Cart Judicial Reviews through the Lens of the Upper Tribunal

Cart judicial reviews constitute the largest single group of claims for judicial review in the High Court. However, they are difficult to study because they rarely result in a High Court judgment and thus are reflected in the popular case law databases. I analysed the best public source of information on the fate of successful Cart judicial reviews: Upper Tribunal decisions following successful Cart claims from 2018 to 2020 selected programmatically from a dataset of over 42,000—mostly unreported—decisions. Thus, I can discuss the key aspects of successful Cart challenges, including the question what claims succeed in meeting the ‘second-tier appeals’ test for permission set in CPR 54.7A(7). This study complements previous quantitative work on rates of success in Cart and non Cart judicial reviews and contributes to the broader discussion on the appropriateness of retaining the Cart procedure in the context of the proposal to discontinue Cart claims made by the Independent Review of Administrative Law and adopted by the Government in the Judicial Review and Courts Bill.

Introduction 1. Cart judicial reviews ('Cart challenges' or 'Cart claims') constitute the largest single group of claims for judicial review in the High Court -20% of all judicial reviews and 24.8% of claims that reached the permission stage (since 2014). 1 Cart challenges are judicial reviews of Upper Tribunal (UT) decisions to refuse permission to appeal a decision of the First-tier Tribunal (FtT). 2 Despite being so numerous, Cart challenges are difficult to study because a very small percentage of them result in a judicial decision accessible from the public generalist case law databases. The Independent Review of Administrative Law (IRAL) fell into that trap by attempting to estimate the rate of success for Cart challenges based on judgments available on BAILII and Westlaw. 3 The problem with access to relevant data also limited previous academic studies. 4 2. I adopted a different method and I analysed the best public source of information on the fate of Cart challenges: UT decisions following successful Cart claims from 2018 to 2020 selected programmatically from a dataset of over 42,000mostly unreported -UT decisions. 5 This article uses the content analysis methodology. 6 This analysis allows me to discuss the key aspects of successful Cart challenges, including the question of what claims succeed in meeting the 'second-tier appeals' test for permission set in CPR r 54.7A(7). This study complements the previous quantitative study on rates of success in Cart and non-Cart judicial reviews 7 and contributes to the broader discussion on the appropriateness of retaining the Cart procedure in the context of the proposal to discontinue Cart claims made by IRAL and adopted by the government in the Judicial Review and Courts Bill. 8 3. The bulk of this article focuses on the problems with the Cart procedure, but this should not distract from the fact that many more Cart-following appeals cases succeed in the UT than fail even though failed cases are reported at much higher rates than successful ones (at least in the UT's Immigration and Asylum Chamber). My discussion of the problems, especially with the application of CPR r 54.7A (7), is meant to help identify ways in which the Cart procedure could be improved.
4. The article proceeds as follows. First, I give an overview of the studied sample of Cartfollowing decisions of the UT. I then turn to the main issue: the problems with the application of the test for permission in Cart claims beginning with examples of what factors led the courts to grant Cart permissions. Subsequently, I discuss the specific problems with the application of the 'second-tier appeals' limb of the test for permission. I also point out the more general problem with the lack of clarity in Cart permission decisions. I then consider two further issues: raising new grounds in Cart judicial reviews and the appropriateness of Cart judicial reviews following a partial ground of permission to appeal by the UT. I end with suggestions of amendments to CPR r 54.7A(7) that may be worth considering in the light of my study.
Overview of the studied Cart-following appeals in the Upper Tribunal 5. I identified 81 UT decisions, promulgated from 2018 to 2020, which followed Cart judicial reviews successful in quashing of an UT decision to refuse permission to appeal. 9 Of those cases 14 were reported (67 unreported explicitly critical of the Cart process were not reported, 11 but the imbalance does suggest that the problems with the Cart procedure are more salient to IAC judges than its successes, even though the clear majority -66% in the studied sampleof Cart-following appeals succeeded.
