I. Introduction
The dramatic rise in executive lawmaking has become a widespread problem globally. This has been exacerbated by the changing practices and expectations of legislatures, often in response to political circumstances. In the UK, executive-made legislation has long since overtaken primary legislation as the dominant mechanism of lawmaking. However, the use of executive-made legislation (known as “delegated” legislation in the UK) has further accelerated in quantity and scope in recent years, particularly following the UK’s decision to leave the EU. Delegation of broad rule-making powers is enabled under the doctrine of parliamentary sovereignty, which recognises that the UK Parliament has unlimited legislative authority. As the powers granted to make delegated legislation have become so much broader in scope and frequency, greater clarity on their nature and regulation has become increasingly essential.
The debates regarding delegated legislation have tended to focus on the adequacy of parliamentary scrutiny and the scope of the powers being granted; these concerns go directly to the UK’s compliance with rule of law principles. However, they are also predicated on the understanding that delegated legislation is a clearly defined category of instruments and that all delegations of legislative power and delegated legislation can therefore be identified and subjected to such scrutiny and like processes in accordance with rule of law principles. This article shows that this understanding is incorrect by raising another pressing concern that has been substantially overlooked, namely the distinction between administrative powers and legislative powers. The problem of this distinction is particularly acute in cases where the enabling Act granting the executive powers is ambiguous as to their nature. Such ambiguity causes a difficulty in identifying whether some delegated powers are legislative or administrative in nature.
Indeed, a recent example, discussed in this article, saw the same type of instrument issued under the same powers treated by the Scottish and UK Governments as an administrative decision but by the Welsh Government as delegated legislation. This difference affects the type and level of parliamentary scrutiny accorded to the instrument: delegations of legislative powers and the instruments made under such powers are ordinarily subject to set processes of parliamentary scrutiny, whereas little to no parliamentary scrutiny might be applied to an instrument that is delegated legislation but not identified as such. This categorisation also informs how the instrument is reviewed by the courts and whether it has mandatory force of law. Uncertainty on such matters therefore raises considerable concerns regarding democratic oversight and constitutional legitimacy, and can undermine the application of rule of law principles to certain whole categories of instrument. This could, as will be shown, significantly affect individuals’ rights under the law, their position within governmental processes and their ability to challenge governmental decision-making.
This article will consider the adequacy of the existing tests for identifying delegated legislation and will highlight limitations in these tests relating to instruments made under enabling Acts that outline the powers in an ambiguous way. This article will, accordingly, propose a new two-stage test for identifying delegated legislation where the nature of those powers under the enabling Act is ambiguous. The long-standing source-based test should be expanded to become a “source-and-form” test as the primary or first-stage test. However, where this fails to identify the nature of the instrument because the powers granted in the enabling Act are ambiguous, this article proposes a new second-stage test: the “legislative-character” test. This article will then apply the new test to three types of instruments of an ambiguous nature, revealing that they should be understood as examples of what this article identifies as a new category of “innominate” delegated legislation.
This article advances our understanding of both the nature of delegated legislation and its role within the UK territorial constitution. Its findings have implications for future parliamentary scrutiny of and the courts’ approach to such innominate delegated legislation and for the wider public law principles that should underpin the granting and use of delegated legislative powers. This article specifically focuses on identifying delegated legislative powers arising under enabling Acts; issues relating to prerogative legislation can be distinguished and will be investigated by the authors elsewhere.
II. Existing Tests for Identifying Delegated Legislation
Definitions of primary legislation are long-standing. Dating back to the eighteenth century, the eminent English jurist, William Blackstone, discussed “the leges scriptae, the written laws of the kingdom, which are statutes, acts, or edicts, made by the king’s majesty by and with the advice and consent of the lords spiritual and temporal and commons in parliament assembled”.Footnote 1 Such primary legislation created by Parliament can be distinguished from delegated legislation. “Delegated”, “secondary” and “subordinate” legislation are interchangeable terms denoting a broad category of law. Typically, a piece of primary legislation – known as either the “parent Act” or “enabling Act” – will “delegate” lawmaking powers to another person or body, typically a government minister, on a particular topic; hence the term “delegated” legislation. The exercise of this power, because it is set out in statute, is limited by the terms of the statute, typically making the exercise of any such delegated legislative power highly circumscribed. With the arguable exception of devolved legislation and Acts of Parliament passed under the Parliament Acts 1911 and 1949,Footnote 2 it has long been acknowledged that delegated legislation should usually be defined and identified negatively by its dependency upon primary legislation. Because it is dependent upon, and limited by, primary legislation, delegated legislation is not accorded the same legal status as primary legislation and is therefore hierarchically lower than it (hence the terms “secondary” or “subordinate” legislation). With reference to the Interpretation Act 1978, Nick Barber and Alison Young acknowledge this as the long-standing “source-based test”: “any legislation that depends on an Act of Parliament for its legal validity constitutes delegated legislation.”Footnote 3
Barber and Young observe further that delegated legislation is reviewable by the courts, by which “the scope of the legislative power is determined by the empowering provision”.Footnote 4 Their assessment implies that delegated legislation is amenable to review by the courts on common law grounds, and consequently, unlike primary legislation passed by the UK Parliament, is capable of being struck down.Footnote 5 This is the case irrespective of any parliamentary approval of the delegated legislation.Footnote 6 However, Barber and Young recognise that reviewability is “neither a necessary nor a sufficient test. It is not necessary because courts may be debarred from considering the vires of a provision by a statutory ouster clause. It is not sufficient because some, rare, forms of primary legislation can have their vires turn on their subject-matter”.Footnote 7 Indeed, it is doubtful whether this additional reviewability test (if it can be framed as such) is a competent mechanism for identifying delegated legislation.
