following the New Zealand judgment in Re G.J. [1998] 1 N.L.R. 387 I regard it as established that depending on the evidence homosexuals may in some countries qualify as members of a particular social group. Yet some homosexuals may be able to escape persecution because of their relatively privileged circumstances. By itself that circumstance does not mean that the social group of homosexuals cannot exist. Historically, under even the most brutal and repressive regimes some individuals in targeted groups have been able to avoid persecution. Nazi Germany, Stalinist Russia and other examples spring to mind. To treat this factor as negativing a Convention ground under article 1A(2) would drive a juggernaut through the Convention. My Lords, on careful reflection there is no satisfactory answer to the argument that the social group is women in Pakistan.

The narrower group

If I had not accepted that women in Pakistan are a "particular social group," I would have held that the appellants are members of a more narrowly circumscribed group as defined by counsel for the appellants. I will explain the basis of this reasoning briefly. It depends on the coincidence of three factors: the gender of the appellants, the suspicion of adultery, and their unprotected position in Pakistan. The Court of Appeal held (and counsel for the Secretary of State argued) that this argument falls foul of the principle that the group must exist independently of the persecution. In my view this reasoning is not valid. The unifying characteristics of gender, suspicion of adultery, and lack of protection, do not involve an assertion of persecution. The cases under consideration can be compared with a more narrowly defined group of homosexuals, namely practising homosexuals who are unprotected by a state. Conceptually such a group does not in a relevant sense depend for its existence on persecution. The principle that the group must exist independently of the persecution has an important role to play. But counsel for the Secretary of State is giving it a reach which neither logic nor good sense demands. In A. v. Minister for Immigration and Ethnic Affairs 142 A.L.R. 331, 359 McHugh J. explained the limits of the principle. He said:

"Nevertheless, while persecutory conduct cannot define the social group, the actions of the persecutors may serve to identify or even cause the creation of a particular social group in society. Left-handed men are not a particular social group. But, if they were persecuted because they were left-handed, they would no doubt quickly become recognisable in their society as a particular social group. Their persecution for being left-handed would create a public perception that they were a particular social group. But it would be the attribute of being left-handed and not the persecutory acts that would identify them as a particular social group."

The same view is articulated by Goodwin-Gill, The Refugee in International Law, 2nd ed., (1996) at p. 362. I am in respectful agreement with this qualification of the general principle. I would hold that the general principle does not defeat the argument of counsel for the appellants.

My Lords, it is unchallenged that the women in Pakistan are unprotected by state and public authorities if a suspicion of adultery falls on them. The reasoning in Acosta, which has been followed in Canada and Australia, is applicable. There are unifying characteristics which justify the conclusion that women such as the appellants are members of a relevant social group. On this additional ground I would hold that the women fall within the scope of the words "particular social group."

The third issue: The causation test

Having concluded on a two-fold basis that the appellants are within the scope of the words "particular social group," it is necessary to consider whether they have a well founded fear of being persecuted "for reasons of" their membership of the group in question. A question of causation is involved. Here a further legal issue arose. Counsel for the appellants argued that a 'but for' test is applicable. He relied on the adoption of such a test in the sex discrimination field: see James v. Eastleigh Borough Council [1990] 2 AC 751; and compare Hathaway, The Law of Refugee Status, 1991, at 140. Counsel for the Secretary of State challenged this submission. He argued that in the different context of issues of refugee status the test of effective cause--and there may be more than one effective cause--is the correct one. In the present case it makes no difference which test is applied. It matters not whether causation is approached from the vantage point of the wider or narrower social group I have identified. In either event it is plain that the admitted well founded fear of the two women is "for reasons" of their membership of the social group. Given the central feature of state-tolerated and state-sanctioned gender discrimination, the argument that the appellants fear persecution not because of membership of a social group but because of the hostility of their husbands is unrealistic. And that is so irrespective whether a "but for" test, or an effective cause test, is adopted. In these circumstances the legal issue regarding the test of causation, which did not loom large on this appeal, need not be decided.

The view of UNHCR

My Lords, Mr. Peter Duffy Q.C., counsel for the UNHCR, placed before the House all the relevant background materials and produced a valuable written review supplemented by helpful oral argument. Except to point out that the UNHCR view is wider than the grounds upon which I have reached my conclusion I do not propose to express a concluded view on the UNHCR position. My diffidence on this point is reinforced by an observation by Sedley J. in Shah. Commenting on the unique complexity of such issues Sedley J. said [1997] Imm.A.R. 145, 153:

"Its adjudication is not a conventional lawyer's exercise of applying a legal litmus test to ascertain facts; it is a global appraisal of an individual's past and prospective situation in a particular cultural, social, political and legal milieu, judged by a test which, though it has legal and linguistic limits, has a broad humanitarian purpose."

Political opinion

In the Islam case there was also a discrete issue as to whether the appellant can rely on the Convention ground of political opinion. Given my conclusions this issue falls away. Nevertheless, I must make clear that I was not attracted by this argument. The special adjudicator and the I.A.T. decided this issue against the appellant. The findings of fact were open on the evidence. There were no misdirections. In agreement with all members of the Court of Appeal I regard this ground of appeal as unsustainable.

Conclusion

In the Islam case I would allow the appeal and make a declaration in accordance with section 8(2) of the Asylum and Immigration Appeals Act 1993 that it would be contrary to the United Kingdom's obligations for her to be required to leave the United Kingdom. In the Shah appeal I would allow the appeal to the extent of setting aside the order of the Court of Appeal and restoring the order of Sedley J. remitting her case to the Immigration Appeal Tribunal.

LORD HOFFMANN

My Lords,

In Pakistan there is widespread discrimination against women. Despite the fact that the constitution prohibits discrimination on grounds of sex, an investigation by Amnesty International at the end of 1995 reported that government attempts to improve the position of women had made little headway against strongly entrenched cultural and religious attitudes. Women who were victims of rape or domestic violence often found it difficult to obtain protection from the police or a fair hearing in the courts. In matters of sexual conduct, laws which discriminated against women and carried severe penalties remained upon the statute book. The International Bar Association reported in December 1998 that its mission to Pakistan earlier in the year "heard and saw much evidence that women in Pakistan are discriminated against and have particular problems in gaining access to justice." (Report on Aspects of the Rule of Law and Human Rights in the Legal System of Pakistan, p. 29).

