M v London Borough of Croydon

This appeal raises the issue as to the proper approach to awarding costs in judicial review proceedings, where the defendant public authority effectively concedes some or all of the relief which the claimant seeks. As with any question relating to costs, the issue is both highly fact-sensitive and very much a matter for the discretion of the first instance tribunal. However, a degree of consistency of approach is self-evidently desirable, and the issue gives rise to some points of principle and policy.

The appellant, known in these proceedings as M, arrived unaccompanied in the UK from Afghanistan on 19 March 2008. He immediately applied for asylum, informing the Secretary of State for the Home Department that he was aged 12, so his date of birth was recorded as 1 January 1996. He was then referred to the respondents, the London Borough of Croydon, one of the so-called gateway local authorities for young asylum seekers. The respondents accepted responsibility for the appellant pursuant to section 20 of the Children Act 1989 (‘section 20’), and placed him with foster parents and at a school (‘the school’). (The appellant has now been granted leave to remain in the UK as a refugee).

Under section 20, the nature and extent of the duty of a local authority depends on the age of an individual, and in the past few years, there have been a significant number of cases brought in the Administrative Court by claimants challenging a decision of a local authority as to their age.

In this case, the school questioned the appellant’s age in early June 2008, suggesting that he was 14. That same month, Bennett J gave judgment in R (M) v Lambeth LBC; R (A) v Croydon LBC [2008] EWHC 1364 (Admin), in which he concluded that the question whether an individual was a child for the purposes of section 20 was a matter for the relevant local authority rather than the court, although the court could review the local authority’s decision on normal judicial review principles.

On 20 August 2008, two experienced social workers employed by the respondents assessed the appellant’s age as over 14. Soon afterwards the Refugee Council referred the appellant to his present solicitors. In mid-September 2008, those solicitors wrote the first of a series of letters challenging the respondents’ assessment. The respondents replied, disclosing their reasoning, whereupon the appellant’s solicitors requested an assessment from someone they believed was an expert in the field, Dr Birch.

On 7 October 2008, Dr Birch produced a report assessing the appellant’s age to be between 12 and 14, from which she derived a mean age of 12 years and 11 months. On 30 October, the appellant's solicitors sent a very detailed pre-action protocol letter to the respondents, together with Dr Birch’s report and two witness statements from staff at the Refugee Council. They asked the respondents to set aside the age assessment of 20 August 2008, and to accept that the appellant was aged 12 or to carry out a further age assessment.

The respondents asked for further time to consider the issue. On 17 November 2008, a member of the respondents’ Unaccompanied Minors Team carried out a review and affirmed that the appellant was 14. The respondents then wrote to the appellant’s solicitors, rejecting Dr Birch’s methodology, stating that they stood by their original age assessment, and disclosing their file and a care plan.

On 28 November 2008, in R (A by Valbona Mejzninin) v Croydon LBC [2008] EWHC 2921 (Admin), Mr Stephen Morris QC, sitting as a deputy High Court Judge, quashed the respondents’ decision as to another child’s age, because they had not had adequate regard to a report from Dr Birch. He held that the respondents should not have rejected her expert medical opinion without ‘sound reason’. Further correspondence ensued with no change of position on either side.

On 18 December 2008, this court handed down judgment in R (A) v Croydon LBC and R(M) v Lambeth LBC [2008] EWCA Civ 1445, upholding the decision of Bennett J. Thus, it was apparent that, if the issue of the appellant’s age was to be litigated, it would be on the basis of a judicial review. So, in order to succeed, the appellant would have to establish that the respondents’ decision that he was born in 1994 rather than 1996 was ‘Wednesbury unreasonable’ i.e. that the decision was irrational, in the sense of being one which no reasonable local authority could have reached.

In January 2009, in R (A) v Croydon and R (WK) v Kent County Council, Holman J directed a stay of all age assessment cases, pending a trial which would involve consideration of the reliability and value of expert evidence as to the age of a child.

By the end of that month, the appellant’s solicitors had resolved that they wished formally to challenge the respondents’ decision of 17 November 2008, so they had to issue any claim by 17 February 2009. Accordingly they sent another pre-action protocol letter on 30 January, contending that the respondents’ failure to accept the appellant was born in 1996 was irrational, in the light of the report of Dr Birch and statements from the workers at the Refugee Council. By the letter, the appellant’s solicitors required the respondents to accept that the appellant was born in 1996, or to carry out a further assessment, and requesting a response within two weeks.