However, even the source-based test has been challenged previously by Aileen McHarg as “clearly inadequate, or at least incomplete”.Footnote 8 Her concern arises because it implies that the Acts of the devolved parliaments – which were themselves created by statute and remain dependent on statute – are therefore delegated legislation rather than primary, a conclusion with which she disagrees. As she has observed, with reference to the Parliament Act 1911, the House of Lords held that “the political intention behind [it] is indeed what matters most in determining the nature of the legislative power it created”.Footnote 9 McHarg instead proposes an alternative test for identifying delegated legislation derived from the assumptions that underpin the general legal controls normally applying to delegated legislation both pre- and post-enactment. This includes parliamentary scrutiny and the amenability of delegated legislation to judicial review by the courts.Footnote 10 According to McHarg, there are two overlapping justifications behind the more restrictive legal controls being applied to delegated legislation. The first concerns the separation of functions: because the power to make delegated legislation is given to the executive branch, the process of making such legislation is perceived as being both less legitimate than and at greater risk of abuse than the making of primary legislation. The second is that the executive is given the power to make delegated legislation for a specific purpose only, and is thus acting on behalf of the primary legislator in exercising that power; the use of delegated legislation should therefore be limited to those specific purposes.Footnote 11 As a result, McHarg concludes that “[f]rom this analysis of the legal controls which normally apply to delegated legislation we may therefore conclude that it is a form of executive power and one which must be used to further specific purposes laid down in the instrument of delegation”.Footnote 12
Of the two justifications, McHarg argues that the existence of a specific purpose is the most determinative test of the existence of delegated legislation. As she notes, concerns over the relationship between the executive and the courts “is applicable to all forms of executive rule-making, whether or not they derive from specific grants of legislative power. By contrast, it is only the existence of specific legislative goals that tells us what is distinctive about delegated legislation”.Footnote 13 However, the specific purposes test proposed by McHarg is just as limited as the source-based test. As she focuses on the rationale behind the legal controls on delegated legislation, the test becomes difficult to apply and would result in a very significant reliance on comparison with Hansard. Additionally, the limitation on powers to their awarded purpose, as she acknowledges, applies to any exercise of executive power, so it is difficult to understand how this makes legislative powers distinct from any other executive power. Part of that quandary is resolved by the context of McHarg’s focus: her hypothesis is designed to distinguish Acts of the Scottish Parliament from delegated legislation only, so applies to a very specific context. Her analysis pre-dated judicial determination of the nature of Acts of the devolved parliaments and subsequent case law on this point also raises concerns with this distinction.Footnote 14
III. A New First-Stage Test: Source and Form
The previous section has shown that none of the above tests for identifying delegated legislation are satisfactory. The source-based test – the dependency on an enabling Act – clearly must be considered within any test to identify delegated legislation. To this, however, needs to be added consideration of form, which equally will in most cases be specified by the enabling Act, but which is not mentioned by Barber and Young.
Delegated legislation can take many forms and this will often be specified by the enabling Act.Footnote 15 Certain forms will routinely be recognised as delegated legislation. Section 21(1) of the Interpretation Act 1978 states, for example, that “‘subordinate legislation’ means Orders in Council, orders, rules, regulations, schemes, warrants, byelaws and other instruments made or to be made under any Act” and was updated post-Brexit to include the additional categories of legislation “made or to be made on or after IP completion day under any assimilated direct legislation other than assimilated direct CAP legislation as so defined or made or to be made on or after exit day under assimilated direct CAP legislation as defined in section 2 of the Direct Payments to Farmers (Legislative Continuity) Act 2020”. Form is thus highly indicative of whether an instrument is delegated legislation.
It is therefore proposed that the source-based test should be expanded and have increased specificity as a two-pronged “source-and-form” test: the enabling Act as the source of the power (the first prong) will normally indicate whether the power is legislative by identifying the form (the second prong). Source necessarily comes before form: if the instrument is not made under an enabling Act then it cannot be considered delegated legislation. However, in isolation, even the expanded source-and-form test is incomplete because the second prong – form – can fail.
First, the 1978 Act does not provide a closed definition, as evidenced by the inclusion of “and other instruments made or to be made under any Act”.Footnote 16 Source and form alone therefore cannot and should not be the sole determining factors in identifying whether an instrument is legislative in nature. Delegated legislation is a broad and imprecise category of law, which would logically need to be written but which otherwise could take on forms not yet fully acknowledged.
Second, the source-and-form test cannot be successfully applied where the primary Act, although delegating a power, is silent or ambiguous on the nature of that power and the instrument made under it. Statutes will delegate a wide range of administrative powers and functions as well as legislative powers. The distinction between legislative and administrative powers might have a significant impact on the validity of an instrument issued under such ambiguous powers, and can pose challenges for parliamentary scrutiny and the future exercise of those powers.
The question then arises as to what alternative or additional criteria should be used to identify whether an instrument issued under delegated powers is legislative or administrative in nature. Or, in other words, because source and form are not indicative in all cases, an additional test is required to identify whether an instrument is delegated legislation.
A new second-stage test for identifying delegated legislation, which can apply in ambiguous cases, is therefore much needed. It is submitted that, in such circumstances where the primary Act is silent or ambiguous, the nature of the powers delegated should be determined with reference to their effect – and specifically whether the instrument that would be issued changes legislation.
IV. A New Second-Stage Test: Legislative Character
The difficulty in identifying delegated legislation goes back to at least the Donoughmore Committee in the early 1930s, which acknowledged that the terms relevant to delegated legislation (regulations, rules, orders) were being applied to instruments that were not delegated legislation in terms of their nature or effect.Footnote 17 This requires us to go back to consider the primary function of delegated legislation. The Donoughmore Committee, when reviewing ministerial powers and working to distinguish delegated legislation, highlighted that “[i]t is indeed difficult in theory and impossible in practice to draw a precise dividing line between the legislative on the one hand and the purely administrative on the other”.Footnote 18 The Donoughmore Committee therefore indicated that one should consider whether an instrument had “a legislative character” to identify delegated legislation.Footnote 19 The Committee did not define what might be included within this term, but rather provided a list of standard executive decisions that should not be seen as having such a character, including commands, licensing powers, remission of penalties, inspections or the power to inquire.Footnote 20
In searching for a potential definition of “legislative character”, it is notable that such a test was formerly introduced in Australia, in what is now known as the Legislation Act 2003,Footnote 21 following a 1992 report by the Australian Administrative Review Council.Footnote 22 However, the definition adopted was cyclical: whether an instrument was legislative was solely defined by whether it had legislative character and was made using a power granted by Parliament. This limited and circular definition inevitably caused problems in practice.Footnote 23 The definition of a legislative instrument was initially amended in 2005,Footnote 24 and then again by the Acts and Instruments (Framework Reform) Act 2015. The 2015 Act renamed the 2003 Act, provided a new test for legislative instruments and introduced a new category of notifiable instruments.Footnote 25
Australia now defines a legislative instrument in the 2003 Act by reference to four criteria, namely where it is either: deemed to be such under the 2003 Act,Footnote 26 declared as such in the enabling Act,Footnote 27 registered as a legislative instrument,Footnote 28 or if it has been made with a power delegated by Parliament, making or altering the law that affects privileges, interests, obligations or rights.Footnote 29 This final criterion appears to be a non-explicit definition of legislative character, even though the phrase “legislative character” no longer forms an explicit part of the test. The Act also implies that legislative character is not required or determinative for legislative instruments identified using the first three criteria,Footnote 30 and limits the inferences that can be drawn from other powers granted in the enabling Act or other primary legislation as to the legislative character of an instrument identified under the fourth criterion.Footnote 31 Moreover, the Act makes clear that even where the implied legislative-character test is met, an Act can nevertheless declare it not to be a legislative instrument; such an instrument can still become a legislative instrument if registered, subject to exceptions.Footnote 32 This further erodes the residual category based on legislative character.