These appeals concern two women who became victims of domestic violence in Pakistan, came to the United Kingdom and claimed asylum as refugees. Shahanna Islam is a graduate school teacher from Karachi. In 1990 she became involved in a playground dispute between rival gangs of politically motivated boys. Those supporting the Mohaijur Quami Movement or "MQM" told her husband, who belonged to the same party, that she had been unfaithful to him. As a result he gave her severe beatings which eventually drove her out of the house. The other woman, Syeda Shah is simple and uneducated. She was frequently beaten by her husband and eventually, when pregnant, turned out of the house. She too came to the United Kingdom, where her child was born.

Both women were given limited leave to enter the United Kingdom as visitors. Afterwards they claimed the right to remain as refugees under the 1951 Geneva Convention relating to the Status of Refugees as amended by the 1967 Protocol. The United Kingdom is a party to this Convention, which has been incorporated into domestic law by section 8 of the Asylum and Immigration Appeals Act 1993. Subsection (2) provides that a person who has limited leave to enter the United Kingdom may appeal to a special adjudicator against a refusal to vary the leave "on the ground it would be contrary to the United Kingdom's obligations under the Convention for him to be required to leave the United Kingdom after the time limited by the leave." Both accordingly appealed to a special adjudicator.

The Convention defines a refugee in article 1A(2) as a person who:

"owing to well founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable, or owing to such fear is unwilling, to avail himself of the protection of that country . . ."

The question for the special adjudicators was whether Mrs Islam and Mrs Shah came within this definition.

Domestic violence such as was suffered by Mrs Islam and Mrs Shah in Pakistan is regrettably by no means unknown in the United Kingdom. It would not however be regarded as persecution within the meaning of the Convention. This is because the victims of violence would be entitled to the protection of the state. The perpetrators could be prosecuted in the criminal courts and the women could obtain orders restraining further molestation or excluding their husbands from the home under the Domestic Violence and Matrimonial Proceedings Act 1976. What makes it persecution in Pakistan is the fact that according to evidence which was accepted by the special adjudicator in Mrs Islam's case and formed the basis of findings which have not been challenged, the State was unwilling or unable to offer her any protection. The adjudicator found it was useless for Mrs Islam, as a woman, to complain to the police or the courts about her husband's conduct. On the contrary, the police were likely to accept her husband's allegations of infidelity and arrest her instead. The evidence of men was always deemed more credible than that of women. If she was convicted of infidelity, the penalties could be severe. Even if she was not prosecuted, as a woman separated from her husband she would be socially ostracised and vulnerable to attack, even murder, at the instigation of her husband or his political associates. The special adjudicator said:

"On the evidence, the agents of persecution are the MQM boys who made false allegations against her to her husband, and/or her husband who had subjected her to violence. In order for them to be regarded as agents of persecution the appellant has to show that the authorities in Pakistan were unable or unwilling to offer her protection. It is the appellant's case that in her particular circumstances, given the structure of society and the attitude of the authorities towards domestic violence and given the impunity with which MQM members have acted and still act in Pakistan, that the authorities in Pakistan are both unable and unwilling to offer her protection. I find on the evidence that this is indeed the case."

The Immigration Appeal Tribunal summed up her position as follows:

"She cannot return to her husband. She cannot live anywhere in Pakistan without male protection. She cannot seek assistance from the authorities because in Pakistan society women are not believed or they are treated with contempt by the police. If she returns she will be abused and possibly killed."

In the case of Mrs Shah, the evidence of the legal and institutional background was much more sketchy. She said that she was afraid that if she returned to Pakistan her husband would deny paternity of the child to which she had given birth in England and either assault her himself or charge her with immorality before a religious tribunal. In her case, the special adjudicator found that she was simply a battered wife. Although as a matter of ordinary language her husband might be said to have persecuted her, it was not persecution within the meaning of the Convention.

The question in both cases was therefore whether the women had a well founded fear of persecution within the meaning of the Convention and, critically, whether such persecution was for one of the five enumerated reasons, namely, "race, religion, nationality, membership of a particular social group or political opinion." Of these, the only serious candidate for consideration was that they feared persecution because they were members of a "particular social group." There was an attempt to argue that Mrs Islam was being persecuted by the MQM and her husband for her political opinions, but I agree with my noble and learned friend Lord Steyn that this was not made out on the evidence.

The problem for both women was to specify the "social group" of which they claimed their membership had given rise to persecution. Mrs Shah's counsel seems to have tried to persuade the special adjudicator that "women who had suffered domestic violence" were a social group. This submission was rejected and the application dismissed. Her application to the Immigration Appeal Tribunal for leave to appeal was refused on the ground that the special adjudicator had found against her on the facts. She moved for judicial review before Sedley J., where counsel for the Home Office conceded that this reason was a bad one: [1997] Imm.A.R. 145. The special adjudicator had entirely accepted her version of the facts but ruled that the reason for her apprehended persecution was not her membership of anything which could as a matter of law qualify as a "social group." Sedley J. then proceeded to consider whether she had any reasonable prospect of satisfying the tribunal that she was being persecuted for such a reason. Her counsel made several attempts to define the group: one was "women who are perceived to have transgressed Islamic mores" and another was "women rejected by their husbands on the ground of alleged adultery." Sedley J. was sceptical of both these formulations but said that there was nevertheless a sufficiently arguable case to go before the Tribunal, which would be entitled to hear evidence about the social and legal background which had not been before the special adjudicator. "Its adjudication", said the judge, at p. 153:

"is not a conventional lawyer's exercise of applying a legal litmus test to ascertained facts; it is a global appraisal of an individual's past and prospective situation in a particular cultural, social, political and legal milieu, judged by a test which, though it has legal and linguistic limits, has a broad humanitarian purpose."

He therefore made granted mandamus ordering the Tribunal to hear and determine the appeal.