On 3 February, the appellant’s solicitors served the respondents with a draft claim form and a chronology. On 12 February, the respondents conducted a case review, and, four days later, they wrote saying that, given the volume of documentation they had received, they were unable to respond by 13 February. The appellant’s solicitors then issued and served a claim form, seeking permission to apply for an order quashing the respondents’ decision of 17 November 2008 to refuse to revisit their assessment that the appellant was born in 1994. The relief sought was a (i) declaration that “any rational authority would have withdrawn their decision of 20 August and reassessed the claimant” in the light of Dr Birch’s report and “the witness statement of the Refugee Council”, and (ii) a mandatory order directing a reassessment within 14 days. The respondents’ acknowledgement of service should have been filed by 9 March 2009, but it was not. Nevertheless the parties continued to correspond about the appellant's education and future, while the appellant’s claim, like all other age assessment cases remained stayed in accordance with the direction of Holman J.

In March 2009, Lisa Thomas, Head of Year at the appellant’s school again confirmed that in her opinion the appellant was born in 1994. Thus, to date the respondents had received opinions from the appellant’s teachers and from experienced social workers that he was born in 1994, and opinions from Dr Birch and two workers at the Refugee Council that he was born in 1996.

Meanwhile, on 8 May 2009, Collins J gave judgment in R (A) v Croydon and R (WK) v Kent County Council [2009] EWHC 939 (Admin). He was very critical of Dr Birch’s methodology, and consequently he disagreed with the approach adopted by Mr Morris QC (who had not had the benefit of detailed evidence and argument as Collins J had done). Collins J directed that the outstanding age assessment cases be listed as soon as possible.

Because the appellant’s solicitors had relied on the views of Dr Birch, the respondents, on 27 May 2009, invited them to withdraw the appellant’s claim. However, on 15 June 2009, the appellant's solicitors invited the respondents to accept that the appellant was born in 1996 or to re-assess his age having regard to an addendum statement from Dr Birch and statements made by workers at the Refugee Centre.

Correspondence ensued during which the respondents indicated they were in the process of reviewing the outstanding cases where there were medical reports. A further review was conducted on 11 August 2008, when a practice manager, Eileen Lawrence, and Ms Henry confirmed the respondents’ original assessment. At the end of August 2009 the appellant’s solicitors wrote expressing why, in their view, the assessment of 11 August was flawed and asked for yet another review claiming the previous reviews had not been properly conducted. They alleged the previous reviews had simply provided ex post facto reasons for the previous decisions. The respondents were asked to disclose particulars of their contact with the appellant's school and disclosure of all relevant records. They were disclosed.

On 26 November 2009, the decision of the Court of Appeal, upholding Bennett J, was reversed by the Supreme Court in R (A) v Croydon LBC and R (M) v Lambeth LBC [2009] UKSC 8 [2009] 1 WLR 2557 (‘R(A) v Croydon’). The Supreme Court concluded that the determination of age is a fact precedent to the exercise of a local authority’s powers under section 20, so that, in the case of any dispute, the responsibility for assessing an individual’s age under that section was for the court. As a result, what had been the appellant's traditional judicial review application was effectively converted into an age assessment claim under the enhanced jurisdiction of the Administrative Court.

The beleaguered staff at the respondents were working their way through literally hundreds of claims involving disputes as to age assessment for the purpose of section 20. They chose to do so on the basis of first issued first dealt with. That is presumably why there appears to have been little signs of activity in this case for more than eight months after November 2009.

Meanwhile, on 26 July 2010, His Honour Judge McMullen QC, sitting as a High Court Judge, granted the appellant’s application to issue proceedings against the respondents. In granting permission, he noted that, in the absence of an acknowledgment of service, there was an arguable case, and directed, inter alia, that, doubtless based on the decision of the Supreme Court, the case should proceed under the enhanced fact finding jurisdiction of the Administrative Court.

On 6 August 2010, the respondents notified the appellant’s solicitors that an age assessment interview had been arranged for 10 August but the parties disagreed on the terms of the interview and it never happened. The respondents’ acknowledgement of service and grounds of defence were served on 13 August 2010. The grounds of defence explained why the respondents asserted the appellant was two years older than he claimed, and included reference to the fact that Dr Birch herself conceded there was a wide margin of error in these cases. Accordingly, it was said, there was a 50% chance the respondents were correct, even on the appellant’s evidence.

The respondents had meanwhile instructed a Dr Stern. On 21 August 2010, he produced a report, in which he (i) criticised Dr Birch’s report and methodology, (ii) considered witness statements of Refugee Council workers, and (iii) concluded that ‘it is likely … that [the appellant] was younger than 14 years when he was first assessed and may have been 12 years old at that time’.