The 2003 Act as amended also introduced the category of notifiable instruments: a mutually exclusive category with legislative instruments.Footnote 33 Notifiable instruments require registration but are not subject to parliamentary scrutiny. Although the Act is silent on this point, the inference is that this category should include administrative matters rather than instruments with a legislative character. A third category of instruments are those non-legislative, non-notifiable instruments;Footnote 34 these are subject neither to parliamentary scrutiny nor to a registration requirement unless specified in an Act. Whether such instruments in practice have a legislative character has again been deemed irrelevant to their categorisation and interpretation.Footnote 35
All this shows that legislative character arguably has a less central role in the identification of Australian legislative instruments than when originally envisaged and it persists now only in a reduced role, underpinning the understanding of the new residual category.Footnote 36 Additionally, the difference in operation of the two systems has important differences for certain types of instruments in practice. For example, a Commencement Order would be recognised as delegated legislation in the UK, but a Commencement Instrument is merely a notifiable instrument in Australia.Footnote 37
While a fuller comparison of the two jurisdictions would be valuable, these initial findings suggest that the Australian model cannot easily be adopted in the UK without wider consequential impacts. What is required is a broader, more inclusive and jurisdictionally appropriate understanding of “legislative character” that better captures the nature, use and effect of delegated legislation in the UK. It is submitted that some assistance in identifying what might be considered as a “legislative character” is provided in the work of Sir Cecil T. Carr, “one of the founders of administrative law in England”, who was a prominent thinker on delegated legislation in the early twentieth century.Footnote 38 He said that “[d]elegated legislation is allowed to affect the statute book in three different ways […] They are, firstly, direct amendment of an Act; secondly, the creation of some additional machinery affecting the commencement, duration or application of an Act; and, thirdly, the elaboration of detail […] to facilitate the objects or working of an Act”.Footnote 39
It is submitted that this legislative character – the effect of changing legislation in these three ways: amendments or modifications to legislation, changes in the application of legislation and changes to legislative frameworks – can be given normative value: it should be the second-stage test for identifying whether an instrument is delegated legislation. For an instrument to have the effect of changing legislation validly, it must be legislative in nature. By extension, if a power granted to the executive has that legislative character, then it must be a legislative power – even if not identified as a legislative power in the enabling Act.
Building on this conclusion, Carr’s three categories of changes to previous legislation can be used as criteria for identifying whether instruments issued under delegated powers are legislative or administrative in nature. It is thus important to consider them closely. Of course, applying the below categories of legislative change to any particular instrument might be clear cut in some cases and there may be some overlap in other cases, but in such cases it remains clear that these types of change are indicative of such an instrument having a legislative character.
A. Category One: Changing the Text or Interpretation of Legislation
In the first category of change, Carr considered delegated legislation that amended, substituted or added to the wording of statutes.Footnote 40 The examples he gave are modest, but in the century since he wrote such revisions have become more pronounced and common, and have amended other delegated as well as primary legislation and – particularly during the Brexit process – retained or assimilated EU law. He later built on his own work identifying different methods of textual amendment.Footnote 41
He considered “legislation without reference” or “amendment which is not direct amendment” to be “amend[ing] a previous statute by implication and without referring to it”.Footnote 42 The example he provided involves obscure provisions of Tudor statutes on horse-stealing,Footnote 43 but more recent and prevalent examples can be found. Following the Brexit process, sections 5(1) and 5(4) of the Retained EU Law (Revocation and Reform) Act 2023 have provided that any statutory language associated with retained EU law “is to be read, as regards all times after the end of 2023, as a reference to” the relevant term associated with assimilated law.
Carr contrasted such “legislation without reference” with “legislation by reference”, namely changes made to a specific named statute.Footnote 44 However, there were already alternative connotations of this phrase by the beginning of the twentieth century.Footnote 45 As time progressed, the term “legislation by reference” became increasingly opaque, as was highlighted by the Renton Review in 1975.Footnote 46 While the term has since fallen out of use and notwithstanding examples such as the 2023 Act, changing legislation with explicit reference to the relevant provisions is now standard practice.