7. The 51 decisions where an FtT decision was set aside should not be treated as the total number of positive results in Cart cases in the studied years. Unreported decisions of other UT chambers than IAC, to which I did not have access, may contain additional positive results. Also, in 2017-2019, 2.3% of Cart challenges led to the UT setting aside an FtT decision; however, as many as 6.9% of all Cart challenges may have settled in favour of the claimant, which is clearly a positive result. 12 Moreover, it is arguable that even failed Cart-following appeals may be beneficial to the administration of justice, for example, by giving the UT an opportunity to clarify an important point of law. A good example of this is Chowdhury, where a Cart permission was granted because the High Court judge believed that an aspect of an immigration regulation required clear guidance from the UT, which the UT provided while dismissing the appeal. 13 8. I will now present some of the general characteristics of the studied appeals before turning to a more detailed discussion of Cart permissions.
9. Asylum and international protection appeals constituted the largest group among positive results (23 cases, ie 45% of positive results) followed by leave to remain (12 cases, 24%), and deportation and removal (nine, 18%). Among negative results, leave to remain cases dominated (15, 50% of negative results), followed by four asylum and protection cases (13%) and four deportation cases. Hence, in the sample, asylum and protection appeals were more likely to succeed than leave to remain appeals.
10. Table 1 illustrates the genders of appellants. 'Both' refers to decisions with several appellants, with at least one female and one male. In the sample, both lone female and lone male appellants were more likely to succeed than to fail. In decisions considering several appeals of appellants of mixed genders there were as many successes as failures.
11. Over two-thirds of appeals in the sample raised human rights grounds ( art 3 (seven cases alone). Moreover, six successful appeals relied on the Refugee Convention.
12. I also attempted to identify protected characteristicsunder s 4 of the Equality Act 2010raised as issues in the studied appeals. I counted 12 cases where disability was relevant, mostly in the form of mental health issues. Moreover, five cases concerned age (all dealing with children), three concerned sexual orientation and three alleged religious discrimination.
13. At least a strong suspicion of criminal behaviour of the appellant was noted as relevant in 11 cases with positive results (22%) and eight cases with negative results (27%). The difference of proportions is so small that no conclusions can be drawn from this observation.
The Cart test for permission (CPR r 54.7A(7)) in practice 14. Given that all the studied appeals followed grants of Cart permission to apply for judicial review -five in the Court of Appeal and 76 in the High Courtthe UT decisions can inform our understanding of the Cart permission process. Cart permission decisions are almost always determinative of a Cart claim: just 11 out of all 323 Cart claims granted permission reached the substantive hearing stage and only two hearings concluded with a dismissal of a claim. 14 In accordance with CPR r 54.7A(9), once permission is granted and unless a party requests a substantive hearingwhich happens very rarelya quashing order follows automatically.
15. CPR r 54.7A(7) sets the following strict ('second-tier appeals' 15 ) requirements for grant of permission in a Cart judicial review: (7) The court will give permission to proceed only if it considers -(a) that there is an arguable case, which has a reasonable prospect of success, that both the decision of the Upper Tribunal refusing permission to appeal and the decision of the First Tier Tribunal against which permission to appeal was sought are wrong in law; and (b) that either -(i) the claim raises an important point of principle or practice; or (ii) there is some other compelling reason to hear it.
16. In this section, I consider examples of factors that led the High Court or the Court of Appeal decide that the 'second-tier appeals' test (from subpara (b)) was satisfied. The following sections are devoted to the difficulties with Cart permissions that can be seen in UT decisions.
17. The first limb of the 'second-tier appeals' test is that 'the claim raises an important point of principle or practice'. One way in which this standard was met in the studied cases was due to decisions of higher courts interpreting the relevant law and post-dating the impugned FtT decision. Eight of the studied appeals succeeded on this ground. For example, in Khalid, the Court of Appeal granted permission because an FtT decision from September 2015 was arguably inconsistent with the approach taken by the Court of Appeal in a judgment from July 2016. 16 In NA, in a similar situation, the UT noted that the FtT judge made an error of law 'through no fault of her own' given that she 'cannot be faulted for applying the law as it was thought to be at the time of her decision'. 17 Those cases show that both Cart claims and Cart-following appeals sometimes succeed on new grounds, which could not have been raised before the FtT -I come back to this issue later in this article.
