The Commissioner is unsure of the policy to whose formulation or development the Assessment is said to relate. On the assumption that it is the selection of towns for funding through the £3.6 billion Towns Fund, the Commissioner’s position is that the Assessment does not relate to the formulation or development of that policy.

In the Commissioner’s view the Assessment was an administrative document concerned with the implementation of the Towns Fund policy rather than its formulation or development. It did not influence the design of the Towns Fund policy or the process by which towns were selected for funding. It merely recorded and explained these steps. It was not concerned with reviewing or improving the existing policy. The towns had already been selected by Ministers.

The Commissioner submits that the record of the qualified person’s (QP’s) opinion dated 26 April 2021 does not include any reasons from the Minister. The submission to the Minister simply asserted that the necessary prejudice/inhibition would be likely to occur rather than explaining how. The Commissioner maintains that the opinion was not substantively reasonable where the QP lacked any evidence on which to base that opinion and provided no reasons for that opinion.

[This is no longer in issue]

The Commissioner’s position is that the public interest favours disclosure. There is a general public interest in transparency. There is a legitimate interest in the process by which the £3.6 billion Towns Fund was allocated. There was a plausible suspicion of wrongdoing over the process by which that money was allocated, indicated by the PAC Report. These factors outweigh the public interest factors in favour of withholding the information.

The Department submits that there is no requirement that the policy-making process still be live in order for the exemption to bite: Cabinet Office v Information Commissioner [2018] UKUT at paragraph 29.

It submits that the question is whether the QP’s opinion is substantively, not procedurally, reasonable. The opinion is not conclusive but must be accorded a measure of respect.

The Department submits that in considering a possible ‘chilling effect’, that effect need not be proved by evidence in any particular case (Department of Health and Social Care v Information Commissioner [2020] UKUT 299 at paragraphs 28 – 30) but the tribunal is entitled to scrutinise claims that there would be such an effect in context (Davies v Information Commissioner [2019] 1WLR 6641 at paragraphs 25 – 30).

The Department submits that the departmental Accounting Officer (“AO”) is personally responsible and accountable to Parliament for their department’s use of public money (see Chapter 3 of Managing Public Money (“MPM”)). The AO is normally also the Permanent Secretary and therefore also owes a duty in that capacity to serve the Departmental Minister. The production of an AO assessment is part of the process of assurance that the standards in MPM are met.

The Department sets out the relevant extracts from the 2017 AO Guidance and notes that the Guidance recognises that the AO assessment itself may affect the formulation or development of the policy.

The Department submits that the 2017 AO Guidance was clear that it would not normally be appropriate for the full assessment to be published. That reflects a settled understanding of the importance of ensuring a safe space for the AO in relation to the production of an AO assessment, having regard to the personal nature of the opinion and the role of the AO assessment in ensuring that the standards in MPM are met.

The AO guidance was revised in December 2021 (“the 2021 AO Guidance”). Paragraph 1.11 notes the expectation that a summary of the AO assessment will be prepared and published.

The Department submits that if the AO concludes that the standards are not met, that may require that the relevant policy (in this case the method and selection of towns for the Towns Fund) is modified. The Department submits that this is not implementation but part of ensuring that policy is formulated and developed in accordance with MPM. The Assessment is developed alongside policy discussions, rather than being written only after the policy was settled. It plainly related to the formulation and development of government policy even if it did not conclude that the policy needed to be changed.

The Department submits that the QP’s opinion that disclosure would lead to prejudice was a reasonable one to hold. The Commissioner focusses on putative flaws in the process and/or record of reasoning rather than on its substantive reasonableness. The QP’s opinion is consistent with the opinions of others and with the AO Guidance.

The Department submits that the opinion was plainly objectively and substantively reasonable. The preservation of a safe space is particularly important because :

[No longer in issue]

The Department submits that the interest in transparency is outweighed because:

We read an open and a closed bundle and a number of additional documents including an additional QP’s opinion (and submissions to the QP) dated 23 January 2023. With the agreement of the Department, the submissions to the QPs and both opinions are now to be treated as open material.

The bundles include an open and a closed witness statement from David Thomas, senior civil servant, dated 14 October 2022. We heard oral evidence from David Thomas. Mr Thomas was an impressive witness and appropriately indicated where matters were outside his knowledge. He made concessions where appropriate. We attach significant weight to his experience and expertise and have highlighted the reasons for any departure from his evidence within our discussions and conclusions below.

After the hearing we were provided with an amended marked-up copy of the Assessment which forms part of the closed material.

The closed bundle consists of:

It is necessary to withhold the above information because it refers to the content of the withheld information, and to do otherwise would defeat the purpose of the proceedings.

The following is a gist of the closed session, prepared by the parties and approved by the tribunal:

The relevant parts of sections 1 and 2 FOIA provide:

APPGER v Information Commissioner and Foreign and Commonwealth Office [2016] AACR 5 gives guidance on how the balancing exercise required by section 2(2)(b) of FOIA should be carried out:

Section 35(a) FOIA provides as follows:

Section 35 is a class-based exemption: prejudice does not need to be established for it to be engaged. It is not an absolute exemption. The tribunal must consider if, in all the circumstances of the case, the public interest in maintaining the exemption outweighs the public interest in disclosure.

Case law has established in the FOIA context that “relates to” carries a broad meaning (see APPGER at paragraphs 13-25). In UCAS v Information Commissioner and Lord Lucas [2015] AACR 25 at paragraph 46 the Upper Tribunal approved the approach of the FTT in the APPGER case where it said that “relates to” means that there must be “some connection” with the information or that the information “touches or stands in relation to” the object of the statutory provision.