Whether with or without reference to the provisions affected, there have broadly been two mechanisms of making such changes. In the language of the Office of the Parliamentary Council, these are direct textual “amendments” such as substitutions or deletions on the one hand and, on the other, “modifications” that are not direct edits of the text but which change the provision’s interpretation.Footnote 47 The Renton Review classified interpretative modifications as “non-textual amendments”, because the original text was not changed, in contrast to “textual amendments” where there was instruction to vary the text.Footnote 48 This separation of textual amendments and interpretative or non-textual modifications is a significant development from Carr’s earlier thinking, which elided the two. In a recent example, Duncan Berry, while the Consultant Parliamentary Counsel to the Irish Office of the Parliamentary Council to the Government, adopted this language in the Irish context as “the textual or direct method” and “the non-textual or indirect method […] [being] a discursive statement or narrative of the effect of the amendment on the old law”.Footnote 49
The Renton Review highlighted that an advantage of modifying the interpretation of the earlier statute rather than amending its text was that the changes were more easily understood by MPs; this was accordingly regarded as “the Westminster tradition”.Footnote 50 This advantage was recognised long before the Review. Indeed, Sir Courtenay Ilbert, writing in 1901 while Parliamentary Counsel to the Treasury, said: “The ordinary method of amending an Act is to state in the amending Bill the effects of the amendment proposed to be made. This is the commonest mode and for Parliamentary purposes the most convenient, because under it every Member of Parliament who knows anything of the subject, learns at once the nature of the amendment proposed.”Footnote 51
However, as the original legislative provision itself is not amended, this can result in potential confusion where there are numerous or complex modifications by reference. Hence, Ilbert recommended “that where it has been adopted, the proper course is to throw the law, as soon as practicable, into a simpler and more intelligible form by passing a measure of consolidation”.Footnote 52
In the end, the Renton Review recommended that direct textual amendments be used as far as possible.Footnote 53 More recently, Stephen Laws, while Head of the Office of Parliamentary Council, said this is “departed from only for good reason”.Footnote 54 However, although changes in technology mean that direct textual amendments are now more easily incorporated into the original text than they were a century ago, the difficulty in understanding the impact of the direct textual amendments to be made can remain. A very recent example of this difficulty was again seen in the Brexit process, where often long lists of particular phrases of retained EU law were revised via delegated legislation. In such cases, the actual change being made could not be understood easily without cross-referencing the delegated legislation with the original EU legislation.Footnote 55
However, interpretative modifications remain common, particularly where their effect will be temporary. Two important examples of this trend have been seen both in the Brexit process, as discussed above, as well as during the pandemic: the Coronavirus Act 2020 provided various changes to other legislation using interpretative modifications.Footnote 56 This mechanism is also used in delegated legislation: the Education Act 2002 (Transitional Provisions) (Wales) Regulations 2002 made various interpretative modifications to other legislation using the phrase “shall have effect as though”.Footnote 57
Whatever the relative benefits of these two mechanisms of change, it is notable that both have the same effect despite having different means for achieving that effect. The fundamental point is that such provisions change legislation, either by instructing the editor to make the change to the text or by providing an alternative interpretative framework to the reader.
B. Category Two: Changing Legislation’s Applicability
Carr’s second category of change made by delegated legislation included those instruments that change a statute’s applicability, such as those to commence, suspend or revoke it, as well as those that vary its scope or extent – whether geographically or to different industries or groups.Footnote 58 It is notable that he treats commencement orders as important and making a change to legislation because they have an impact on the statute book by providing primary legislation with legal effect. This stance has not been without detractors and some have instead regarded commencement orders as a purely administrative decision.Footnote 59
However, the increased importance of commencement orders since the time Carr was writing has proven him correct. There are various recent examples of the importance of commencement orders – and the potential controversy that commencement might cause. For example, in R. v Secretary of State for the Home Department, ex parte Fire Brigades Union, the minister’s decision not to commence a statutory scheme in favour of a different scheme was successfully judicially reviewed.Footnote 60 More recently again, the Sunak Government’s decision to hold a consultation prior to commencing buffer zones around abortion clinics after the passing of the Public Order Act 2023 drew criticism in Parliament.Footnote 61
Likewise, delegated legislation can revoke or repeal both primary and delegated legislation. For example, the Poole Harbour Revision Order 2012, made under section 14 of the Harbours Act 1964, revoked or amended various articles of one earlier Order and repealed three Acts of Parliament in their entirety, three Acts to the extent that they confirmed earlier orders and another Act but for four sections.Footnote 62 Likewise, section 43 of the Heritable Jurisdictions (Scotland) Act 1746, which provided for the levying of fines by sheriffs, was “omitted” or repealed by the Scotland Act 1998 (Consequential Modifications) (No.2) Order 1999, made under various powers granted by the Scotland Act 1998.Footnote 63 The granting of such powers would likely gain more attention and scrutiny now than powers to commence legislation, particularly as these specific powers would be categorised as so-called Henry VIII powers. However, Carr would view all the above examples as significant and as having a legislative character because all of these change the applicability of legislation.
C. Category Three: Supplementing Legislative Frameworks
Carr’s third category of changes includes the “delegation of power to make rules, regulations and orders which elaborate, supplement or help to work out some principle which Parliament has laid down”.Footnote 64 He terms this “supplementary legislation”,Footnote 65 although this is now better understood in connection with the phrase coined by Lord Herschell L.C. as “skeleton” primary legislation.Footnote 66 This category of change might overlap with the first category but can be distinguished: the first category of changes are made to the relevant Act itself; the third category are free-standing changes supplementing the relevant Act with additional legislative provisions.
The scope of what might be included in such a legislative framework has shifted significantly since Carr’s time. In 1927, M.L. Gwyer indicated, for example, that:
I am confident that no Department regards a mere power to make regulations for the purpose of carrying an Act into effect as conferring any power of making regulations creating criminal offences; that can only be done by an Act of Parliament itself. The Act may confer the power of making regulations and go on to provide that a breach of regulations so made shall be a punishable offence; but no regulations could themselves create offences or impose penalties.Footnote 67
However, today the creation of offences in delegated legislation even by negative procedure has become common.Footnote 68
In some ways, this effect is the most difficult to detect and distinguish from those instruments whose character is administrative. Administrative documents issued to interpret legislation for use by public officials or the public can be highly technical and authoritative and are thus sometimes referred to as quasi-legislation or soft law.Footnote 69 Such quasi-legislation includes various forms of guidance, codes of practice, circulars, measures and like documentation that can also be issued to elaborate on the law.Footnote 70 The crucial question for identifying whether the instrument has a legislative character is whether that elaboration changes the legislative framework or simply guides discretionary decision-making, interprets the law or outlines a public body’s approach in applying the law in a particular area.Footnote 71
V. Applying the New Two-Stage Test: Innominate Delegated Legislation
Building on the above analysis, a two-stage test can be proposed for identifying whether powers are legislative. First, the source-and-form test would be applied: normally the language of the enabling Act as the source and the specified written form will resolve the question of the nature of the instrument. However, where the enabling Act is ambiguous, a second-stage test can be used, namely whether the exercise of the powers has the effect of changing legislation. This second-stage test can be termed the “legislative-character” test. Ministers and other such individuals cannot change legislation through their decisions alone, so any power to do so must be a legislative power and the instrument of its exercise must be delegated legislation. Three categories of change that might be particularly instructive in this identification are, as outlined above: first, the amendment of legislation, whether by reference, by direct textual amendment or by varying the text’s interpretation; second, affecting the legislation’s applicability, whether its commencement or revocation, or changing its extent or scope; or, third, by changing the legislative framework by elaborating on or supplementing the legislation.