Within a few days of receiving it, the respondents served Dr Stern’s report, and the witness statements upon which he relied, on the appellant’s solicitors. As a result, the appellant’s solicitors invited them to concede the claim. On 15 October 2010, the respondents conceded that the appellant was born on 1 January 1996. (It may be of some interest to any reader who has had the stamina to follow this saga so far that, in the subsequent case of R(R) v Croydon LBC [2011] EWHC 1473 (Admin), Kenneth Parker J was highly critical of the evidence given by Dr Birch in purporting to assess the age of an individual, and he much preferred the evidence of Dr Stern.)

Although the respondents were prepared to concede on the substantive issue between the parties, namely the appellant’s age, they were not prepared to agree to pay his costs of the proceedings, to which he said he was entitled as a result of his claim having been conceded by the respondents. Accordingly, on 11 February 2011, the court sealed a consent order, which recorded that the respondents agreed that the appellant was born on 1 January 1996, and by which the appellant withdrew his claim, leaving the question of costs to be determined on paper by the court.

On 18 May 2011, having received written submissions on behalf of both parties, Lindblom J made no order for costs, for the following reasons:

The appellant applied for permission to appeal on two grounds, namely:

The appeal is of no significance to the appellant himself. It does, however, have significance for his lawyers. If the decision that there should be no order for costs stands, they will recover substantially less than if a costs order was made in the appellant's favour.

The appellant’s case is that the Judge ought to have awarded him the costs of the proceedings, because, from the initiation of these proceedings in February 2009 until the respondents conceded, the only real issue between the parties was whether the appellant was born in 1994, as the respondents contended, or in 1996, as he contended, and he had been wholly successful on that issue. The respondents support the order made by Lindblom J, and, in that connection, they rely on (i) the well-established and salutary rule that an appellate tribunal does not normally interfere with a judge’s decision on costs, (ii) the principle that, where a case settles, it is normally wrong for the court to award costs unless it is tolerably clear that one party would have won and it is not clear in this case, and (iii) both the facts and the law relating to the issue between the parties were unclear and changing during the period of the dispute, so that it would be unfair to penalise the respondents in costs.

In order to consider the issue properly, it is necessary first to refer to the relevant provisions of the CPR, and the relevant cases in, or on appeal from, the Administrative Court. After then considering the normal approach to costs in civil cases which go to trial, and in those that settle, I will turn to discuss the right approach to be adopted by the Administrative Court when asked to deal with costs in cases in which the parties settle all the other issues, and finally turn to the disposition of this appeal.

The ‘general rule’ as to costs is to be found in CPR 44.3 (2), which provides that (a) ‘the unsuccessful party will be ordered to pay the costs of the successful party’ but (b) ‘the court may make a different order’ if the circumstances demand. CPR 44.3(4) provides that such circumstances include (a) ‘the conduct of all the parties’, (b) the extent to which the successful party has succeeded, and (c) any admissible offer to settle, including any offer under CPR 36.

CPR 44.3(5) explains that the reference to ‘conduct’ includes conduct before, as well as during, the proceedings, and includes (a) ‘the extent to which the parties followed the Practice Direction (Pre-Action Conduct), or any relevant pre-action protocol’, (b) the reasonableness of raising or defending a particular issue, (c) the manner in which the litigation was conducted, (d) whether a successful claim had been exaggerated,

The application of this rule to compromised judicial review claims was considered by Scott Baker J in R (Boxall) v Waltham Forest LBC (2001) 4 CCLR 258 (‘Boxall’). Having analysed a number of earlier decisions, he set out at para 22, certain principles, or guidelines, which he derived from those cases and which, Ms Rowlands for the respondents maintains, her clients assumed would apply when they compromised the instant proceedings.

The guidelines set out by Scott Baker J were as follows:

These guidelines pre-dated the Pre-Action Protocol for Judicial Review introduced in December 2001 (‘the Protocol’), which came into effect one year after the decision in Boxall. So far as relevant, it provides.

The Boxall guidelines were considered in R(Scott) v Hackney LBC [2009] EWCA Civ 217 (‘Scott’), where the local authority had conceded a substantial part of the claimant’s case. The Administrative Court Judge, had decided that there should be no order for costs on the ground that the local authority had ‘reasonable points of defence’, he could not ‘confidently conclude’ who would have won if the matter had not settled, he should not ‘discourage reasonable settlements’, and ‘certain aspects of the original claim … were not pursued’.