The question of whether the policy-making process is still ‘live’ is an issue that goes to the assessment of the public interest balancing test, and not to whether the section 35(1)(a) exemption is engaged in the first place (Morland v Cabinet Office [2018] UKUT 67 (AAC).

The intersection between the timing of the FOIA request and its relevance to the public interest balancing test is helpfully analysed by the First-tier Tribunal in Department for Education and Skills v Information Commissioner and the Evening Standard (EA/2006/0006) (“DFES”) at paragraph 75(iv)-(v) (a decision approved in Office of Government Commerce v Information Commissioner [2008] EWHC 774 (Admin); [2010] QB 98 (“OGC”) at paragraphs 79 and 100-101):

The public interest can wax and wane and the need for a safe space changes over time in relation to development of policy.

If disclosure is likely to intrude upon the safe space then there will, in general terms, be significant public interest in maintaining the exemption, but this has to be assessed on a case by case basis.

In considering the factors that militate against disclosure the primary focus should be on the particular interest which the exemption is designed to protect, in this case the efficient, effective and high-quality formulation and development of government policy.

In relation to ‘chilling effect’ arguments, the tribunal is assisted by the following paragraphs from the Upper Tribunal decision in Davies v IC and The Cabinet Office [2019] UKUT 185 (AAC):

Section 36 provides in material part that:

It is for the tribunal to assess whether the qualified person’s (QP’s) opinion that any of the listed prejudices/inhibitions would or would be likely to occur is reasonable, but that opinion ought to be afforded a measure of respect: Information Commissioner v Malnick [2018] UKUT 72 (AAC), [2018] AACR 29 at paragraphs 28-29 and 47.

It is not an absolute exemption.

It was largely agreed between the parties as to what was in scope of the principle of parliamentary privilege and the approach to be taken by the tribunal, namely that in so far as what is relied on are statements of fact and in particular statements of agreed fact, there is no objection to relying on that material. It was agreed, in essence, that the tribunal could treat the NAO report as covered by parliamentary privilege without the need to make a ruling on that issue.

The Commissioner submitted that the date on which the tribunal is required to conduct the public interest balance depends on whether the public authority responds to the request within the statutory timeframe set out in sections 10 and 17 FOIA. If the response is in time, it is considered at the date of that response. Where the response is late, it is considered at the latest date by which the public authority should have responded. In this case, the responses were late and the dates are therefore 9 December 2020 for the first request and 16 December 2020 for the second request. The disclosure of the Summary cannot therefore be taken into account.

In support of this submission Mr. Jackson highlighted the reasoning of the Upper Tribunal at paragraphs 65-67 of Montague v Information Commissioner [2022] UKUT 104 (AAC), which relies on the actions that the public authority is required to take by Part I FOIA.

Mr. Anderson submitted that the requirement with which the Upper Tribunal was concerned was the requirement to respond to the request. Once the public authority has done so, it has done what FOIA requires. That is true even if the response was later than it was required to be. There is nothing in Montague to support the position that the relevant time is when a request ought to have been made even if, in fact, there was no response by that time.

It was common ground that the question was whether or not the particular information in dispute should be disclosed, not whether AO assessments in general should be disclosed. Mr. Anderson submitted that, nonetheless, there are particular structural features of AO assessments and the role of AOs that are relevant. Mr. Jackson accepted that wider points can be relevant, but reminded the tribunal that our decision does not set a precedent therefore wider concerns about future AO assessments being disclosed are not relevant.

Mr. Jackson made the following overarching points:

Mr. Jackson submitted that there is nothing unusual about the personal responsibility of the AO. This is not in itself a factor that attracts any public interest or legal consequence one way or another. The AO is the permanent secretary and therefore would be least susceptible to chilling effect concerns.

Mr. Anderson submitted that the timing of the request does not give the answer to whether the information relates to the formulation or development of government policy. Whether or not the policy formulation or development was live at the date of the request is irrelevant to the engagement of the exemption. ‘Relates to’ has a broad meaning – it has to have some connection with the information.

Mr. Anderson submitted that the information did not simply relate to the implementation or administration of policy. The Assessment is an integral part of the policy formulation and development process. It is not a retrospective audit of a decision that has been taken, it is one of the ‘keys’, along with treasury approval, to getting to a final decision.

Mr. Jackson submitted that although ‘relates to’ has a broad meaning, it “should not be read with uncritical liberalism as extending to the furthest stretch of its indeterminacy, but instead must be read in a more limited sense so as to provide an intelligible boundary, suitable to the statutory context” (Department of Health v ICO [2017] 1 WLR, quoting with implied approval from the FTT decision).

Mr. Jackson submitted that previous iterations and drafts of the Assessment may well relate to the formulation and development of policy. The Assessment does not. He argued that the policy is not the entire process of the Towns fund over 5 years, the policy in question is the ministerial selection process of which the results were announced on 6 September 2019. This is relevant to whether the Assessment relates to the policy in question.

The Department does not dispute that there is a general public interest in disclosure and that there are features of this case which might stand in favour of disclosure, including the amount of money at stake, a degree of media attention and political comment. Mr. Anderson submitted that even to make a finding of a reasonable suspicion of wrongdoing would be to go beyond what can properly be found on the evidence, in particular given the limitations of the extent to which the tribunal can rely on the various reports to which parliamentary privilege attaches. It is fair to note that there were members of the public who were concerned that there may have been wrongdoing, or who raised those concerns in correspondence.