This new two-stage test encapsulates the common forms of delegated legislative instrument, such as statutory instruments and Orders in Council. However, unlike the source-and-form test operating in isolation, or the specific purposes test proposed by McHarg, this new test, which includes the effect of the instrument being legislative in character, also recognises other less common forms of instrument where the powers granted in the enabling Act are ambiguous.
This new second-stage legislative-character test for identifying delegated legislation is more expansive than previous tests. It will therefore capture some types of instruments that would not have been considered delegated legislation previously. These additional instruments can be said to be “innominate” delegated legislation, because they are not immediately identifiable as delegated legislation by the terms of the relevant enabling Acts. The examination below applies this new test to three types of instruments made under ambiguous powers, revealing them as innominate delegated legislation.
A. Parliamentary Resolutions
Parliamentary resolutions are the first type of instrument that in some instances might newly be identified as “innominate” delegated legislation under this new test. Carr identified that resolutions of Parliament could have legislative effect, in that they could amend the text of statutes where the enabling Act provided for this.Footnote 72 He implicitly recognised such resolutions as delegated legislation and expressed concern that they were not published in an appropriately accessible manner.Footnote 73
He gave two examples of such powers being granted under enabling Acts.Footnote 74 First, section 1(b) of the Coal Mines Act 1919 provided that:
if, after the end of the year [1920], a Resolution is passed by both Houses of Parliament that the economic position of the coal industry is such as to allow a further reduction of working hours, then, as from the thirteenth day of July [1921], the Act shall have effect as if in subsections (1) and (2) of section one thereof for “eight hours” there were substituted “six hours”, and as if in proviso (a) to subsection (7) of section one thereof for “nine hours and a half” there were substituted “seven hours”.
Second, section 2 of the Emergency Powers Act 1920 provided the Government with a very wide power to make regulations to respond to an emergency situation. It indicated that the regulations made under that section “may be added to, altered, or revoked by resolution of both Houses of Parliament” as well as by subsequent regulations. Resolutions passed under these two Acts would therefore be able to amend primary and secondary legislation respectively.
A more modern and regularly exercised example of parliamentary resolutions having legislative effect is provided in those resolutions related to the budget. Section 1(2) of the Provisional Collection of Taxes Act 1968 provides that a resolution that renews, varies or abolishes a tax and “contains a declaration that it is expedient in the public interest that the resolution should have statutory effect under the provisions of this Act” will have legislative effect for the period identified; section 5 has a similar effect for motions by the Chancellor of the Exchequer. Such resolutions are made each year to pass the Government’s budget: the budget is debated, after which the resolution or motion is presented and voted on and the tax rates set out in that resolution or motion are given statutory effect. This allows the budget to have immediate effect, with the relevant Act of Parliament to follow.Footnote 75
The phrasing of “statutory effect” is unclear, but it can be interpreted with reference to the similar provision found in section 50 of the Finance Act 1973, which provides resolutions with “statutory effect as if contained in an Act of Parliament” with respect to the variation or abolition of stamp duty. The wording of the 1968 and 1973 Acts suggests that the resolutions have legislative character and should be viewed as delegated legislation that can change other legislation as though they were primary legislation. However, the phrasing of “statutory effect” should not be viewed as making the resolutions primary legislation themselves. Applying the first-stage test – the source-and-form test – indicates that such resolutions are dependent on an enabling Act, but this first-stage test alone fails to identify their nature. This phrasing is ambiguous and it is submitted that it should be seen as indicative of the extent of the resolutions’ power, namely the potential for such resolutions to include Henry VIII powers to further delegate legislative powers that can change statute. An example of this is in the budget resolutions outlined in the next paragraph. Indeed, this phrase has even been used in delegated legislation when subdelegating legislative powers: the Medical Act 1983 (Amendment) Order 2002, which recognised its own article 16(4) as having “effect as if contained in an Act of Parliament” to delegate to the Privy Council certain Order-making powers under a saving provision.Footnote 76
Applying then the second-stage test – the legislative-character test – it is clear that the resolutions have a legislative character and should therefore be viewed as delegated legislation. The content of the budget resolutions made under the 1968 and 1973 Acts can include specific textual changes to previous legislation, thus falling under Carr’s first category of change and can specify varying commencement dates, thus falling under the second category. For example, the resolutions moved by the Chancellor on 6 March 2024 amended various prior statutes: they changed thresholds in the Income Tax Act 2007, provided for textual changes to the Income Tax (Earnings and Pensions) Act 2003 and substituted text in the Taxation of Chargeable Gains Act 1992, among other things.Footnote 77 Some resolutions for amending previous legislation might be set out in very broad terms or even have retrospective effect. Thus, the same 2024 budget resolutions included that “provision may be made amending Chapter 2 of Part 13 of the Income Tax Act 2007 to deal with cases where assets are transferred by closely-held companies” and “provision (including provision having retrospective effect) may be made which is incidental to, or consequential on, provision authorised by any other Resolution”.Footnote 78 As per the previous paragraph, they can also subdelegate legislative powers: for example, resolution 19 devolved powers to the Treasury to make regulations relating to co-ownership schemes.
While some of these resolutions outline what might be expressions of policy to be taken forward within the intended Act of Parliament, other legislative changes explicitly take effect almost immediately. For example, resolution 22 in the 2024 budget resolutions amended the Finance Act 2009 with effect from 6 March 2024. This example is particularly notable in that these amendments to statute took effect before the relevant Act of Parliament – the Finance (No. 2) Act 2024 – was introduced to the House. It was introduced on 13 March 2024, passed its third reading in the House of Commons on 23 May 2024 and was then rushed through the House of Lords on 24 May before the proroguing of Parliament that same day.Footnote 79 It is notable that these changes were not implemented on legislation.gov.uk prior to the 2024 Act being passed.
Such revisions to legislation by parliamentary resolution have been upheld by the courts. In Leeds City Council v Revenue and Customs Commissioners,Footnote 80 Revenue and Customs’s decision to reject the Council’s request for reimbursement of overpaid VAT was upheld. The reason for this was that the request was not made within a statutory limitation period that had been reduced by a parliamentary resolution made under the 1968 Act.