In the Court of Appeal, it was common ground that the Boxall guidelines were valid and should be applied (see para 38). The points raised by the claimant appear to have been (i) that the Protocol should be taken into account, (ii)(a) that the judge had set too high a test when he said that he could not ‘confidently conclude’ who would have won, and (b) that it was clear that the claimant would have won (see paras 41, 42 and 45). In her judgment (with which Sir Andrew Morritt C and Richards LJ agreed) at para 41, Hallett LJ said that adherence to the Protocol was plainly a relevant factor to be taken into account when considering an application for costs, but she was unimpressed with the other two points made by the claimant.

At para 49, she expressed the view that ‘there was here no real winner and no real loser’, and that ‘[n]o order for costs was the only way to do overall justice in this case’. Two paragraphs later, she said that:

The next significant development was the review of costs conducted by Sir Rupert Jackson, who published his Final Report in December 2010. At paras 4.12-13, Sir Rupert said this:

Thereafter, in Bahta, this court heard five appeals, in all of which the claimants had brought proceedings challenging the refusal of the United Kingdom Borders Agency (‘UKBA’) to accord them permission to work in this country. In due course, UKBA conceded each claim, but refused to pay the claimant’s costs. In each case, an Administrative Court Judge had made no order for costs, applying the Boxall guidelines, holding that, although the claimant had substantially obtained the relief he or she sought, the outcome was not plain, or, if it was plain, that was only because the law had been changed or clarified between issue and settlement – see paras 8-12 (and three first instance cases going the other way were cited in the next three paragraphs).

In his judgment in Bahta (with which Sullivan LJ and Hedley J agreed) Pill LJ said that the decision as to costs must be made by reference to the circumstances as at the date of assessment, and must take account of ‘the whole sequence of events and the conduct of the parties throughout’ (see para 58).

At para 60, he said that ‘[n]otwithstanding the heavy workload of UKBA, and the constraints upon its resources, there can be no special rule for government departments in this respect.’ As he explained, ‘[o]rders for costs, legitimately made, will of course add to the financial burden on the Agency. That cannot be a reason for depriving other parties, including publicly funded parties, of costs to which they are entitled.’ In the following two paragraphs, Pill LJ made it clear that, consistent with other judicial dicta, the fact that a party was publicly funded was no reason to depart from normal principles.

At para 63, Pill LJ expressed ‘serious misgivings about UKBA's claim to avoid costs when a claim is settled for “purely pragmatic reasons”’, explaining that ‘[t]he expression “purely pragmatic” covers a multitude of possibilities’. He then said that ‘[a] clear explanation is required, and can expect to be analysed, so that the expression is not used as a device for avoiding an order for costs that ought to be made’. At para 64, he said that ‘what needs to be underlined is the starting point in the CPR that a successful claimant is entitled to his costs and the now recognised importance of complying with Pre-Action Protocols’. Accordingly, he explained in the following paragraph, ‘[w]hen relief is granted, the defendant bears the burden of justifying a departure from the general rule that the unsuccessful party will be ordered to pay the costs of the successful party and that the burden is likely to be a heavy one if the claimant has, and the defendant has not, complied with the … Protocol’.

Pill LJ said at para 66 that he would not ‘tack on words to the Boxall guidelines’. Such a course could, he suggested, lead to ‘a formula [which] would carry the danger of being used mechanistically when what is required is an analysis of the circumstances of the particular case, applying the principles now stated.’ He also repeated ‘the warning in Scott [2009] EWCA Civ 217, that a judge should not be tempted too readily to adopt a fall back position.’

The outcome of the five appeals was stated atpara 70 to be that all the claimants were entitled to their costs, on the ground that they ‘were entitled to work and there was no justification for withholding that right from them. Proceedings were instituted and pursued because the right was withheld’, and there was ‘no justification for resorting to the fall back position stated in Boxall’.

There are three relevant general principles which appear to me to apply to awards of costs after a trial in ordinary civil litigation. The first is that any decision relating to costs is primarily a matter for the discretion of the trial judge, which means that an appellate court should normally be very slow indeed to interfere with any decision on costs. However, while wide, the discretion must be exercised rationally and in accordance with certain generally accepted principles. To a large extent, those principles are set out in CPR 44.3, and in particular, paras (2), (4), (5), and (6). If the trial judge departs from rationality or the correct principles then it is legitimate for an appellate court to interfere with his conclusion.