Mrs Islam also argued before the special adjudicator that she feared persecution because she belonged to a social group defined as "Pakistani women subject to domestic violence, namely wife abuse." The special adjudicator found that such a group, defined by reference to its fear of persecution, could not constitute a social group for the purposes of the Convention. The argument was a circular one: if one belonged to a group because one shared a common fear of persecution, one could not be said to be persecuted because one belonged to that group. This decision was upheld by the Immigration Appeal Tribunal.

The Secretary of State appealed against the decision of Sedley J. and Mrs Islam appealed against the decision of the Immigration Appeal Tribunal. The Court of Appeal (Staughton, Waite and Henry L.JJ.) heard both appeals together. By this time, the definition of the social group had been greatly elaborated. Mr. Blake Q.C. who appeared, as he did before your Lordships, for the women, defined it as "Pakistani women . . . accused of transgressing social mores (in the instant case, adultery, disobedience to husbands) . . . who are unprotected by their husbands or other male relatives." This, he submitted, was not a group defined by its common fear of persecution. It had objective distinguishing features in its sex, isolation (being abandoned by the husband and having no male relative to turn to) and being ostracised because perceived to be deserving of condemnation by the community for infringement of the sexual code for woman. But the Court of Appeal [1998] 1 W.L.R. 74 held that these features (apart from sex) were all the product of the persecution itself. If one took away the persecution, then, as Waite L.J. said, at p. 87G, "the stigma and the isolation necessarily depart with them. They are not the independent attributes of a particular group." Staughton and Henry L.JJ. agreed but Staughton L.J., at p. 93D and possibly also Henry L.J. (the judgment is not altogether free from ambiguity on this point) also rejected the alleged social group on the additional ground that it lacked "cohesiveness, co-operation or interdependence." Its members were solitary individuals having no contact with each other. The Court of Appeal therefore allowed the appeal against the judgment of Sedley J. in favour of Mrs Shah and dismissed Mrs Islam's appeal from the Immigration Appeal Tribunal. Against those decisions, Mrs Shah and Mrs Islam appeal to your Lordship's House. In hearing the appeal, the House has been greatly assisted by the intervention of the United Nations High Commissioner for Refugees, who was represented by Mr. Peter Duffy Q.C., whose untimely death since the hearing in this appeal has deprived the Bar and the cause of human rights of one of its brightest talents,

The travaux preparatoires for the Geneva Convention shed little light on the meaning of "particular social group." It appears to have been added to the draft at the suggestion of the Swedish delegate, who said that "experience had shown that certain refugees had been persecuted because they belonged to particular social groups." It seems to me, however, that the general intention is clear enough. The preamble to the Convention begins with the words:

Considering

In my opinion, the concept of discrimination in matters affecting fundamental rights and freedoms is central to an understanding of the Convention. It is concerned not with all cases of persecution, even if they involve denials of human rights, but with persecution which is based on discrimination. And in the context of a human rights instrument, discrimination means making distinctions which principles of fundamental human rights regard as inconsistent with the right of every human being to equal treatment and respect. The obvious examples, based on the experience of the persecutions in Europe which would have been in the minds of the delegates in 1951, were race, religion, nationality and political opinion. But the inclusion of "particular social group" recognised that there might be different criteria for discrimination, in pari materiae with discrimination on the other grounds, which would be equally offensive to principles of human rights. It is plausibly suggested that the delegates may have had in mind persecutions in Communist countries of people who were stigmatised as members of the bourgeoisie. But the concept of a social group is a general one and its meaning cannot be confined to those social groups which the framers of the Convention may have had in mind. In choosing to use the general term "particular social group" rather than an enumeration of specific social groups, the framers of the Convention were in my opinion intending to include whatever groups might be regarded as coming within the anti-discriminatory objectives of the Convention.

The notion that the Convention is concerned with discrimination on grounds inconsistent with principles of human rights is reflected in the influential decision of the U.S. Board of Immigration Appeals in In re Acosta (1985) 19 I. & N. 211 where it was said that a social group for the purposes of the Convention was one distinguished by:

"an immutable characteristic . . . [a characteristic] that either is beyond the power of an individual to change or that is so fundamental to his identity or conscience that it ought not to be required to be changed."

This was true of the other four grounds enumerated in the Convention. It is because they are either immutable or part of an individual's fundamental right to choose for himself that discrimination on such grounds is contrary to principles of human rights.

It follows that I cannot accept that the term "particular social group" implies an additional element of cohesiveness, co-operation or interdependence. The fact that members of a group may or may not have some form of organisation or interdependence seems to me irrelevant to the question of whether it would be contrary to principles of human rights to discriminate against its members. Among the other four categories, "race" and "nationality" do not imply any idea of co-operation; "religion" and "political opinion" might, although it could be minimal. In the context of the Convention it seems to me a contingent rather than essential characteristic of a social group. In the opinion of Judge Beezer for the U.S. Court of Appeals in Sanchez Trujillo v. Immigration and Naturalization Service (9th Cir. 1986) 801 F.2nd 1571) it was said that "'particular social group' implies a collection of people closely affiliated with each other, who are actuated by some common impulse or interest." This remark has been taken up in some (but not all) other U.S. cases. It has however been rejected by the Supreme Court of Canada in Attorney-General of Canada v. Ward (1993) 103 D.L.R. (4th) 1 and the High Court of Australia in A. v. Minister for Immigration and Ethnic Affairs [1998] I.N.L.R. 1 I would reject it also. I agree with La Forest J. in the Ward case when he said (at p. 34) that "social group" could include individuals fearing persecution on "such bases as gender, linguistic background and sexual orientation." None of these implies any form of interdependence or co-operation.

To what social group, if any, did the appellants belong? To identify a social group, one must first identify the society of which it forms a part. In this case, the society is plainly that of Pakistan. Within that society, it seems to me that women form a social group of the kind contemplated by the Convention. Discrimination against women in matters of fundamental human rights on the ground that they are women is plainly in pari materiae with discrimination on grounds of race. It offends against their rights as human beings to equal treatment and respect. It may seem strange that sex (or gender) was not specifically enumerated in the Convention when it is mentioned in article 2 of the Universal Declaration of Human Rights. But the Convention was originally limited to persons who had become refugees as a result of events occurring before 1 January 1951. One can only suppose that the delegates could not think of cases before that date in which women had been persecuted because they were women. But the time limit was removed by the 1967 New York Protocol and the concept of a social group is in my view perfectly adequate to accommodate women as a group in a society that discriminates on grounds of sex, that is to say, that perceives women as not being entitled to the same fundamental rights as men. As we have seen, La Forest J. in the Ward case had no difficulty in saying that persecution on grounds of gender would be persecution on account of membership of a social group. I therefore think that women in Pakistan are a social group.