It is conceded that the Court of Appeal in Leeds City Council v Revenue and Customs Commissioners did not explicitly recognise the resolution as legislation. However, it could be said that, by confirming that the resolution had validly amended statute, the court implicitly acknowledged its legislative character. Put another way, parliamentary resolutions can have a legislative character particularly associated with the first category of change. In those cases where resolutions have that effect, it is submitted that they should be recognised as a form of delegated legislation, even if these are not forms that would normally constitute or be recognised as delegated legislation and which are not subject to scrutiny as delegated legislation.
Of course, contrary to this position, the UK Supreme Court in R. (on the Application of Miller) v Secretary of State for Exiting the European Union clearly stated that “[a] resolution of the House of Commons is not legislation”.Footnote 81 However, the resolution at issue recognised the outcome of the EU referendum and called on the Government to provide Parliament with a scrutiny opportunity prior to invoking Article 50 of the Treaty on European Union,Footnote 82 and was not – unlike those resolutions discussed above – made under an enabling Act that provided power to effect legislative change. It is submitted, therefore, that the UK Supreme Court’s statement should be viewed as a sweeping statement attempting to capture the ordinary nature of resolutions, as well as particular to the circumstances of the case, and not one that was sufficiently contemplated to present a barrier to the recognition of those exceptional resolutions with legislative character as being delegated legislation.
B. Approvals
Focusing on legislative character can also better identify a second type of instrument that has been understood differently by the UK’s different Governments and courts, namely approvals made under the Abortion Act 1967. That Act restricts the locations where abortions can lawfully be performed via sections 1(3) and 1(3A), the latter of which was introduced in the Human Fertilisation and Embryology Act 1990. Section 1(3) permits abortions to take place in various categories of hospitals and further authorises the relevant government minister to approve “a place” for this purpose. Section 1(3A) provides that “[t]he power under subsection (3) of this section to approve a place includes power, in relation to treatment consisting primarily in the use of such medicines as may be specified in the approval and carried out in such manner as may be so specified, to approve a class of places”.
The powers granted under section 1(3A) were exercised for the first time in October 2017. Making explicit reference to both section 1(3) and 1(3A), the Scottish Ministers issued an instrument (which they termed an Approval) to approve as a “class of places” the home of a pregnant woman undergoing abortion treatment insofar as it would allow her to self-administer misoprostol, the second pharmaceutical given in an early medical abortion, at home.Footnote 83 The Scottish Approval was the model for a near-verbatim instrument issued in Wales in June 2018,Footnote 84 and the UK Government issued its own Approval in December 2018.Footnote 85 During the pandemic, each Government issued a further Approval to allow telemedical abortion with both pharmaceuticals taken at home.Footnote 86 After the pandemic, telemedical abortion was retained in Scotland using that second Approval, in Wales initially through a third Approval, and was then placed on a statutory footing for England and Wales in the Health and Care Act 2022.Footnote 87
The language of an approval power is vague and has been used differently in different Acts. Sometimes the power to approve is specifically identified as the power to make delegated legislation. That might be, for example, to apply conditions to a particular piece of equipment under a legal framework,Footnote 88 or to approve additional services being provided by an authority,Footnote 89 or to approve a scheme.Footnote 90 Sometimes the power to approve has been a power to issue a certificate of approval, such as to license a particular device.Footnote 91 Understanding what type of approval power has been provided, and specifically whether that is a legislative power, is thus highly contextual to the circumstance of each enabling Act.
The power granted under section 1(3A) is inherently ambiguous: the Act is silent on its nature. The implications of the insertion of section 1(3A) into the Abortion Act were little noted in contemporary secondary literature when the 1990 Act was passed.Footnote 92 However, while section 1(3) had been interpreted as a licensing power, section 1(3A) is ambiguous in terms of what power is being provided to the minister. This can probably at least in part be attributed to the section’s origin as a backbench amendment to an otherwise unrelated Government bill and the possible modelling of its wording on that of section 1(3) by that MP despite the difference in scope. The proponent, Robert Key MP, did not speak to the amendment in debate,Footnote 93 and Parliament potentially created these powers without necessarily fully comprehending what it had created. This sort of situation is perhaps a further challenge to the argument advanced by McHarg: where the nature of the powers is unclear, it is possible – and perhaps even likely – that the specific purpose will likewise be opaque.
When the various Governments issued their Approvals in 2017 to 2018, they had to decide how to interpret those powers. The Scottish and UK Governments issued their Approvals as executive decisions. The Scottish Approval was subject to an unsuccessful judicial review action,Footnote 94 which has been analysed elsewhere.Footnote 95 The nature of the powers granted under section 1(3A) was not at issue in the case: both parties and the Lord Ordinary at first instance agreed that the Scottish Approval was “simply a decision and not delegated legislation, there being no regulation making powers conferred by the 1967 Act”.Footnote 96 The question of the nature of these instruments was not reopened on appeal to the Court of Session’s Inner House, or in the separate judicial review action raised against the UK Government’s pandemic Approval for England – although it is notable that the Court of Appeal referred to the instruments variously as “the Approvals” or “the Decisions”. Yet, in Wales, the Government and Senedd (Parliament) considered the Approvals to be delegated legislation, and they were published as delegated legislation and scrutinised as such by the Legislation, Justice and Constitution Committee.Footnote 97 There is therefore a critical disconnect in understanding between the three Governments as to the nature of the Approvals that they issued. Therefore, the first-stage test – the source-and-form test – fails to identify the nature of the instrument issued either as administrative or legislative.
The nature of a section 1(3A) approval therefore falls to be identified by the second-stage test, the legislative-character test. Abortions are an offence under the criminal law – by statute in England, Wales and Northern Ireland, and by common law in ScotlandFootnote 98 – unless performed in accordance with the requirements of the Abortion Act. By extending the lawful locations where an abortion can be performed, the minister is exempting those locations from the offence, or at least where the offence might be committed.
Indeed, it is submitted that the 1967 Act cannot be properly understood without separate reference to any additional approved class of place and one could only understand how to proceed legally (and without being liable to prosecution) if such reference was made. An approval thus, in effect, allows a change to be made to section 1(3) to add the approved class of place to the list of places where abortions can lawfully be performed, if only within the geographical extent of each approval.Footnote 99 This therefore changes the general understanding of the law, distinct from (for example) administrative decisions such as the provision of licenses to individual premises.