Mr Anderson submitted that a number of factors diminished the public interest in disclosure:

In relation to the factors in favour of withholding the information the arguments related to the potential chilling effect and the importance of maintaining a safe space.

Mr. Anderson submitted that the tribunal must avoid two extreme positions:

It was submitted on behalf of the Department that the tribunal must avoid either extreme and focus on the particular information as well as the structural features of the AO in that role. The tribunal must not put out of its mind the risks that arise from compelled disclosure of such assessments in relation to how future assessments might be prepared. That is always the risk that arguments on chilling effect and safe space go to.

Mr. Anderson submitted that the tribunal has heard clear and candid evidence from Mr. Thomas, a senior civil servant, that if those drafting the submissions in the Assessment had to have more in mind the risk of disclosure to a wider audience, that may have an effect in terms of couching how their submissions were made. Any person communicating will tailor the way they communicate to who they believe their audience will be. This will be different if the anticipated audience is effectively contained to those at a high level within the policy making process compared to the wider world. An important feature of the AO process is that it gives the opportunity for the AO to say ‘pause’ or to highlight the issues that need particular care and attention.

He highlighted other features of the role that point to why ensuring a safe space is particularly important in this context:

Mr. Anderson submitted that the Assessment is a good example of a case where those features are salient, not least because it involved a lot of money on an issue that was potentially controversial and therefore where it was important to be as frank as possible at that stage in the process.

Whilst Mr. Anderson accepted that it was right that disclosure does not set a legal precedent for disclosure of future AO assessments, the tribunal has to consider if there may be indirect and wider consequences as a matter of fact when considering the potential chilling effect of a decision (DHSC paragraph 26).

The Department does not rely on the wider policy connections to establish that the section 35 exemption is engaged, it is more relevant to liveness in relation to the public interest balance.

It was submitted that relying on the Guidance as a factor in favour of the respondent’s case is peculiar when the thrust of that guidance plainly emphasises that generally AO assessments won’t be published in full.

Mr. Anderson argued that the conscientiousness of civil servants cuts both ways. The tribunal has heard evidence from a conscientious senior civil servant who has taken time out from his work to explain why particular effects may arise from disclosing certain types of information and his evidence should be given significant weight. Mr. Thomas was a candid and helpful witness, who could speak both to the general structure and in relation to the specific information in issue.

Mr. Anderson submitted that there is nothing as a matter of law that says that a tribunal should reject arguments on a chilling effect. It is a matter of fact to be determined in the circumstances of every case.

The tribunal asked Mr. Anderson about the effect on arguments based on the potential chilling effect of the tribunal’s decision of the fact that the NAO decides what parts of the Assessment would be put into the public domain, without any veto for the AO.

Mr. Anderson submitted that, as Mr Thomas said in his evidence, there is a degree of dialogue between the NAO and the AO in relation to the final form of the report which may include considerations of confidentiality amongst other things so there is scope within that for the AO to have a role in deciding what will go in the final report. Further, publication as part of the same accountability structure is quite different from the prospect of material going out into the wider world before the wider public when it is not subject to that filter of the NAO considering what is relevant and what should not be published in its report, and which was necessary to be published in its report for the purposes of the functions it is performing.

Mr. Jackson submitted that the public interest lies in favour of disclosure, taking account of the following:

In relation to the counter arguments, Mr. Jackson submitted:

Mr. Jackson noted that the chilling effect argument was made in different ways. It would lead to officials couching their language differently. It would undermine the middle space that an AO Assessment occupies before a ministerial direction is sought. It gives an opportunity for policymakers to raise a red flag over something that might be problematic. These are all arguments that if disclosure was ordered, people might behave differently in the future.

Mr. Jackson submitted that this was considered to be a poor argument because of the conscientiousness and competence of the high quality civil servants we are lucky to have. The idea that they would not provide free and frank advice in the future is rightly deprecated. Mr. Thomas was concerned, but he need not be, about the effect on the permanent secretary. It was suggested that the permanent secretary has a relationship with the Minister that needs to be protected, but there is no need to treat permanent secretaries as if they are fragile.

The tribunal asked Mr. Jackson about the effect on arguments based on the potential chilling effect of the tribunal’s decision of the fact that the NAO decides what parts of the Assessment would be put into the public domain, without any veto for the AO.

Mr. Jackson submitted that this does go to show that the chilling effect argument is overstated. Mr Thomas was clear that while there is a negotiation between the AO and the NAO the final decision lies with the Comptroller general of the NAO and if they are of the view that the information should be published then that is the decision that is made. That seems to have been the case for some time and hasn’t caused any great concerns.

The Commissioner accepts that the Government is entitled to a safe space, but on the facts of this case, even if the information did still relate to the formulation and development of policy, the need for a safe space was rapidly diminishing at the date of the Assessment given how soon after the policy was announced. By the time of the response to the request in January 2021 any public interest in maintaining the exemption was significantly diminished.

Mr. Anderson submitted that in accordance with Malnick the test is one of substantive reasonableness:

He submitted that paragraphs 54 and 55 of Malnick highlight the unintended consequences of the alternative approach: first, that it would mean that the decision-making process requirements were more demanding at the initial gateway stage than they were at the substantive stage of considering the public interest test; second, if a procedural error prevents a public authority from relying on section 36, then (absent any other exemption applying) the disputed information must be disclosed, whatever the potential prejudice.

Mr. Anderson argued that this must apply to this appeal, where it is effectively said that the submission underlying the opinion was too thin in its reasons or conflated the public interest test with the prejudice test. In any event, the second QP opinion provides a fuller submission with a wider range of factors.