The second principle is that, as has long been the case in English civil litigation, and is expressly stated in CPR 44.3.2(a), the general rule in all civil litigation is that a successful party can look to the unsuccessful party for his costs. Of course, as CPR 44.3(2)(b), (4), (5) and (6) demonstrate, there may be all sorts of reasons for departing from this principle, but it represents the prima facie position. For instance, the fact that the successful party lost on, or abandoned, an issue, will often involve his being deprived of some, or even all, of his costs (and, in an extreme case, he may even have to pay some of the unsuccessful party’s costs) – CPR 44.3(4)(b). Further, the parties’ conduct is a relevant matter, as CPR 44.3(4)(a) provides, so that failure to adhere to the provisions of any relevant protocol may well affect any decision the court makes on costs.

The third principle is that the basis upon which the successful party’s lawyers are funded, whether privately in the traditional way, under a ‘no win no fee’ basis, by the Community Legal Service, by a Law Centre, or on a pro bono arrangement, will rarely, if ever, make any difference to that party’s right to recover costs. That point appears to me to be plainly right as a matter of principle, and it is supported by the second of the Boxall guidelines, by what was said in this court by Hallett LJ in Scott para 56, and by Pill LJ in Bahta paras 61-2, and by what Lord Hope said in the Supreme Court in Re appeals by Governing Body of JFS [2009] UKSC 1 [2009] 1 WLR 2353, paras 24-45.

It is open to parties in almost any civil proceedings to compromise all their differences save costs, and to invite the court to determine how the costs should be dealt with. The court has jurisdiction in such a case to determine who is to pay costs, but it is not obliged to resolve such a free-standing dispute about costs. Accordingly, by settling all issues save costs, the parties take the risk that the court will not be prepared to make any determination other than that there be no order for costs not only because that is the right result after analysing all the arguments, but also on the ground that such an exercise would be disproportionate.

In BCT Software Solutions Ltd v C Brewer & Sons Ltd [2003] EWCA Civ 939, [2004] FSR 9 (‘BCT’), Chadwick LJ said this at para 24 (which was approved in Venture Finance plc v Mead [2005] EWCA Civ 325):

However, Chadwick LJ immediately went on to say in the next paragraph:

The outcome will normally be different in cases where the consent order does not involve the claimant getting all, or substantively all, the relief which he has claimed. In such cases, the court will often decide to make no order for costs, unless it can, without much effort decide that one of the parties has clearly won, or has won to a sufficient extent to justify some order for costs in its favour. Thus, the fact that the claimant has succeeded in obtaining part of the relief he sought may justify his recovering some of his costs, for instance where the issue on which the claimant succeeded was clearly the most important and/or expensive issue. But in many such cases, the court may consider that it cannot fairly award the claimant any costs because, for instance, it is not easy to assess whether the defendants should have their costs of the issue on which the claimant did not succeed, and whether that would wipe out the costs which the claimant might recover in relation to the issue on which he won.

In many cases which are settled on terms which do not accord with the relief which the claimant has sought, the court will normally be unable decide who has won, and therefore will not make any order for costs. However, in some cases, the court may be able to form a tolerably clear view without much effort. In a number of such cases, the court may well be assisted by considering whether it is reasonably clear from the available material whether one party would have won if the case had proceeded to trial. If, for instance, it is clear that the claimant would have won, that would lend considerable support to his argument that the terms of settlement represent success such that he should be awarded his costs. An example of such a case is Brawley v Marczinski, [2002] EWCA Civ 756, [2003] 1 WLR 813 where the court could determine, without too much effort, who would have won, and then took that into account when awarding costs.

The question which then arises is whether the principles discussed in the preceding section of this judgment should apply in the Administrative Court, just as much as to other parts of the civil justice system: in particular, where the defendants accept that the claimant is entitled to all, or substantially all, the relief which he claims, should the defendants pay his costs, unless they can show good reason to the contrary? At least on the face of it, the fact that a claim is a public law claim should make no difference. Such claims are subject to the CPR, and a successful claimant who has brought such a claim is just as much entitled to his costs as he would be if it had been a private law claims. The court’s duty to protect individuals from being wronged by the state, whether national or local government, is every bit as vital as its duty to enable them to vindicate their private law rights. And the fact that the defendants are public bodies should make no difference, as Pill LJ explained in Bahta atpara 60. However, a number of points could be raised as to why defendants who concede claims in the Administrative Court should be less at risk on costs than those who concede in ordinary civil actions.

First, it may be said that government and public bodies should be encouraged to settle, and should not therefore be penalised in costs if they do so after proceedings have been issued. There are four answers to that. First, if it is a good point, it should apply to any litigation, whether in private law or public law, and in very few, if any, private law cases would such an argument carry any weight. The implication that public authority defendants should be in a more privileged position than other defendants in this connection is not, in my view, maintainable. Secondly, it is simply unfair on the claimant or his lawyers if, at least in the absence of special factors, he does not recover his costs of bringing wholly successful proceedings, provided that they have been properly brought and conducted. Thirdly, while defendants may be more ready to concede a claim rather than fight it if they know that they will not thereby be liable for the claimant’s costs, it can forcefully be said that the fact that, if defendants know they will have to pay the claimant’s costs, it would be a powerful incentive to concede the claim sooner rather than later. Fourthly, if the defendants wish to settle, the time to do so is before proceedings are issued: that is one of the main reasons for the introduction of the Protocol.