The approvals can therefore be said to be making the first category of change. It remains arguable whether approvals can be said to have made these changes by modifying the interpretation of section 1(3),Footnote 100 or should be seen as allowing the new “class of places” to be added to the existing categories of place listed in section 1(3) as direct textual amendment.
Comparison can usefully be made to Lord Herschell’s observation in Institute of Patent Agents v Lockwood, which included consideration of rules for maintaining registration as a patent agent laid down in delegated legislation:
But if you read into the section [of the Act] […] the rules made under sub-sect. 2, then of course every rule which is intra vires […] is to be read into the section, and have just the same effect as if it had been contained in the Act itself; and if so it is impossible to say that he can claim to be registered otherwise than in the manner which the statute, as filled up, if I may say so, by the rules provides.Footnote 101
Much the same can be said of the approvals: the current law of abortion, and the ability of patients and medical practitioners in the UK to proceed without potentially committing an offence, could not be understood without reference to the approvals. The approvals therefore have a legislative character, having the first of the three categories of legislative change. Accordingly, it was suggested elsewhere that the approvals should be considered as delegated legislation.Footnote 102
C. Immigration Rules
The power granted under the Immigration Act 1971 to issue immigration rules is inherently ambiguous. Indeed, there is no explicit power to make immigration rules of any type outlined in the Act. Rather, there is an implicit power in the requirement that statements of any changes to the immigration rules need to be laid before Parliament.Footnote 103 Thus, the first-stage test – the source-and-form test – fails to identify the nature of the immigration rules.
We therefore need to consider whether the immigration rules have a legislative character, per the second-stage test. Whether the immigration rules have a legislative character has been the subject of significant and prolonged debate as to whether they are delegated legislation or some unique category of policy statement with some legal effect, as will now be explored. However, as will be seen, without having the critical framework provided by the legislative-character test, this debate has lacked the necessary precision and focus to reach a satisfactory and widely accepted answer.
The UK Government has described the immigration rules as “some of the most important pieces of legislation that make up the UK’s immigration law”.Footnote 104 The explanatory memoranda that have accompanied statements of changes in the rules have referred to them variously as legislation or simply as law.Footnote 105 The statements have been referred to as legislation, and scrutinised as such, by the UK Parliament’s Secondary Legislation Scrutiny Committee.Footnote 106 However, the courts have extensively debated the nature of the immigration rules and have been reluctant (if not consistently so) to recognise the immigration rules as delegated legislation.
The Court of Appeal has considered the nature of the immigration rules on various occasions, with two prominent cases being heard in the 1970s. In R. v Chief Immigration Officer, Heathrow Airport ex parte Bibi, Lord Roskill commented, obiter, that the immigration rules are part of the law of the country and “just as much delegated legislation as any other form of rule-making activity or delegated legislation which is empowered by Act of Parliament”.Footnote 107 Part of his rationale appears to have been that the Immigration Act 1971 requires a statement of any changes to the immigration rules to be laid according to the negative laying procedure,Footnote 108 and it would be “unheard of” that such a procedure would apply if the rules were not delegated legislation.Footnote 109
However, Lord Roskill’s reasoning was rejected in R. v Secretary of State for the Home Department ex parte Hosenball.Footnote 110 Lord Denning identified them instead as “rules of practice”:Footnote 111 he noted that the courts could have reference to them under section 19(2) of the Act and in judicial review cases, “[b]ut they are not rules in the nature of delegated legislation so as to amount to strict rules of law”.Footnote 112 He gives the example of one of the rules replicating what the Secretary of State said to the House of Commons: “It adds nothing. It subtracts nothing. At any rate, nothing of substance.”Footnote 113 Lane L.J. noted in the same case that the immigration rules are “difficult to categorise or classify. They are in a class of their own”.Footnote 114 Rather, he saw them as “little more than explanatory notes of the Act itself”.Footnote 115 He distinguished the laying of statutory instruments from the laying of guidance, by implication disagreeing with Lord Roskill. Cumming-Bruce L.J. meanwhile highlighted that the rules do not generally have legal force, except when given that effect under section 19(2) in appeal cases before the adjudicator of the tribunal.Footnote 116 The conclusion of the court in R. v Secretary of State for Home Department, ex parte Hosenball has been explicitly followed in various cases since.Footnote 117
More recently, the UK Supreme Court considered the nature of the immigration rules for the first time in Odelola v Secretary of State for the Home Department.Footnote 118 As per the earlier Court of Appeal decisions, the Supreme Court highlighted that “[t]he status of the immigration rules is rather unusual. They are not subordinate legislation but detailed statements by a minister of the Crown as how the Crown proposes to exercise its executive power”.Footnote 119
Indeed, Lord Hoffman acknowledged that “they create legal rights […] [so] there is no conceptual reason why they should not create rights which subsequent rules should not, in the absence of express language, be construed as removing”.Footnote 120 However, Lord Brown identified the immigration rules as decisions made under the prerogative, rather than under the Immigration Act 1971.Footnote 121
The association of the immigration rules with the prerogative was not new,Footnote 122 but this was nonetheless rejected in R. (on the Application of Munir) v Secretary of State for the Home Department, which held that “the power to make immigration rules under the 1971 Act derives from the Act itself and is not an exercise of the prerogative”.Footnote 123 Indeed, although “there is no provision in the 1971 Act which in terms confers on the Secretary of State the power or imposes on her the duty to make immigration rules” there was nonetheless “implicit in the language of the Act that she is given such a power and made subject to such a duty under the statute”.Footnote 124 The Court determined that even though the explicit duty is to make a statement about the rules, this should be read broadly to include making rules themselves to avoid frustration of parliamentary intention. Overall, the Supreme Court concluded that its earlier decision in Odelola was probably incorrect on its historical understanding of the origin of the power to make the immigration rules, but in any case, commented that the remarks about the rules being secondary legislation were obiter.Footnote 125 While this comment provided scope for the Supreme Court to reconsider the nature of the immigration rules, when the court next did so, in the case of Ali v Secretary of State for the Home Department, it declined to do so. Lord Reed said: “The Rules are not law (although they are treated as if they were law for the purposes of section 86(3)(a) of the 2002 Act) […] but a statement of the Secretary of State’s administrative practice.”Footnote 126
The courts have therefore generally been of the view that the immigration rules are not legislation. Indeed, academic commentators have now concluded that the viewpoint that the immigration rules are not legislation “appears to have been widely accepted”.Footnote 127