Mr. Jackson submitted that in accordance with Davies v ICO [2019] 1 WLR for an opinion under section 36 to be substantively reasonable it must appropriately identify, prove, explain and examine the harm and it must address in a properly reasoned and balanced objective way the weaknesses in the chilling effect argument. The first opinion clearly falls short. The second opinion falls short where the reasons are exactly of the generic and general kind that was deprecated in Davies.

In relation to Davies and the approach to the reasonableness of the QP opinion, it was submitted by Mr. Anderson on behalf of the Department that the issue in that case concerned the reasons the first tribunal had given for its conclusion that the QP’s opinion was a reasonable opinion to hold and effectively the Upper Tribunal concluded that such reasons hadn’t been given. This appeal is very different in that there are plainly sufficient reasons for concluding that prejudice of the necessary sort would arise, which demonstrates that the opinion was substantively reasonable. The alternative would be to make the gateway to section 36 significantly elevated and burdensome. Mr. Anderson submitted that it is plain the opinions were substantively reasonable.

It was agreed that the public interest cannot be aggregated, but that the factors under each section were the same in this appeal.

The tribunal’s remit is governed by section 58 FOIA. This requires the tribunal to consider whether the decision made by the Commissioner is in accordance with the law or, where the Commissioner’s decision involved exercising discretion, whether he should have exercised it differently. The tribunal may receive evidence that was not before the Commissioner and may make different findings of fact from the Commissioner.

The issues we have to determine are:

The parties were in agreement that certain paragraphs of the Assessment were exempt under section 42, and, having reviewed the withheld information, the tribunal was content to proceed on that basis.

The tribunal adopts the agreed position from the parties. In effect, we may rely on the undisputed facts contained in the NAO and PAC reports (and any other material covered by parliamentary privilege) but not on the opinions contained therein.

In essence the conclusion in Montague was based on the function of the Commissioner under section 50(1) FOIA which was to decide whether a request for information had been dealt with in accordance with the requirements of Part 1 of FOIA. As there was no requirement in Part 1 FOIA for the public authority to review its decision refusing the request, the relevant date cannot be the date of any such review. Once a decision has been made and communicated, this brings an end to that which the public authority is required by law to do by Part 1 FOIA.

We do not agree that anything in the reasoning in Montague supports an argument that the relevant date is the date at which the public authority should have responded. There a number of references in the decision to an ‘in-time’ review decision. The Upper Tribunal could have used, but did not use, these words in relation to the initial refusal decision.

In contrast, Montague states repeatedly that the relevant date is the date of the initial refusal decision. We consider that we are bound by the Upper Tribunal decision to assess the public interest at the date of the initial refusal, not at the date by which the Department should have refused. Whether or not this should be the position is not a matter for this tribunal.

The relevant date for the purposes of this appeal is therefore 14 January 2021.

We find that all the withheld information relates to the formulation or development of government policy. The tribunal notes that on the latest version of the document entitled “Marked up copy of AO advice” the coloured highlighting indicates that part of paragraph 3 is withheld only under section 36 and not under section 35. This was not the position in the earlier version of that document in the closed bundle, where it was stated to be withheld under both sections.

We understood Mr. Anderson’s submissions to be that the entire Assessment related to the formulation of government policy, and he confirmed this in response to a question from the Judge. Our conclusions in this part of the decision relate to all the withheld information, including that in paragraph 3.

The relevant policy is the ministerial selection process of those towns to be invited to bid for funding. We accept that the Assessment took place right at the end of the process of development and formulation of the relevant policy. We accept that the Assessment is largely an update on AO issues relating to the planned announcement on the selection of 100 towns, such selection having already been made. The recommendation in the Assessment is to “note the Accounting Officer risks and associated legal advice”. No subsequent policy development or formulation took place. This is subject to a qualification that we deal with in our closed reasons.

We accept Mr. Thomas’ evidence that the Assessment relates to the formulation and development of that policy because its content is the outcome of an iterative process involving discussions that took place during the process of development and formulation of policy. Further although it takes place at the end of the policy development and formulation, it looks back on and evaluates that process. The question of whether the policy making process was still live goes to the public interest test not to whether the section is engaged.

For those reasons we accept that the exemption is engaged in relation to all the withheld information.

The purpose of section 35 is to protect the effective, efficient and high-quality formulation and development of government policy and to protect good government. It reflects and protects some longstanding constitutional conventions of government. It reserves a safe space to consider policy options in private – civil servants and subject experts need to be able to engage in free and frank discussion of all the policy options internally, to be able to expose their merits and demerits and possible implications. It is in the public interest that officials and ministers have “time and space…to hammer out policy exploring safe and radical options alike, without the threat of lurid headlines depicting what has merely been broached as agreed policy” (DfES para 75(iv), approved in OGC).

Under section 35 there is no space where confidentiality can be assured because it is not an absolute exemption.

The need for a safe space is much greater when development of that policy is nearer the live end of the spectrum at the relevant date.

In considering the weight of the safe space in this appeal, we have taken particular account of the following.

The AO assessment is an important gateway in the policy formulation process – it is one of the keys to a final decision. It is important that the AO assessment properly identifies any issues and appropriately challenges the policy under consideration. The AO is personally responsible to Parliament which underscores the importance of the AO assessment and the importance of getting frank advice. The safe space of the AO assessment is important as an intermediate ‘safe space’ before officials take the step of seeking a published Ministerial direction. The AO, as permanent secretary, has to give advice in a way that manages to be full and frank and avoids undermining the policies that they are required to serve and implement. If individuals are more conservative in how they couch their language in an AO assessment that diminishes the value of the free and frank exchange of views and information in the intermediate space and is inimical to good policy making. We accept that all those matters would carry very significant weight, particularly if the policy in question were live at the relevant date.