As we have seen, however, the appellants in the Court of Appeal did not say that they feared persecution simply on the ground that they were women. They produced a much more restricted and complicated definition of the social group to which they claimed to belong and membership of which was said to be the ground for their persecution. In so doing, they introduced into the definition elements which the Court of Appeal regarded as arbitrary except by reference to the persecution they feared. Thus they found that the parts of the definition which restricted the group to anything narrower than the entire sex were essentially circular and incapable of defining a group for the purposes of the Convention.

The reason why the appellants chose to put forward this restricted and artificial definition of their social group was to pre-empt the question of whether their feared persecution was "for reasons of" their membership of the wider group of women. It was argued for the Secretary of State that they could not fear persecution simply for the reason that they were women. The vast majority of women in Pakistan conformed to the customs of their society, did not chafe against discrimination or have bullying husbands, and were not persecuted. Being a woman could not therefore be a reason for persecution. The question is essentially one of causation. Being a woman does not necessarily result in persecution and therefore cannot be the reason for those cases in which women are persecuted. The appellants' argument in the Court of Appeal accepted this reasoning and tried to confess and avoid by opting for a sub-category of women.

I do not need to express a view about whether this strategy should have succeeded because, as I shall explain in a moment, I think that the argument on causation which it was designed to meet is fallacious. The question is therefore capable of being given a much simpler answer. The strategy probably derives from conclusion 39 "Refugees, Women and International Protection" adopted by the Executive Committee of the United Nations High Commission for Refugees in 1985, which read as follows:

". . . [S]tates in the exercise of their sovereignty, are free to adopt the interpretation that women asylum seekers who face harsh or inhuman treatment due to their having transgressed social mores of the society in which they live may be regarded as a 'particular social group' within the meaning of article 1 A(2) of the 1951 United Nations Refugee Convention."

This was a well-meaning attempt to encourage a more liberal treatment of women refugees, who frequently did not conform to the standard characteristics of the male refugees, fleeing for racial or political reasons, with which national authorities were familiar. But I think that, whether right or wrong, it unnecessarily overcomplicates the matter.

I turn, therefore, to the question of causation. What is the reason for the persecution which the appellants fear? Here it is important to notice that it is made up of two elements. First, there is the threat of violence to Mrs Islam by her husband and his political friends and to Mrs Shah by her husband. This is a personal affair, directed against them as individuals. Secondly, there is the inability or unwillingness of the State to do anything to protect them. There is nothing personal about this. The evidence was that the State would not assist them because they were women. It denied them a protection against violence which it would have given to men. These two elements have to be combined to constitute persecution within the meaning of the Convention. As the Gender Guidelines for the Determination of Asylum Claims in the UK (published by the Refugee Women's Legal Group in July 1988) succinctly puts it (at p. 5): "Persecution = Serious Harm + The Failure of State Protection."

Answers to questions about causation will often differ according to the context in which the question is asked. (See Environment Agency (formerly National Rivers Authority) v. Empress Car Co. (Abertillery) Ltd. [1998] 2 WLR 350). Suppose oneself in Germany in 1935. There is discrimination against Jews in general, but not all Jews are persecuted. Those who conform to the discriminatory laws, wear yellow stars out of doors and so forth can go about their ordinary business. But those who contravene the racial laws are persecuted. Are they being persecuted on grounds of race? In my opinion, they plainly are. It is therefore a fallacy to say that because not all members of a class are being persecuted, it follows that persecution of a few cannot be on grounds of membership of that class. Or to come nearer to the facts of the present case, suppose that the Nazi government in those early days did not actively organise violence against Jews, but pursued a policy of not giving any protection to Jews subjected to violence by neighbours. A Jewish shopkeeper is attacked by a gang organised by an Aryan competitor who smash his shop, beat him up and threaten to do it again if he remains in business. The competitor and his gang are motivated by business rivalry and a desire to settle old personal scores, but they would not have done what they did unless they knew that the authorities would allow them to act with impunity. And the ground upon which they enjoyed impunity was that the victim was a Jew. Is he being persecuted on grounds of race? Again, in my opinion, he is. An essential element in the persecution, the failure of the authorities to provide protection, is based upon race. It is true that one answer to the question "Why was he attacked?" would be "because a competitor wanted to drive him out of business." But another answer, and in my view the right answer in the context of the Convention, would be "he was attacked by a competitor who knew that he would receive no protection because he was a Jew."

In the case of Mrs Islam, the legal and social conditions which according to the evidence existed in Pakistan and which left her unprotected against violence by men were discriminatory against women. For the purposes of the Convention, this discrimination was the critical element in the persecution. In my opinion, this means that she feared persecution because she was a woman. There was no need to construct a more restricted social group simply for the purpose of satisfying the causal connection which the Convention requires.

Mr. Blake, in supporting this argument, suggested that the requirement of causation could be satisfied by applying a "but for" test. If they would not have feared persecution but for the fact that they were women, then they feared persecution for reason of being women. I think that this goes from overcomplication to oversimplification. Once one has established the context in which a causal question is being asked, the answer involves the application of common sense notions rather than mechanical rules. I can think of cases in which a "but for" test would be satisfied but common sense would reject the conclusion that the persecution was for reasons of sex. Assume that during a time of civil unrest, women are particularly vulnerable to attack by marauding men, because the attacks are sexually motivated or because they are thought weaker and less able to defend themselves. The government is unable to protect them, not because of any discrimination but simply because its writ does not run in that part of the country. It is unable to protect men either. It may be true to say women would not fear attack but for the fact that they were women. But I do not think that they would be regarded as subject to persecution within the meaning of the Convention. The necessary element of discrimination is lacking. (Compare Gomez v. Immigration and Naturalization Service (1991) 947 F.2d 660).