Secondly, it may be said that, because of the three month time limit, there will often be less time available for defendants in a public law claim to consider the merits of the claimant’s case than in a private law claim, where the more generous time limits in the Limitation Act 1980 normally apply. In my opinion, in some cases, that factor might justify making a more generous order for costs from the defendants’ perspective than the analysis in the previous section of this judgment might otherwise suggest. It would not be good enough for defendants to say that they had not got round to dealing with the claimant’s claim because of [their] ‘heavy workload’ or ‘constraints upon [their] resources’ (see Bahta para 63). However, where the claim is one which reasonably requires more time to investigate than is available before the three month period runs out, there may be a powerful case for defendants who thereafter concede the claim not being liable for any or some of the claimant’s costs. However, that does not seem to me to give rise to a difference in principle between Administrative Court litigation and other civil litigation – for instance, where the letter before action is written very shortly before the expiry of the limitation period.

A third argument is that defendants sometimes concede claims in the Administrative Court simply because it is not worth the candle fighting the case, or because the claim is justified only on a relatively technical ground, such as a procedural defect. In the first type of case, it is said to be unfair to penalise the defendants in costs for taking a view which, while not necessarily reflecting the legal merits, is realistic and proportionate. In the second type of case, the court normally then remits the decision to the defendants, who then go on to reconsider and often arrive at the same substantive conclusion as before. In the main, it seems to me that the answer to this is that the defendants should make up their mind to concede the claim for such reasons before proceedings are issued. That is one of the main purposes of the Protocol, and, if defendants delay considering whether they should concede a claim, that should not be a reason for depriving the claimant of his costs. If, in fact, the only reason the defendants did not take that course was that they had insufficient time to consider the claimant’s claim, one is back to the point discussed in the preceding paragraph. In some cases, Pill LJ’s scepticism about this argument, as expressed at para 63 of Bahta, will apply; in others, the defendants may be short of resources, but, as mentioned, that is not a good reason for depriving the claimant of his costs.

A fourth argument is that, in some public law cases, the law (or what is understood to be the law) changes after the issue of proceedings, so that what appears to be a weak claim becomes transformed into a strong claim. An obvious example is where the Supreme Court overrules previous Court of Appeal authority, so that the defendants who (justifiably) thought they had a very strong case suddenly realise that they are very much on the back foot. In Bahta,Pill LJ was unimpressed with the UKBA’s argument that they were entitled to refuse to agree the claimants’ cases on the ground that, although Court of Appeal authority was against the UKBA, they were entitled to act on the assumption that the Supreme Court might take a different view, which, in the event, they did not. By parity of reasoning, it may seem rather harsh to visit defendants with liability for all the claimant’s costs, because they assumed that the law was as the Court of Appeal had decided, until the Supreme Court took a different view. In such a case, however, while the defendants have a real argument for saying that they should not pay all the claimant’s costs, the claimant can nonetheless raise all the normal reasons for receiving his costs. This argument would apply equally to ordinary civil cases.

A fifth argument, which also applies to ordinary civil cases, is based on a number of miscellaneous possible factual situations which arise in Administrative Court cases. They involve various failings on the part of the claimant, such as not having set out his case clearly in his letter before action, adding to his evidence well after the issue of proceedings, including a claim which does not succeed, or pursuing the claim in an unreasonable manner. In cases where such an argument is raised by the defendants, the court may well be persuaded either that it would be wrong to award the claimant any costs for the reasons canvassed by Chadwick LJ in BCT at para 24, or that the claimant should only receive a proportion of his costs. As in any civil litigation, a claimant who succeeds is only entitled to his costs in the absence of good reason to the contrary. Thus, where the claim has been conceded in a consent order which does not deal with costs, the court will not award the claimant all or any of his costs save to the extent that it is satisfied, without looking at matters in detail, that the claimant is so entitled.