18. However, not every Cart-following appeal on this ground succeeds. In Shah, while dismissing the appeal the UT criticised the appellant's lawyers for misleading the High Court into thinking that the Court of Appeal granted permission on similar grounds to theirs in two other cases. 18  20. Turning to the 'compelling reason' limb of the 'second-tier appeals' test, perhaps the clearest example of a case that satisfies it is what Lord Dysonquoting Laws LJreferred to as 'a wholly exceptional collapse of fair procedure'. 21 However, either public authorities chiefly the Home Officesettle such cases out of court at a considerably higher rate than other Cart cases, or such situations are quite rare, because I identified only four instances in the studied sample. Those cases involved: clerical errors in the FtT, 22 a denial of adjournment of a hearing for medical reasons, 23 denial of permission to appeal despite evidence that the appellant failed to attend his hearing due to being hospitalised, 24 and a bank's fraud-prevention mechanism stopping the payment of an application fee. 25 21. The other type of a 'compelling reason' according to Lord Dyson is 'an error of law which has caused truly drastic consequences'. 26  Problems with the 'second-tier appeals' test 24. First, some Cart permission decisions expressly consider only the requirement of arguability from subpara (a) of CPR r 54.7A(7), with no express consideration of the 'second-tier appeals' test from subpara (b) or at least with no specific reasons given in respect to this test. This led UT judges to question whether the High Court properly applies the subpara (b) test. 25. For example, in Amin and Usman, the High Court did not expressly consider the requirements from subpara (b) at all in its permission decision. 28 Similar problems were apparent in Ejiogu, 29 S, 30 and P. 31 This is particularly concerning because, as Joanna Bell argued, 'the logic of Cart requires that questions of arguability are addressed after the applicant clears the hurdle laid down by the second-appeals criteria'. 32 One would expect the High Court to explain in a permission decision why the second-tier appeals criteria are met first and foremost, but I have seen little evidence of such explanations. It is, however, possible that more robustly reasoned permission decisions lead to out-of-court settlements and hence such cases tend not to reach the UT. 26. A specific problem with the High Court's approach to the 'second-tier appeals' criteria in Cart cases concerns the second limb of subpara (b) -'there is some other compelling reason to hear it'. In Thakrar, a case concerning potential removal of a Kenyan citizen from the UK, the UT reported that the High Court took the view that 'the consequences of removing the appellant would be "so momentous … that I can safely say there is a compelling reason for an appeal to be heard"'. 33 As the UT noted, the problem with that reasoning is that 'all appeals to the First-tier Tribunal against the respondent's refusal of a human rights or a protection claim involve the prospect of removal'. 34 Thakrar was not the only case in the sample where the question whether the severity of removal from the UK was a 'compelling reason' in the sense of subpara (b) arose. 35 27. It is possible that some High Court judges see all arguable errors of law in removal cases as situations that 'cry out for consideration' or instances of potentially 'truly drastic consequences'. 36 However, the vast majority of FtT appeals 'involve the prospect of removal'even if indirectly. The sum of asylum, protection, human rights and deportation appeals constituted 80% of the appeals received in the FtT in 2019-2020. 37 Moreover, 29% of UT IAC decisions in 2018-2020 expressly discussed 'removal'. 38 This raises the question whether the discussed approach of some High Court judges is consistent with the Supreme Court's decision in Cart v Upper Tribunal, which stressed the exceptional nature of what we now know as Cart challenges. 39 The readings of Cart adopted by Bell in her Public Law article 40 and by Carnwath LJ in PR (Sri Lanka) 41 suggest that this approach is inconsistent with Cart.
General lack of clarity of reasons why permission granted 28. Some permission decisions do no more than merely repeat the wording of CPR r 54.7A(7), without giving any specific reasons related to the claim in questionalso regarding the first limb of the test (arguability of an error of law), from subpara (a). Given that permission decisions tend to be dispositive of Cart claims, arguably they should be better reasoned than ordinary permission decisions in judicial review. In MA, the UT complained that it was unable to ascertain what arguable error of law the FtT made according to the High Court because the grant of permission contained only the following sentences: Permission is hereby granted.