I am conscious, as the example which I have just given will suggest, that there are much more difficult cases in which the officers of the State neither act as the agents of discriminatory persecution nor, on the basis of a discriminatory policy, allow individuals to inflict persecution with impunity. In countries in which the power of the State is weak, there may be intermediate cases in which groups of people have power in particular areas to persecute others on a discriminatory basis and the State, on account of lack of resources or political will and without its agents applying any discriminatory policy of their own, is unable or unwilling to protect them. I do not intend to lay down any rule for such cases. They have to be considered by adjudicators on a case by case basis as they arise. The distinguishing feature of the present case is the evidence of institutionalised discrimination against women by the police, the courts and the legal system, the central organs of the State.

Finally, I must say something about the general implications of this case. The Chairman of the Immigration Appeal Tribunal which heard Mrs Islam's case was dismissive about the evidence of discrimination against women in Pakistan. He said that it contained:

"overt and implicit criticisms of Pakistani society and the position of women in that and other Islamic states. We do not think that the purpose of the Convention is to award refugee status because of a disapproval of social mores or conventions in non-western societies."

There was in my view no suggestion that a woman was entitled to refugee status merely because she lived in a society which, for religious or any other reason, discriminated against women. Although such discrimination is contrary not merely to western notions but to the constitution of Pakistan and a number of international human rights instruments, including the Convention on the Elimination of All Forms of Discrimination Against Women, which Pakistan ratified in 1996, it does not in itself found a claim under the Convention. The Convention is about persecution, a well founded fear of serious harm, which is a very different matter. The discrimination against women in Pakistan found by the special adjudicator to exist there is relevant to show that the fear of persecution is on a Convention ground but is not in itself enough. Furthermore, the findings of fact as to discrimination have not been challenged. They cannot be ignored merely on the ground that this would imply criticism of the legal or social arrangements in another country. The whole purpose of the Convention is to give protection to certain classes of people who have fled from countries in which their human rights have not been respected. It does not by any means follow that there is similar persecution in other Islamic countries or even that it exists everywhere in Pakistan. Each case must depend upon the evidence.

I would therefore allow the appeals. In the case of Mrs Islam, I would make a declaration in accordance with section 8(2) of the Asylum and Immigration Appeals Act 1993 that it would be contrary to the United Kingdom's obligations under the Convention for her to be required to leave the United Kingdom. In the case of Mrs Shah, I would restore the order of Sedley J. remitting her case to the Immigration Appeal Tribunal.

LORD HOPE OF CRAIGHEAD

My Lords,

I have had the advantage of reading in draft the speeches which have been prepared by my noble and learned friends Lord Steyn and Lord Hoffmann. I agree with them that these appeals should be allowed, and I would make the same orders as they have proposed. I also agree with what they have said on the questions of causation and political opinion. I should like to make these observations on the question of "particular social group".

Article 1A(2) of the United Nations Convention Relating to the Status of Refugees, 1951 defines the term "refugee" for the purposes of the Convention. The relevant part of that definition, as amended by the 1966 Protocol, states that "refugee" means any person who:

". . . owing to well founded fear of being prosecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable or, owing to such fear, is unwilling to avail himself of the protection of that country."

The issue which is common to these appeals is whether the appellants, who are both married women of Pakistani nationality, are members of a "particular social group." The "particular social group" to which they claim to belong is said to comprise women in Pakistan accused of transgressing social mores who are unprotected by their husbands or other male relatives.

I would make three general points about the definition before I examine the phrase "particular social group". The first is that the characteristics, commonly referred to as the Convention reasons, which are set out in paragraph (2) of article 1A are designed to provide an inclusive list of the reasons for which the person who is seeking refugee status must claim that he has a well founded fear of being prosecuted. The list is not, as one finds in some human rights instruments, an illustrative one. This means that it is necessary for the person to be able to show that his fear is of persecution for a Convention reason, not just that he has a fear of being persecuted.

The second relates to a feature which is common to all five of the Convention reasons which are set out in the paragraph. The first preamble to the Convention explains that one of its purposes was to give effect to the principle that human beings shall enjoy fundamental rights and freedoms without discrimination. This principle was affirmed in the Charter of the United Nations and in the Universal Declaration of Human Rights approved by the General Assembly of the United Nations on 10 December 1948. If one is looking for a genus, in order to apply the eiusdem generis rule of construction to the phrase "particular social group," it is to be found in the fact that the other Convention reasons are all grounds on which a person may be discriminated against by society.

The third point is that, while the risk of discrimination by society is common to all five of the Convention reasons, the persecution which is feared cannot be used to define a particular social group. The rule is that the Convention reasons must exist independently of, and not be defined by, the persecution. To define the social group by reference to the fear of being persecuted would be to resort to circular reasoning: A. and Another v. Minister for Immigration and Ethnic Affairs and Another (1997) 142 A.L.R. 331, 358, per McHugh J. But persecution is not the same thing as discrimination. Discrimination involves the making of unfair or unjust distinctions to the disadvantage one group or class of people as compared with others. It may lead to persecution or it may not. And persons may be persecuted who have not been discriminated against. If so, they are simply persons who are being persecuted. So it would be wrong to extend the rule that the Convention reasons must exist independently of, and not be defined by, the persecution so as to exclude discrimination as a means of defining the social group where people with common characteristics are being discriminated against. That would conflict with the application of the eiusdem generis rule, and it would ignore the statement of principle which is set out in the first preamble to the Convention.

I turn now to the phrase "particular social group." As a general rule it is desirable that international treaties should be interpreted by the courts of all the states parties uniformly. So, if it could be said that a uniform interpretation of this phrase was to be found in the authorities, I would regard it as appropriate that we should follow it. But, as my noble and learned friend Lord Steyn has demonstrated in his review of the United States, Australian and Canadian case law, no uniform interpretation of it has emerged. The only clear rule which can be said to have been generally recognised is that the persecution must exist independently of, and not be used to define, the social group. I agree that the traveaux preparatiores of the Convention are uninformative. But it is more important to have regard to the evolutionary approach which must be taken to international agreements of this kind. This enables account to be taken of changes in society and of discriminatory circumstances which may not have been obvious to the delegates when the Convention was being framed.