We have considered the liveness of the policy at the relevant date. It was necessary to put some of our reasoning on this issue in the closed annex, because otherwise the withheld information would have been disclosed. Our open findings on the liveness of the policy are as follows.

First, the specific policy to which the Assessment related was no longer live at the relevant date. The Assessment related specifically to the planned announcement on the selection of towns to work with and develop town deals and AO issues relating to that announcement. It considered the methodology developed by officials for selecting eligible towns, initially to allocate some of the capacity funding to help places to develop new economic visions for their areas and to apply for substantive funding.

Even at the stage that the Assessment was written on 5 September 2019, the towns had already been selected. Whilst the policy formulation and development process was not technically over, because treasury approval had not yet been given, it was substantively over. Therefore even at the date the Assessment was drafted, put inelegantly, the process was some way from the livest end of the spectrum.

The final list of towns was announced on 6 September 2019, 4 months prior to the relevant date. At the relevant date we find that policy discussion and development in relation to the selection of the initial list of towns was not in any sense live.

We accept that the ‘liveness’ of a policy is not black and white. Further we accept that the public interest in maintaining a safe space waxes and wanes and does not evaporate the moment a policy is announced. We take into account the fact that in January 2021 related work was ongoing in relation to the next stages of the towns fund process and we find that this is likely to have involved matters that are more properly described as policy than implementation. However we heard no evidence on any specific harm or impact that disclosure of this specific information might have had on any ongoing related policy work.

In relation to the specific content of the withheld information we have made further findings but it was necessary to put this reasoning in the closed annex, otherwise the withheld information would have been disclosed.

In the circumstances, we find that the need for a safe space had reduced to some extent by 14 January 2021. Although some broadly related policy work was likely to have been continuing, the specific issues with which the Assessment was concerned were not. Further, for the reasons set out in the closed annex we do not think that the specific information contained in the Assessment was likely to lead to harm if published at the relevant date.

Despite those findings, given the particular features of AO assessments identified above, we still place significant weight on preserving the safe space in the public interest balance.

The Department also relied on the effect that disclosure of the requested information would have on the conduct or the candour of civil servants in the future when drafting Accounting Officer assessments. We have had regard to the experience and expertise of Mr. Thomas, and of the measured and considered way in which he gave evidence and answered questions.

We accept that he has formed the view that, were Accounting Officer assessments to be published, contributors would self-censor to avoid creating material for the public domain that attacks or undermines the government’s position. His view is that self-censorship would be an inevitable consequence of expecting that AO assessments could be published against the wishes of the AO and Ministers.

We accept that this view is honestly held and is formed in the light of his experience and expertise, to which we give due respect. However, having considered Mr. Thomas’ evidence carefully we are not persuaded that our decision risks having those consequences for the following reasons.

In relation to the Assessment specifically, Mr Thomas’ opinion was as follows (paragraph 49 of the open witness statement):

Mr. Thomas also highlighted the particular risks arising from the tension between the AO’s duty to the minister as a Permanent Secretary and to Parliament as AO (paragraph 50 of the open witness statement):

In open oral evidence he added the following:

Mr. Thomas outlined a further risk that ‘routine publication’ of full AO advice risks blurring the importance of a ministerial direction at paragraph 51.

When it was put to Mr. Thomas in cross-examination that it was not right to say that civil servants would ‘self-censor’ because of the risk of freedom of information disclosure, because that would go against their role and responsibilities, he replied in open:

He also stated in open, when it was put to him that civil servants would be aware of the risk of disclosure from their knowledge of FOIA and because it was contemplated in the Guidance:

Mr. Thomas gave the following additional open evidence at paragraphs 61 and 62 of his open witness statement:

The Judge asked Mr. Thomas why the tribunal reaching a particular conclusion as to where the public interest lay in a particular case, taking into account the specifics of that case including, for example, the sensitivity or otherwise of the material, the liveness of the policy etc. would be likely to alter the way that AO assessments were written in the future, given that the FOIA framework is already known about by the people who are writing the assessments.

Mr. Thomas replied, in open:

We have the utmost respect for Mr. Thomas’ experience and expertise, and he gave evidence in a measured and considered way. However, we do not think that his evidence answered the question why a decision by this tribunal, on these particular facts, to the effect that the public interest favoured disclosure, would lead to a change in behaviour.

If Accounting Officers are properly informed, and, at that level of seniority, we assume that they will be, they will appreciate that the greater the public interest in the disclosure of confidential, candid and frank information, the more likely it is that it will be disclosed.

Mr. Thomas is right in his assumption that ‘there tends to be a bit more protection’ afforded to documents such as ministerial submissions and AO assessments than other papers the government might hold. The nature of the AO assessment is what gives rise to the engagement of section 35, and the nature of the document will also feed into the public interest test. The need for a safe space in relation to such documents will carry greater weight in the public interest balance and forms part of our conclusions above. A decision by the tribunal that the public interest favoured disclosure in a particular case would not alter that. There always was and remains a ‘threat’ or a ‘risk’ of publication against the wishes of the AO under FOIA. A decision to disclose in an individual appeal would not increase that risk.

Any future effects said to flow from the fact that “colleagues knew the Assessment would be published” or the AO being “aware that her assessment would be published” do not flow from our decision. A decision to disclose in this appeal does not mean that other AO assessments “would be published”, or indeed that other AO assessments are more likely to be published.