Observations: The Applicant has demonstrated a reasonable prospect of success in establishing that both the FTT and the UT made serious legal errors and the claim crossed the threshold on the basis of compelling reason. 42 29. This is not an isolated case. For example, in Das the UT noted regarding the Cart permission decision: '[t]he only reason given for that decision was that "The grounds reach the threshold for obtaining permission"'. 43 The permission decision reported in Singh was slightly longer, but similarly unhelpful in identifying the reasons why permission was granted. 44 30. There is a large discrepancy between some Cart permission decisions that are robustly reasoned and those which contain little or no indication of reasons. Given that I did not study a statistically representative sample of the High Court's permission decisions in Cart cases, I cannot say how common the latter problem is. having access to documents from proceedings in which the UT is always the defendant. It is one thing that the UT takes a neutral position in Cart proceedings, but it is unclear why the UT could not receive and appropriately file the documents served on it as a party to those proceedings. This looks like a problem with internal clerical procedures in the UT, not something to be addressed by imposing additional duties on the parties.

New grounds in Cart claims
33. It is well-settled 'that the arguments which can be raised on appeal are limited by the grounds of appeal for which permission has been granted'. 48 Relatively often claimants succeed in Cart judicial reviews on different grounds than those used previously before the FtT and the UT. However, in such cases the UT requires appellants to apply for permission to amend groundsa Cart permission and a successful quashing of a UT refusal of permission to appeal do not constitute a permission to appeal, much less permission to appeal on new grounds. It seems that applications to amend grounds are normally granted and refused only if the UT is convinced at the outset that the grounds are hopeless.
34. In Shah, the UT stated that a Cart challenge is 'emphatically not an opportunity for a party to raise new grounds of appeal against the decision of the First-tier Tribunal'. 49 In that case, the claimant succeeded in a Cart judicial review on new grounds, but the follow-up appeal failed in the UT. This approach was confirmed in Das and in MA, both reported. 50 In Ahammad, the UT cited Shah and noted that the appellant attempted to advance new grounds, but the tribunal seems to have allowed it, which did not affect the result as the appeal failed. 51 In Aslam, the UT also allowed the appellant's counsel to proceed on new grounds even though no specific application to amend the grounds was madehere too the appeal failed. 52 In both Ahammad and Aslam the UT noted that the respondent did not object to proceeding on the basis of new grounds. Other examples of cases where the UT also allowed amendments of grounds of appeal are N and P. 53 35. On the other hand, Ejiogu provides an example of a refusal of application to amend grounds of appeal. 54 This application was opposed by the respondent, but the key problems were that the UT did not see the merit in the amended grounds and that the counsel did not satisfy the UT in his explanation 'why the grounds upon which he now sought to rely could not have been advanced within the time limited for the application to the Upper Tribunal'. 55 36. In S the UT raised a possibility of an approach that is difficult to square with Shah, Das and MA and with the requirement of permission to amend the grounds of appeal: Given that the High Court found that ground arguable, and in the result persuasive, leading to the quashing of the decision, I have no doubt that the appellants are entitled to rely on that ground before the UT once permission has been granted subsequently. It may be, and I express no concluded view on this as it was not touched on in the submissions, that he [sic] High Court's decision acts as an implied amendment or variation (by addition) of the original grounds to the UT. 56 37. Cojan highlights a difficulty with the seemingly permissive approach to using new grounds accepted by the High Court in a grant of a Cart permission. 57 The UT said: Since that basis had not featured in the grounds of permission to appeal to the Upper Tribunal, it is difficult to see how the decision of the Upper Tribunal, in refusing permission, could have been 'wrong in law', as required by CPR 54.7A(7)(a). 38. The UT dealt with this problem also at least two times in 2021 (my main sample only covered cases until the end of 2020, but I was later informed about the more recent cases). 58 In Osefiso, the appellants applied to amend the grounds of appeal before the UT after the UT had already decided to refuse permission to appeal. 59 The appellant's application to amend grounds was, according to the UT, substantially identical to their subsequent successful Cart application, and relied on entirely new grounds. 