In general terms a social group may be said to exist when a group of people with a particular characteristic is recognised as a distinct group by society. The concept of a group means that we dealing here with people who are grouped together because they share a characteristic not shared by others, not with individuals. The word "social" means that we are being asked to identify a group of people which is recognised as a particular group by society. As social customs and social attitudes differ from one country to another, the context for this inquiry is the country of the person's nationality. The phrase can thus accommodate particular social groups which may be recognisable as such in one country but not in others or which, in any given country, have not previously been recognised.

Mr. Pannick Q.C. said that a social group normally required cohesion between its members, and that if it lacked cohesion this was a very strong indication that it was not a group. But I think that this cannot be so in all cases. There are various ways in which a social group may be formed. It may be voluntary and self-generating. In that event it makes good sense to say, as Staughton L.J. said in the Court of Appeal at p. 93D, that it must have some degree of cohesiveness, co-operation or interdependence among its members. But, in the context of article 1A(2) of the Convention, I do not think that it needs to be self- generating. It may have been created, quite contrary to the wishes of the persons who are comprised in it, by society. Those persons may have been set apart by the norms or customs of that society, so that all people who have their particular characteristic are recognised as being different from all others in that society. This will almost certainly be because they are being discriminated against by the society in which they live as they have that characteristic. I do not think that the fact that it is discrimination which identifies the group to which these people belong as a "particular social group" within that society offends against the rule that the group must exist independently of, and not be defined by, the persecution. As I said earlier, people can be and often are discriminated against without being persecuted.

The rule that the group must exist independently of the persecution is useful, because persecution alone cannot be used to define the group. But it must not be applied outside its proper context. This point has been well made by Guy S. Goodwin-Gill, The Refugee in International Law, 2nd ed. (1996). At pp. 47-48 he observes that the importance, and therefore the identity, of a social group may well be in direct proportion to the notice taken of it by others. Thus the notion of social group is an open-ended one, which can be expanded in favour of a variety of different classes susceptible to persecution. In a footnote at p. 361, under reference to the analysis in Ward v. Attorney-General of Canada (1993) 103 D.L.R. (4th) 1, he notes that the 'grouping' will often be independent of will, so that the requirement of voluntary association relationship, if adopted in all cases, would introduce an unjustified, additional evidential burden on the person who seeks protection under the Convention. At p. 362, after further discussion, he concludes that to treat persecution as the sole factor which results in the identification of the particular social group is too simple. Persecution may be but one facet of broader policies and perspectives, all of which contribute to the group and add to its pre-existing characteristics.

The unchallenged evidence in this case shows that women are discriminated against in Pakistan. I think that the nature and scale of the discrimination is such that it can properly be said the women in Pakistan are discriminated against by the society in which they live. The reason why the appellants fear persecution is not just because they are women. It is because they are women in a society which discriminates against women. In the context of that society I would regard women as a particular social group within the meaning of article 1A(2) of the Convention.

In the decision of the U.S. Board of Immigration Appeals in In re Acosta (1985) 19 I. & N. 2H, it was recognised that, on the application of the eiusdem generis principle, the shared common, immutable characteristic which would qualify to form a particular social group could include the person's sex. La Forest J. in Attorney-General v. Ward (1993) 103 D.L.R. (4th) 1, 34 accepted that a particular social group could include persons who feared persecution because they were being discriminated against on the basis of gender. So to hold that the appellants were members of a particular social group in Pakistan because they are women and because women are discriminated against in that country would be consistent with previous authority. I do not think that it is necessary in this case to define the social group more narrowly. As the particular social group must be identified in each case in the light of the evidence, the fact that women in Pakistan belong to a particular social group because of way people of their gender are treated in their society does not mean that the same result will be reached in every other country where women are discriminated against. In other cases the evidence may show that the discrimination is based on some other characteristic as well as gender. If so, some other definition will be needed to identify the group. But that problem does not arise in this case.

LORD HUTTON

My Lords,

I have had the advantage of reading in draft the speech of my noble and learned friend Lord Steyn, and I agree that the appeals should be allowed on the ground described by him as "the narrower ground" which was relied on by the appellants. I prefer to express no view on the wider issue whether women in Pakistan constitute "a particular social group" within the meaning of article 1A(2) of the Convention.

LORD MILLETT

My Lords,

The question in these appeals is whether the appellants are refugees within the meaning of the Geneva Convention Relating to the Status of Refugees (1951) as amended by the 1967 Protocol. Article 1A of the Convention defines the term "refugee" to mean any person who:

"owing to well founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable or, owing to such fear, is unwilling to avail himself of the protection of that country"

The appellants are women from Pakistan who claim to have a well founded fear of persecution by reason of their membership of a particular social group. They identify the particular social group to which they belong as consisting of (i) women in Pakistan (ii) accused of transgressing social norms (in the present case by adultery and disobedience to their husbands) and (iii) who are in consequence unprotected by their husbands or other male relatives. Both appellants have satisfied the authorities that they have a fear of persecution and that this fear is well founded. The Court of Appeal rejected their claims to asylum on the ground that the class of persons they identified as liable to persecution was not a social group. The appeal thus calls for an examination of the concept "membership of a particular social group" in the context of the 1951 Convention.

Persecution may be indiscriminate. It may be for any reason or none. It is not, however, enough for an applicant for asylum to show that he or she has a well founded fear of persecution. The persecution must be discriminatory and for a Convention reason. By limiting the persecution in this way, the Convention contemplates that the possibility that there may be victims of persecution who do not qualify for refugee status. Furthermore, if the reason relied upon is membership of a particular social group, it is not enough that the applicant is a member of a particular social group and has a well founded fear of persecution. The applicant must be liable to persecution because he or she is a member of the social group in question.

In A. v. Minister for Immigration and Ethnic Affairs (1997) 142 A.L.R. 331, 355 McHugh J. said:

"Courts and jurists have taken widely differing views as to what constitutes 'membership of a particular social group' for the purposes of the Convention. This is not surprising. The phrase is indeterminate and lacks a detailed legislative history and debate."