Accordingly, I conclude that the position should be no different for litigation in the Administrative Court from what it is in general civil litigation. In that connection, at any rate at first sight, there may appear to be a degree of tension between this conclusion, which applies the ‘general rule’ in CPR 44.3(2)(a), and the fifth guideline in Boxall, at least in a case where the settlement involves the defendants effectively conceding that the claimant is entitled to the relief which he seeks. In such a case, the claimant is almost always the successful party, and should therefore, at least prima facie, be entitled to his costs, whereas the fifth guideline seems to suggest that the default position is that there should be no order for costs. Similarly, there could be said to be a degree of tension between what was said in paras 63-5, and the view expressed in para 66, of Bahta .

In my view, however, on closer analysis, there is no inconsistency in either case, essentially for reasons already discussed. Where, as happened in Bahta, a claimant obtains all the relief which he seeks, whether by consent or after a contested hearing, he is undoubtedly the successful party, who is entitled to all his costs, unless there is a good reason to the contrary. However, where the claimant obtains only some of the relief which he is seeking (either by consent or after a contested trial), as in Boxall and Scott, the position on costs is obviously more nuanced. Thus, as in those two cases, there may be an argument as to which party was more ‘successful’ (in the light of the relief which was sought and not obtained), or, even if the claimant is accepted to be the successful party, there may be an argument as to whether the importance of the issue, or costs relating to the issue, on which he failed.

Thus, in Administrative Court cases, just as in other civil litigation, particularly where a claim has been settled, there is, in my view, a sharp difference between (i) a case where a claimant has been wholly successful whether following a contested hearing or pursuant to a settlement, and (ii) a case where he has only succeeded in part following a contested hearing, or pursuant to a settlement, and (iii) a case where there has been some compromise which does not actually reflect the claimant’s claims. While in every case, the allocation of costs will depend on the specific facts, there are some points which can be made about these different types of case.

In case (i), it is hard to see why the claimant should not recover all his costs, unless there is some good reason to the contrary. Whether pursuant to judgment following a contested hearing, or by virtue of a settlement, the claimant can, at least absent special circumstances, say that he has been vindicated, and, as the successful party, that he should recover his costs. In the latter case, the defendants can no doubt say that they were realistic in settling, and should not be penalised in costs, but the answer to that point is that the defendants should, on that basis, have settled before the proceedings were issued: that is one of the main points of the pre-action protocols. Ultimately, it seems to me that Bahta was decided on this basis.

In case (ii), when deciding how to allocate liability for costs after a trial, the court will normally determine questions such as how reasonable the claimant was in pursuing the unsuccessful claim, how important it was compared with the successful claim, and how much the costs were increased as a result of the claimant pursuing the unsuccessful claim. Given that there will have been a hearing, the court will be in a reasonably good position to make findings on such questions. However, where there has been a settlement, the court will, at least normally, be in a significantly worse position to make findings on such issues than where the case has been fought out. In many such cases, the court will be able to form a view as to the appropriate costs order based on such issues; in other cases, it will be much more difficult. I would accept the argument that, where the parties have settled the claimant’s substantive claims on the basis that he succeeds in part, but only in part, there is often much to be said for concluding that there is no order for costs. That I think was the approach adopted in Scott. However, where there is not a clear winner, so much would depend on the particular facts. In some such cases, it may help to consider who would have won if the matter had proceeded to trial, as, if it is tolerably clear, it may, for instance support or undermine the contention that one of the two claims was stronger than the other. Boxall appears to have been such case.

In case (iii), the court is often unable to gauge whether there is a successful party in any respect, and, if so, who it is. In such cases, therefore, there is an even more powerful argument that the default position should be no order for costs. However, in some such cases, it may well be sensible to look at the underlying claims and inquire whether it was tolerably clear who would have won if the matter had not settled. If it is, then that may well strongly support the contention that the party who would have won did better out of the settlement, and therefore did win.

Having said that, I should add that I have read what Stanley Burnton LJ says in his judgment, and I agree with it.

Having given such general guidance on costs issues in relation to Administrative Court cases which settle on all issues save costs, it is right to emphasise that, as in most cases involving judicial guidance on costs, each case turns on its own facts. A particular case may have an unusual feature which would, or at least could, justify departing from what would otherwise be the appropriate costs order.

In this case, the only real issue was whether the appellant was born in 1994 or 1996. His case from the start has been that the latter date was correct, and there can be no doubt but that he had to bring these proceedings to establish that case. The respondents had ample time, at least on the face of it, to deal with the issue, not only before the claim was issued in February 2009, but also over the exceptionally long period between that date and the date when permission was granted, July 2010. Nonetheless, it was only in February 2011 that the appellant’s claim was conceded. The appellant’s case was clearly spelt out in two successive letters which complied with the Protocol, whereas the respondents failed to acknowledge service until well out of time (although this caused no prejudice and gave rise to no costs implications).