60 This irregular application to amend was likely an implicit recognition of the problem noted in Cojan. That is, that the High Court cannot grant a Cart permission on grounds which the UT did not need to considerbecause they were not pleaded and not 'Robinson obvious' 61because in such a situation it can hardly be said that the UT was 'wrong in law' regarding those grounds. In Osefiso, the UT denied that it has jurisdiction to hear this kind of a late application to amend in the absence of a procedural irregularity (r 43 of the Tribunal Procedure (Upper Tribunal) Rules 2008). 62 The UT expressed the hope that 'no one should repeat those steps in future'. 63 However, despite the irregularity, the UT accepted that there was a Robinson obvious ground in this case, gave permission to amend grounds, and allowed the appeal. 64 39. In another recent case the UT criticised the appellant's counsel in very strong terms ('may have been professionally reprehensible') for what looked like misleading the High Court that a ground had been put to the FtT, when it was not. 65 The High Court granted a Cart permission specifically due to that ground, stating that it arguably should have been explored 'whether the FTT judge erred by failing to take into account' that ground. 66 The appeal failed. 67 40. As I noted earlier, the UT allowed appeals on grounds that were not raised before the FtT or the UT, even in situations where the FtT and the UT decisions only came to be seen as clearly erroneous in the light of a later authoritative interpretation of the relevant law by a higher court. The UT also allowed appeals on grounds of new evidence. There may be some tension between that practice and the understanding of the role of an appellate tribunal implicit in Cojanie that it is meant to decide whether the original decision was erroneous in law at the time it was made.
Are Cart judicial reviews appropriate when the UT partially grants permission to appeal? 41. Another point that seems to have led to some confusion is the question of what path is available to an applicant who is successful in obtaining a permission to appeal from the UT, but only on some grounds? There is some evidence of contradictory approaches taken by different judges.

In
Mossa decision from the UT's Administrative Appeals Chamberthe appellant was successful in his Cart judicial review after being refused permission to appeal by the UT on some grounds onlythe Cart order quashed the UT decision in part in which it denied permission on other grounds. 68 The UT AAC judge subsequently granted permission to appeal on the remaining grounds, without any notes about this process being improper. 69 The UT decision did not register whether the High Court judge made any comments about this either.
43. This approach can be contrasted with HOK (Iraq), where a judge of the UT's Immigration and Asylum Chamber said the following about CPR r 54.7A: There is no reference in the rule to the possibility of an application being made where permission has been granted on limited grounds only. This is plainly an intentional omission, since the CPR makes express provision for such applications in other contexts. Where, for example, permission to proceed with an application for judicial review has been granted 'on certain grounds only', CPR 54.12 makes provision for the partly successful applicant to request that decision to be reconsidered at a hearing. CPR changes worth considering 45. Permission decisions are final decisions in nearly all Cart judicial reviews. Therefore, it is important for High Court judges to provide more robustly reasoned permission decisions that explicitly address how the claim meets both the first and second limb of CPR r 54.7A(7).
46. As an alternative to the proposed sweeping reforms in cl 2 of the Judicial Review and Courts Bill, it may be worth considering whether the Civil Procedure Rules should be amended to include an explicit direction to judges deciding Cart permissions to provide reasons that will allow the UT to identify specifically on what grounds, with reference to the relevant law and facts, the High Court believed that the requirements of both subparas of CPR r 54.7A(7) are satisfied.
47. Given the problems with priority between the two limbs of the test and following Bell's argument, it may also be advisable to reframe CPR r 54.7A(7) to make it clear that the 'second-tier appeals' test must be satisfied before considering the issue of arguability.
48. Finally, the question whether CPR r 54.7A(7) applies in situations where the UT only partially refuses may be calling for clarification.

Conclusion
49. This article focused on problems with Cart claims that are reflected in follow-up appellate decisions of the UT. However, I want to stress that those problems are not evidence that the procedure does not work in general. To the contrary, a clear majority of appellants succeed in their Cart-following UT appeals. 71 The problems I noticed do make it more difficult for UT judges to decide such appeals, but the issues can be addressed with relatively small procedural or organisational changes.