Effects that flow from an “expectation that [AO assessments] would readily be made public” or “the routine publication of full AO advice” do not flow from the outcome of an appeal where the tribunal has carefully balanced the public interests, taken account of the importance of a safe space, but has ultimately concluded that the public interest favours disclosure. It is not the case that AO assessments will either now or in the future “readily be made public” or be subject to “routine publication”.

Mr. Thomas is no doubt right that AO assessments would be drafted differently “if the material was primarily drafted or expected to be released”. Our ruling does not have the effect that AOs should expect their assessments to be released under FOIA. The material does not need to be drafted to be released. It is a matter of fact that in some circumstances the public interest might favour disclosure of AO assessments, because section 35 is not an absolute exemption, but that has always been the case and is not altered by our decision.

Mr. Thomas was specifically asked why our decision on where the public interest lay would have a chilling effect, when awareness of the existence of the risk of disclosure under the public interest balancing test did not. His answer is set out in full above. In essence, his response was that it was understood that documents that contained full and frank discussions and were bound up with fundamental policy decisions tended to have more protection than more anodyne government documents. Whilst that assumption is correct, the answer does not, in our view, address the weakness in the chilling effect argument identified by Charles J. Nowhere in Mr. Thomas’ evidence is this weakness addressed in a properly reasoned way.

Further, there is already a known risk of publication against the AO Officers wishes in the NAO report. Mr. Thomas was asked if he was surprised that the NAO report contained verbatim extracts from the Assessment, including in particular the official’s record of the minister’s selection rationale contained in appendix B of the Assessment, which was initially withheld in response to the request under section 36(2)(b)(i) and section 35(1)(a), and is included wholesale (with minor differences in wording) as Figures 6 and 8 in the NAO report.

His response in open was:

When asked who would have the final say as to what extracts would be included in the NAO report, if the Accounting Officer was not happy with what had been included, he replied in open:

Whilst we accept Mr. Thomas’ evidence that there will usually be a discussion between the AO and the NAO, it is clear that the AO is aware that it is the Comptroller and Auditor General who has the ultimate say about what is published, which could include, as it did in this case, large verbatim extracts from the report which, at least in the Department’s initial view in this case, should be withheld under sections 35(1)(a) or 36(2)(b)(i).

Mr. Thomas was then asked by the Judge whether the fact that it might go in the NAO report might have any impact on the way that people draft the AO assessment. His reply in open was in accordance with what might be expected from courageous, robust and independent senior civil servants, i.e. that they would make sure that they flagged any concerns which they thought the NAO might be interested in:

In our view, the fact there appears to be no generalised chilling effect arising from the risk that verbatim extracts from the AO assessment could be published in the NAO report, even against the wishes of the Department, supports our view that no generalised chilling effect is likely to arise from our decision in this appeal.

In conclusion, taking all the above into account, we respectfully disagree with Mr. Thomas’ opinion that our decision would be likely to have a generalised chilling effect on Accounting Officers. In our view if there is any generalised chilling effect it comes from the passing of FOIA and we rely on the courage and independence of senior civil servants to be robust in the face of the extant risk of publicity to which our decision adds nothing. We do not accept, as a matter of fact, that there will be indirect and wider consequences as a result of any decision by this tribunal to disclose the withheld information.

Although we were not specifically asked to take account of the qualified persons’ opinions under section 35, we find that they are relevant to the public interest balancing exercise under section 35, because the QPs have reached an opinion that disclosure would inhibit the free and frank exchange of advice and the free and frank exchange of views for the purposes of deliberation, which falls within the ambit of the interests protected by section 35.

Whilst we take account of the expertise and experience of both the QP’s, we are not persuaded that there is anything in the opinions that addresses the issues we have identified above in relation to assertions of a generalised chilling effect. There is little of substance in the first QP opinion that adds to Mr. Thomas’ considered and detailed evidence that we have discussed above.

In relation to the second QP opinion we note that the prejudice is said to flow “if the full advice were to be put in the public domain as standard”. As explained above, there is nothing in our decision that could lead any properly informed person to conclude that the full advice of the AO will be put into the public domain “as standard”. It remains the position that every FOIA request for an AO assessment must be subject to a careful balancing of the public interest, and that there is likely to be significant weight in maintaining the safe space as we have concluded above.

Mr. Anderson submitted that we must avoid the ‘extreme position’ of approaching prejudice and risk in a way that is ‘self-defeating’ by accepting the argument that the fact that officials know that FOIA exists and are already aware of the possibility of disclosure diminishes the weight of a safe space argument. Parliament created the section 35 exemption with the intention of ensuring that a safe space was preserved.

We do not accept that we have adopted an extreme position, nor that it, in effect, this approach defeats the purpose of section 35.

First, it may be that evidence in a particular case effectively addresses the weakness identified by Charles J. In those circumstances a generalised chilling effect argument about wider consequences on future discussions is likely to succeed. Whether or not a tribunal accepts arguments of a generalised chilling effect is a matter of fact to be determined in the circumstances of every case.

Second, as the ICO guidance on section 35 acknowledges at paragraph 206 arguments about a chilling effect on ongoing policy discussions are likely to carry significant weight and arguments about the effect on closely related live policies may also carry weight.

Third, whilst we do not accept that there is a risk of a generalised future chilling effect in this case, that does not mean that the need for a safe space is not weighed in the public interest balance. It is in the public interest that civil servants and officials involved in policy making should have a safe-space in which to do so, albeit that the need for a safe space may be diminished or superseded by the finalisation and publication of a policy. (Department of Health and Social Care v Information Commissioner [2020] UKUT 299 (AAC)). In this case we have concluded above that there remained a strong public interest in maintaining that safe space at the relevant date.