In those circumstances, it seems to me that, unless there are good reasons to the contrary, the respondents should have been ordered to pay the appellant’s costs. There are three possible reasons for not making such an order. The first is that the respondents settled on the assumption that there would be no order for costs. The second reason is the change in the perceived legal position as a result of the Supreme Court’s decision in R (A) v Croydon in November 2009. The third reason is the substantial amount of evidence and the surprisingly difficult nature of the issue, including the change in the weight to be given to Dr Birch’s views.

I am not very impressed with the first reason. The respondents well knew that there would be a risk that they would have to pay some costs when they settled, as it would have been clear to them that the appellant was going to ask for his costs. Further, even before the decision in Bahta, nobody had suggested that the Boxall guidelines were immutable; indeed, in Boxall itself, the local authority had to pay some of the claimant’s costs. In addition, the actual decision in Bahta is hard to reconcile with the respondent’s argument.

The second reason is the one which weighed with the Judge, as is clear from his reference to ‘the dynamic development of this area of the law while the claim was live’ in his reasons for making no order for costs. In my view, so long as the law was thought to be whether the respondents’ decision that the appellant was born in 1996 had to be shown to be ‘Wednesbury unreasonable’ before the appellant could succeed, the respondents’ decision to fight the claim was plainly reasonable: indeed, the respondents appear to me to have had a strong case.

For part of that period, the respondents’ case is also assisted by an aspect of the third reason, namely that the appellant’s solicitors’ chosen expert, Dr Birch, appeared, at least to an extent, to support their view. However, once the Supreme Court had decided that the issue of the appellant’s age was for the court to decide, and Dr Birch’s credibility had been severely undermined, it seems to me that the ‘dynamic development’ referred to by the Judge had come to an end. The respondents would then, of course, have had to re-assess the situation.

I do not think that the Judge’s decision to make no order for costs can stand. He adopted the wrong approach in the same way as the five Administrative Court Judges in Bahta and, as Sullivan LJ implied when granting permission to appeal, that is scarcely his fault: Bahta was decided two months after his decision. However, the change in the perceived law by the Supreme Court and the changing status of Dr Birch, who was a witness proposed by the claimant, can properly justify a reduction in the costs awarded to the claimant, especially as the Judge plainly thought so (and we should have regard to that as he is the primary decider).

In my view, it would be right to reflect the points which the respondents have made by ordering that the respondents should pay 50% of the appellant’s costs in respect of the period up to 26 July 2010, when Judge McMullen gave the appellant permission to bring these proceedings, and that the respondents should pay 100% of the appellant’s costs in respect of the period thereafter. In each case, the costs should be assessed, if they cannot be agreed, on the standard basis.

I agree.

I also agree.

The consequence of our decision should be a greater willingness on the part of the parties to judicial review proceedings, at first instance and on appeal, to agree not only the substantive provision of the order to be made by the Court, but also the issue of costs. Settlements in which the question of costs is left to be determined by the Court at a later date are common, and perhaps too common. Parties can no longer assume that the likely order is no order as to costs, even where one party or another has conceded the whole, or substantially the whole, of the other side’s case.

A successful negotiation of costs issues is likely to be cost effective, saving the costs of subsequent written submissions and saving the time of the judge who is required to determine costs. It is in both parties’ interests to address the question of comprehensive settlement as early as possible.

Where the parties are unable to agree costs, and they are left to be determined by the Court, it is important that both the work and costs involved in preparing the parties’ submissions on costs, and the material the judge is asked to consider, are proportionate to the amount at stake. No order for costs will be the default order when the judge cannot without disproportionate expenditure of judicial time, if at all, fairly and sensibly make an order in favour of either party. This is not to say that there are not cases where the merits can be determined and no order for costs can be seen to be the appropriate order; but in such cases that order is not a default order, but an order made on the merits.

In the present case, the Master of the Rolls has set out the procedural history in some detail, but the pertinent facts are few. The respondents’ maintenance of their position was entirely reasonable while the law was as it was generally thought to be before the decision of the Supreme Court in R (A) v Croydon. That decision led eventually to the order His Honour Judge McMullen QC of 26 July 2010. The respondents then had to reconsider their case, if they had not already done so. The appellant’s reliance on the evidence of Dr Birch may have been ill-advised, but ultimately it was his case, based on his account of his age, that prevailed. The respondent agreed not merely to re-assess his age, but that his age was as he contended it to be: i.e., they conceded the entirety of his claim. These are the matters that led me to conclude that the appropriate order as to costs is that identified in paragraph 72 of the judgment of the Master of the Rolls.