The expression has, however, attracted a wealth of judicial analysis in the United States, Canada and Australia, and while considerable difficulties are likely to be encountered in applying the concept in practice, a broad consensus has emerged as to its nature. It is generally agreed that the group must constitute a cognisable group sharing common characteristics which set its members apart from society at large and for which they are jointly condemned by their persecutors. What constitutes a cognisable group is in my opinion a function of the particular society in which it exists. Westernised women may be cognisable as a distinct social group in an Islamic country in the Middle East but not in Israel; just as landowners were such a group in pre-revolutionary Russia but would not be in England today.

The qualifying bases of feared persecution were agreed at the Conference of Plenipotentiaries held at Geneva in July 1951. The inclusion of "membership of a particular social group" was made at the instance of the Swedish delegate because experience had shown that the other categories would be insufficient to cover all the situations intended. Although contemporary records of the discussions are not illuminating, it is not difficult in the light of history to discern what the delegates must have had in mind. Persecution of dissident minorities has often followed in the wake of social or cultural revolution. Class war has not been confined to our own continent and bloodstained century. Aristocrats during the French Terror, Kulaks in pre-war Soviet Russia, the intelligentsia and professional classes in Cambodia, have all been the victims of monstrous persecution not readily covered by the other Convention grounds. It does not, of course, follow that the expression should be confined to the social groups which the framers of the Convention are likely to have had in mind. But it should not be construed to exclude them.

In interpreting the expression "membership of a particular social group" I derive assistance from article 2 of the Universal Declaration of Human Rights. This was adopted by the General Assembly of the United Nations in December 1948, was still recent when the terms of the 1951 Convention were being settled, and is mentioned in the Preamble to the Convention. Article 2 prohibits the denial of the rights and freedoms set forth in the Declaration:

language

or social origin, property, birth or other status

The denial of human rights, however, is not the same as persecution, which involves the infliction of serious harm. The 1951 Convention was concerned to afford refuge to the victims of certain kinds of discriminatory persecution, but it was not directed to prohibit discrimination as such nor to grant refuge to the victims of discrimination. Moreover, while the delegates in Geneva were willing to extend refugee status to the victims of discriminatory persecution, they were unwilling to define the grounds of persecution which would qualify for refugee status as widely as the discriminatory denial of human rights condemned by the Universal Declaration. Discriminatory persecution "of any kind" would not suffice; the Convention grounds are defining, not merely illustrative as in the Universal Declaration. The inclusion of sex as a basis of discrimination in the Universal Declaration and the failure to include it as a ground of persecution in the 1951 Convention is noteworthy. It may be due to the fact that, while sexual discrimination was widely practised in 1951, and women are condemned to a subordinate and inferior status in many societies even today, it is difficult to imagine a society in which women are actually subjected to serious harm simply because they are women. But the words in article 2 which I have emphasised, "language . . . social origin, property, birth or other status", indicate to my mind the kind of characteristics which have commonly been shared by the victims of persecution and which the delegates must have had in mind when including the expression "membership of a particular social group". They are all matters of status rather than association; they have regard to the personal attributes of the victims rather than their behaviour.

It follows that I cannot accept the view of Staughton L.J. in the present case that the expression "particular social group" connotes a number of people joined together in a group with some degree of cohesiveness, co-operation or interdependence. It would exclude the victims of persecution on the ground of birth or social or economic class which was precisely the kind of persecution which the framers of the 1951 Convention are most likely to have had in contemplation. The requirement appears to have originated in the decision of the United States Court of Appeals (Ninth Circuit) in Sanchez-Trujillo v. Immigration and Naturalization Service (1986) 801 F.2d 1571 but the decision has not been followed in other circuits in the United States, and the requirement has been rejected in both Canada and Australia. In my opinion it should be rejected here also. The presence of such a factor may demonstrate that a distinct social group exists; its absence does not demonstrate the contrary.

In identifying the scope of the expression "a particular social group" the decision of the United States Board of Immigration Appeals in In re Acosta (1985) 19 I. & N. 211 has been very influential. This describes a social group as one consisting of persons who share an immutable characteristic, that is to say a characteristic that is beyond the power of the individual to change or is so fundamental to individual identity or conscience that it ought not to be required to be changed. A characteristic which is beyond the power of the individual to change may be innate or may consist of a shared past experience.

The suggestion that the defining characteristic need not be beyond the power of the individual to change has, however, been controversial. In Attorney-General of Canada v. Ward (1993) 103 D.L.R. (4th) 1, 33-34 La Forest J. offered the following as a useful working rule to identify the kinds of group contemplated by the Convention:

(i) groups defined by an innate or unchangeable characteristic;

(ii) groups whose members voluntarily associate for reasons so fundamental to their human dignity that they should not be forced to forsake the association; and

(iii) groups associated by a former voluntary status which is unalterable due to its historical permanence.

La Forest J. later modified this in Chan v. Minister of Employment and Immigration (1996) 128 D.L.R. 213, 248-249. He emphasised that he had offered only a working rule, not a definitive test, and stated unequivocally that there was no requirement that the group should be a voluntary association of like-minded persons. This is plainly right. But I respectfully agree with the analysis of Dawson J. in A. v. Minister for Immigration and Ethnic Affairs (1997) 142 A.L.R. 331, 344-345 that the only relevance of the characterisation of the common element as a fundamental human right is that persons associated for the purpose of asserting that right are more readily recognisable as a distinctive social group. To go further would transform the nature of the Convention by converting it from a measure affording protection to the victims of persecution into a measure aimed at counteracting discrimination.

It is also clear that the group must exist independently of the persecution. The reason for this is explained by Dawson J. in A. v. Minister for Immigration and Ethnic Affairs at p. 341:

Chan v. Canada

per

The group cannot, therefore, consist of the victims of persecution who share no other common characteristic. Nor in my opinion can it consist of victims of persecution who do share a common characteristic if that is not the reason for their persecution. Non-causative characteristics are irrelevant. Battered wives do not form a social group because, if the group is limited to battered wives, it is defined by the persecution, while if it is extended to include all married women, those who are battered are not persecuted because they are members of the group. The Canadian Court of Appeal decided the contrary in Canada (Minister of Employment and Immigration) v. Mayers (1992) 97 D.L.R. (4th) 729, but I agree with McHugh J. in A. v. Minister for Immigration and Ethnic Affairs (at p. 358 note 120) that the decision cannot be supported.