Overall our conclusions are that there is a strong public interest in maintaining the exemption in this case.

In summary, we have concluded there is an extremely strong public interest in disclosure of information relating to the selection of the towns announced on 5 September 2019, but for the reasons set out below we have concluded that this public interest is reduced on the facts to some extent by a number of factors, and that there remained a very strong public interest in disclosure of the withheld information at the relevant date.

Looking first at the factors which have led us to conclude that there is an extremely strong public interest in disclosure, we have taken account of the following.

First, there is, overall, a large amount of public money at stake. There is a large amount of public money potentially available to individual towns to spend for the benefit of particular, not insignificant, sections of the public. Towns not placed on the list are excluded from bidding for a portion of that money, and their inhabitants excluded from benefitting from it. There is therefore very significant public interest in increasing public understanding and accountability in relation to how, overall, that money is spent and in relation to how the selection decision was reached.

Second, whilst we do not find that there was a reasonable suspicion of wrongdoing, we take account of the following matters, which are set out in the internal review request of George Grylls, a journalist at the Times:

Further we take account of the fact that Ministers chose not to follow officials’ suggested regional allocation.

These matters, in the tribunal’s view, very significantly increase the public interest in transparency and accountability. There is an extremely strong public interest in publishing the full analysis by the Accounting Officer of the departure by the Ministers from officials’ recommendations. Disclosure would have the related advantage of avoiding misplaced speculation as to what might have been withheld. There is a strong public interest in informed reporting on this issue in the media.

Third, the levelling up agenda is a subject of importance. There is significant public and political interest in equality of opportunity across the UK. The fact that the Ministers went against officials’ recommendations as to the geographical spread of eligible towns is directly relevant to the levelling up agenda.

The extremely strong public interest in accountability and transparency is reduced to some extent by the accountability mechanisms already in place, including the NAO report and the PAC report, and the fact that the Department had just published a summary of the report. The NAO report and the PAC report are thorough and detailed. The summary lacks much of the detail contained in the full report. In the tribunal’s view, for the reasons set out above, much of the public interest lies in the publication of the full picture, and therefore we take the view that the public interest in publication of the full report remains very strong.

Without attempting to crudely summarise or repeat our detailed discussions and reasoning set out above, our conclusions are that although there is a strong public interest in maintaining the exemption, it is outweighed on these particular facts by the very strong public interest in disclosure. For those reasons the Department was not entitled to rely on section 35 to withhold the requested information.

As we have determined that section 35 is engaged in relation to all the withheld information, section 36 cannot apply. The information is ‘exempt information’ under section 35, even though we have determined that the public interest favours disclosure. For those reasons we find that the Department was not entitled to withhold the information under section 36.

Although we have not considered section 36, given that the parties submissions and the evidence in relation to the public interest balance were materially identical to those under section 35, we would have concluded that the public interest favoured disclosing the information for the reasons set out above.

As we noted above, the NAO report contains extracts from the Assessment, including the officials’ record of the minister’s selection rationale contained in appendix B of the Assessment, which was initially withheld in response to the request under section 36(2)(b)(i) and section 35(1)(a), and is included wholesale (with minor differences in wording) as Figures 6 and 8 in the NAO report.

Mr Thomas was asked by one of the members of the tribunal why the Department did not inform the requestors that some of the information in the Assessment was already in the public domain through the NAO Report. Mr Thomas did not know the answer to this as he was not involved at that stage.

Given the substantive extracts from the Assessment contained in the NAO report, the tribunal’s view is that the Department could and should have referred the requestors to the NAO report in its initial response.

The evidence of Mr. Thomas was that some of the options discussed in the Assessment were not proceeded with, in particular in paragraph 15. Mr Anderson submitted that this shows that the Assessment did not simply reflect the end of the process, but was still a live part of the process, which goes into the public interest balancing test. In so far as the Assessment was dealing with matters that were ultimately discarded, the public interest in those matters is reduced and conversely the interest in protecting policy formulation is greater.

The substantive content of the information helps to illustrate the points made about the way in which advice is couched, for example the way in which paragraph 3 is formulated. This is not the sort of formulation you would expect if it were intended for a wider audience. It is intended to put those reading the document ‘on alert’.

The tribunal should carefully consider the extent to which there is any real volume added to the information that has already been made public, by, for example, the greater detail provided in paragraphs 14, 15 and 17.

The Department has undertaken a careful balancing exercise and it is accepted that the bulk of the document can be disclosed, which diminishes the public interest in requiring the rest of it to be published.

In terms of the liveness of the policy, Mr Anderson submits that there were a number of strands to the Towns fund. Initially it was intended that there were going to be 3 strands. It was on 6 September 2019 that the towns were invited to develop their proposals. It was not until July 2021 that a full list of the funding allocated was published. There was intended to be a third competitive strand of funding but that element of the policy was changed in November 2021.

This is relevant because the detailed information in relation to the decision-making around what money should be allocated in the earlier stream and the possibilities of allocation that were rejected is the sort of thing that may have been relevant or might have been relied upon or put forward in relation to the further stages of the funding. That means that there was still a live element at the time of the decisions to refuse the requests.

In relation to the ‘liveness’ in January 2021, Mr Jackson submits that the policy under consideration is not the entire process of the Towns fund over the course of three years. The policy that is the focus of the Assessment is the ministerial selection process that was undertaken in the summer of 2019 and announced on 6 September 2019. It is possible for information to relate to multiple different stages of policy or multiple different policies, but it is necessary to consider the strength of the connection.