It is in my opinion essential to bear in mind at all times that it is not enough for the applicant for asylum to establish that he or she is a member of a particular social group and is liable to persecution. The applicant must also establish that he or she is liable to persecution because he or she is a member of the group. The applicant must be the subject of attack, not for himself or herself alone, but because he or she is one of those jointly condemned in the eyes of their persecutors for possession of the characteristic which is common to the group.

With these preliminary observations I turn to consider the social group to which the present appellants claim to belong and for membership of which they fear persecution. As formulated by the appellants it consists of women in Pakistan who have been or who are liable to be accused of adultery or other conduct transgressing social norms and who are unprotected by their husbands or other male relatives. It is a subset of the set "women" (and of the subset "married women"). The third qualifying condition can be disposed of at once. The fact that the appellants have no one to protect them helps to show that their fear of persecution is well founded. But it does not help to define the social group to which they belong. I am content to assume that the appellants would not be persecuted if they had someone to protect them from attack. But they are not persecuted because they have no one to protect them; that is not the ground of persecution. The "but for" test of causation, which is always necessary but rarely sufficient, is beguilingly misleading in this context.

This qualifying condition was no doubt included because of an erroneous belief that all the members of the group must be equally liable to persecution. That is not the case. It is no answer to a claim for asylum that some members of the group may be able to escape persecution, either because they have powerful protectors or for geographical or other reasons. Such factors do not narrow the membership of the group, but go to the question whether the applicant's fear of persecution is well founded. Thus I would accept that homosexuals form a distinct social group. In a society which subjected practising homosexuals but not non-practising homosexuals to persecution the relevant social group would still consist of homosexuals, not of the subset practising homosexuals. A non-practising homosexual would have no difficulty in establishing that he was a member of a persecuted group. His only difficulty would be in establishing that his fear of persecution was well founded, having regard to the fact that he was not a practising homosexual. This would be a matter of evidence, but given the hostility encountered by all homosexuals in such a society and the obvious problems the applicant would have in satisfying his tormentors of his own sexual abstinence, I doubt that the difficulty would be a real one.

Whether the social group is taken to be that contended for by the appellants, however, or the wider one of Pakistani women who are perceived to have transgressed social norms, the result is the same. No cognisable social group exists independently of the social conditions on which the persecution is founded. The social group which the appellants identify is defined by the persecution, or more accurately (but just as fatally) by the discrimination which founds the persecution. It is an artificial construct called into being to meet the exigencies of the case.

The appellants contended for the subset because they recognised that the head set of "women" or "married women" would not do. Officially, I understand, Sharia law regards women as "separate but equal," a description which, I observe, was also applied, with scant regard for the truth, to apartheid in South Africa. The evidence clearly establishes that women in Pakistan are treated as inferior to men and subordinate to their husbands and that, by international standards, they are subject to serious and quite unacceptable discrimination on account of their sex. But persecution is not merely an aggravated form of discrimination; and even if women (or married women) constitute a particular social group it is not accurate to say that those women in Pakistan who are persecuted are persecuted because they are members of it. They are persecuted because they are thought to have transgressed social norms, not because they are women. There is no evidence that men who transgress the different social norms which apply to them are treated more favourably.

In the course of argument an illuminating instance was put forward. Suppose, in the early years of the Third Reich, Jews in Nazi Germany were required to wear a yellow star on pain of being sent to a concentration camp and murdered if they did not. Would they have failed to qualify as refugees on the ground that they were not liable to persecution on racial grounds, but because they were defying the law? Of course we know now that they should not have failed to qualify, because the law was not merely discriminatory but a necessary part of the intended persecution. Jews were required to wear a distinguishing badge in order to mark them out for persecution. At the time this would have been a matter of evidence; but given the absence of any other rational explanation for the law, the virulence of the state-inspired racial propaganda which formed the background against which it was enacted, and the wholly disproportionate penalty for disobedience, there should have been no difficulty in satisfying the requirements of the Convention even in the absence of other evidence of persecution (of which there was an abundance). I find the example instructive precisely because of the differences between it and the present case rather than the similarities.

I am accordingly not willing to accept, as a general proposition, the submission that those who are persecuted because they refuse to conform to discriminatory laws to which, as members a particular social group, they are subject, thereby qualify for refugee status. Such persons are discriminated against because they are members of the social group in question; but they are persecuted because they refuse to conform, not because they are members of the social group. Nor am I able to accept the submission that somehow this ceases to be the case where the persecution is sanctioned or tolerated by the state. As I have said, this goes to a different question, whether the applicant's fear of persecution is well founded. It is still necessary to establish that the persecution, whether or not sanctioned by the state, is for a Convention reason.

Of course, the fact that the persecution is sanctioned by the state may make it easier to categorise it as persecution on political grounds. In extreme cases, where persecution is employed as an instrument of state policy and is actively encouraged by the authorities, as in Nazi Germany, the distinction between discrimination and persecution may be a distinction without a difference. Persecution takes many different forms. Where the authorities perceive a particular social group to be hostile, they may persecute its members by openly withdrawing their protection and leaving them to the mercy of criminal elements. The fact that those who take advantage of the situation to use violence against members of the group do so for their own private purposes does not matter; the members should be regarded as the victims of official persecution by the state. To qualify for refugee status, however, they must still prove that the state authorities have withdrawn their protection for a Convention reason.

Such questions will depend on the evidence. The evidence in the present case is that the widespread discrimination against women in Pakistan is based on religious law, and the persecution of those who refuse to conform to social and religious norms, while in no sense required by religious law, is sanctioned or at least tolerated by the authorities. But these norms are not a pretext for persecution nor have they been recently imposed. They are deeply embedded in the society in which the appellants have been brought up and in which they live. Women who are perceived to have transgressed them are treated badly, particularly by their husbands, and the authorities do little to protect them. But this is not because they are women. They are persecuted as individuals for what each of them has done or is thought to have done. They are not jointly condemned as females or persecuted for what they are. The appellants need to establish that the reason that they are left unprotected by the authorities and are liable to be persecuted by their husbands is that they are women. In my opinion they have not done so.

I would dismiss the appeals.