The Commissioner submits that by January 2021 the liveness of the policy had fallen away. The decision had been made. Even if there were further developments in July and November 2021, this information does not relate to those policies.

The Assessment is dated 5 September, the day before the public announcement was made. There may have been earlier drafts or iterations but the request relates to this document. The tribunal can infer that there was no formulation or development in progress at this point.

The specific sections of the Assessment that are contested are paragraphs 3, 14, 15 and 17. To the extent that these record discussions and deliberations that had been happening over the previous months but had now reached a settled position, this no longer relates to the formulation or development of policy. Paragraph 15 deals with implementation of policy rather than formulation. These are referred to as ‘specific delivery questions’. The policy - ministerial selection of towns eligible for this sum of money - had concluded by this point.

In relation to section 36, the first opinion falls far short of the requirements for a reasonable opinion in Davies. It relies on entirely generic concerns about the chilling effect and free and frank advice.

The second opinion reiterates the same general concerns, and does not meet the requirement to be tailored to the specific information in question.

It is not right to infer that the Assessment did not relate to the formulation or development of policy from the fact that the Assessment was the day before the announcement. It does not become an adopted policy of Government until it is signed off, until then it is still at the stage of formulation and development.

Where the Assessment refers to further choice in relation to matters of delivery, this does not mean that it is not part of policy.

Paragraphs 12 and 16 of the second opinion do refer to the particular context of the fund. The submission has to be read in the context of the background and the information with which it is concerned which is annexed to the submission. It is plainly sufficient for the purposes of the section 36 gateway.

In relation to the liveness of the policy to which the Assessment relates, we accept, for the purposes of the engagement of section 35, that the fact that paragraphs 14 and 15 contain aspects which were not taken forward shows that the Assessment did not simply reflect the end of the process, but was still to some extent part of a live process at the date at which it was written. This supports our conclusions in relation to the engagement of section 35.

In relation to liveness at the relevant date for the purposes of the public interest balance, the first policy aspect which was not concluded at the date of the announcement on 6 September 2019 is found in paragraph 14 of the AO and relates to a proposed cluster of two towns:

In the original announcement on 6 September 2019 Keighley and Shipley were announced as a cluster, but the NAO report records that this was not taken forward. We find that this was therefore no longer a live policy at the relevant date, because it had been abandoned by the time of the earlier NAO report.

The second aspect is in paragraph 15:

Unfortunately the evidence before the tribunal on when this proposal was abandoned is unclear. The Judge asked Mr Thomas when the decision was made, but he did not know when that proposal was abandoned, and we were unable to find evidence of the date in the bundle. When giving evidence on why it was not in the public interest for paragraphs 15 and 16 to be disclosed, Mr. Thomas referred to the AO guidance on redacting ‘policies not taken forward’ from the summary.

Mr. Anderson’s submissions on liveness were fairly broad, and his only reference to this specific point was that ‘possibilities of allocation that were rejected may have been put forward or relied on in relation to the further stages of funding’.

Doing our best with the limited evidence before us, and assessing the matter on the balance of probabilities, given the references by Mr Thomas and Mr. Anderson to policies that had being ‘rejected’ or ‘not taken forward’ we find that the proposal in paragraph 15 had been rejected by the relevant date. Our findings as to the liveness of the process as a whole are set out in our open decision.

Mr. Thomas highlighted the use of the phrase ‘The main area of concern is around propriety’ in paragraph 3. He stated that it was written like that to ‘flash up in red lights’ that this is the issue the AO has to think particularly carefully about. He stated that ‘we wouldn’t phrase it in that way’. His concern was that this would, if released, immediately become a newspaper headline. He stated that the NAO might choose to say that, and the Department would not question that, because that is the sort of thing that the NAO are entitled to say.

Given that one of the four established AO standards or criteria is propriety, and that a section on propriety is included in the summary of the AO report it was already a matter of public knowledge at the relevant time that propriety was one of the matters under consideration.

In relation to any generalised chilling effect, we have made our findings on this in open. We do not think that there is any other harm, and there is significant public interest, in the public knowing that this was the AO’s main area of concern, even if this were to appear in a newspaper headline. The AO reports sets out in detail the context and considerations in relation to propriety and why the AO reached the conclusion that the Ministers’ approach was appropriate.

Mr Anderson submitted that it was not in the public interest to publish policy that had not been taken forward. We accept that this carries some weight in the public interest balance, but there was no evidence or submissions identifying the specific harm that the release of information on these specific proposals could have caused to the ongoing process, and we find that the additional weight is not sufficient to outweigh the very strong public interest in disclosure of the full report for the reasons set out in open.

Further we find that there is a public interest in knowing why Keighley and Shipley were originally announced as a cluster, and in understanding the options that were considered in relation to potentially asking towns to work together. These matters give a clearer picture to the public and to the towns concerned of how carefully this was considered by the Secretary of State and will contribute to a fuller understanding of how decisions were reached in relation to the spending of this significant sum of public money.

In addition to the particular points considered above, we have carefully considered the specific content of all the withheld sections of the AO Assessment, looked at in the context of the nature of AO assessments, and in the light of all the evidence before us. The passages consist of further detailed explanation of the process adopted and we find that their disclosure will serve the public interests in disclosure identified in open.

On the evidence before us, and taking into account the nature of the information and the way in which it is expressed, we find that there is no specific additional harm that is likely to flow from the publication of the contents of these particular passages, other than the matters already identified in open.