However, the context in which cases are raised is an important consideration. Judicial challenge of the immigration rules has often been based on changes to them. There are strong policy reasons why Governments would want the immigration rules to be classified as policy instruments rather than delegated legislation, namely that any changes can then be applied immediately – even to cases which were competently filed while the previous version of the rules were in force. This policy rationale was explicitly considered by the court in R. v Immigration Appeals Tribunal ex parte Nathwani. As Brown J. remarked, “bearing in mind that the Rules are not statutes or statutory instruments which give rights to any person, there can be no question here of retrospectivity applying certainly to the time of the application as distinct from the time of the Secretary of State’s consideration of the application and his decision”.Footnote 128 Likewise, Eveleigh L.J. indicated: “The argument of Mr Nathan that in effect this is giving retrospective effect to the Rules, and then by analogy to the interpretation of statutes contending that that was not permissible, is, in my view, a mistaken approach. The Rules and their statutory interpretation depend very largely on vested rights. There were no such rights in the present case.”Footnote 129
Such commentary highlights the complicating factor that if the immigration rules were recognised as delegated legislation then the Government would have to apply the version in effect when any particular immigration application was filed. Interpreting the immigration rules as a sui generis category of policy document provides near-unfettered discretion to the Government to change the rules with immediate effect, irrespective of the potential expectations of any persons with applications already in the system.Footnote 130 This in turn restricts in-court oversight and reduces scrutiny, albeit individual application decisions remain subject to the basic principles of judicial review and the rules of natural justice.Footnote 131
One of the court’s other considerations, particularly emphasised in R. v Secretary of State for the Home Department ex parte Hosenball, has been that the nature of the rules themselves were more a type of guidance or policy statement than law. In that case, Cumming-Bruce L.J. commented that “having scrutinised them, and observed that curious amalgam of information and description of executive procedures, they are not in my view in any sense of themselves of legislative force”.Footnote 132 This point has been oft-cited in the case law since. Indeed, Satvinder Juss (who believed that the immigration rules were not legislation but should need to be issued as such) noted that this kind of informal language had been characteristic of the general instructions that preceded the 1971 Act. This, he said, has “influenced the approach which the courts have taken to [the immigration rules’] interpretation” and has been “used to imply a lack of direct legal effect”.Footnote 133
However, a fresh consideration of the difference between the earliest immigration rules and those issued in the modern era highlights the precarious nature of classifying the immigration rules as mere policy directions. The first version of the immigration rules was published in 1973 and would not easily be regarded as supplementing the law (or, to put another way, would not meet the threshold for Carr’s third category of changes with a legislative character).Footnote 134 Cumming-Bruce L.J. was undoubtedly correct in his conclusion that these had the look of guidance and policy. However, the immigration rules have changed very significantly since that time – in complexity, prescriptiveness, formality, number and length. Some of the provisions remain of the character of a policy statement or guidance, but many of them are highly prescriptive and indeed arguably modify the interpretation of legislation or supplement legislation.Footnote 135 The rationale underpinning the original court judgments in the 1970s cannot be easily applied to the immigration rules as they are now.
Yet, even where there is a high degree of flexibility or discretion provided by the immigration rules, this should not be seen as a bar to their being regarded as legislative in nature. Indeed, also broadly within the space of immigration law, the courts in R. (on the Application of Tigere) v Secretary of State for Business, Innovation and Skills and Jasim v Scottish Ministers held that delegated legislation limiting access to educational funding based on immigration status may fail the test of proportionality unless discretion and flexibility is provided to decision-makers in difficult cases.Footnote 136
This is not to say, of course, that the immigration rules should necessarily be recognised as delegated legislation. However, if the effect of the immigration rules is considered according to the legislative-character test proposed in this article, at least some of the provisions amend, modify or supplement the 1971 Act such that, returning to Lord Herschell’s comments in Institute of Patent Agents v Lockwood quoted above, it is not possible to understand immigration law without reference to them. At least some of the immigration rules could therefore be said to have a legislative character and so meet the second-test to be identified as delegated legislation.
A binary approach to classifying the immigration rules as either policy or legislation would therefore appear unhelpful. Indeed, Juss already observed in 1992 that there were grounds to conclude that “it may be misleading to refer collectively to the rules as having the same legal effect”.Footnote 137 Rather, potentially a more nuanced approach needs to be taken, where the nature and effect of each particular rule is considered rather than taking a blanket approach. Such a casuistic approach compares favourably with Carr’s example of parliamentary resolutions: not all resolutions are delegated legislation, but some of them should be recognised and treated as such. This is therefore potentially a power that always encapsulated administrative and legislative elements, but the legislative component was not recognised or utilised in its early history and the balance of the exercise of those two components has since changed.
VI. Conclusion
This article has proposed a new two-stage test for identifying delegated legislation. The first stage, the source-and-form test, builds on and draws together two long-standing considerations in identifying delegated legislation with reference to the enabling Act. Does the enabling Act as the source of the power (the first prong) specify a form that implies a legislative instrument (the second prong)? If there is no enabling Act, the instrument cannot be delegated legislation; if the Act is ambiguous on the nature of the powers regarding form then this article identifies a second-stage test to apply.
Drawing principally on the work of both the Donoughmore Committee and Cecil T. Carr, this article proposes that this second-stage test should consider whether the instrument made under the power(s) has a legislative character, defined as whether the instrument would change legislation. Three categories of changes are highlighted and given normative force by this article as identifiers of such a legislative character: amendments or modifications to legislation, changes in the application of legislation and changes to legislative frameworks.
This article then applies this legislative-character test to three types of instruments where the first-stage source-and-form test fails to reveal their nature because the powers under which they are made are ambiguous: parliamentary resolutions given statutory effect under enabling Acts; approvals made under section 1(3A) of the Abortion Act 1967; and the immigration rules made under the Immigration Act 1971. This article shows that such parliamentary resolutions, the approvals and some of the immigration rules should be classified as delegated legislation. In so doing, this article reveals that there are more types of delegated legislation than has previously been understood. Because the nature of this delegated legislation is not explicit in the relevant enabling Acts, this article classifies such instruments as “innominate” delegated legislation.
This new two-stage test, specifically the addition of the legislative-character test and the recognition of innominate delegated legislation advances our understanding of the nature and role of delegated legislation within the UK territorial constitution. Part of that better understanding is a greater recognition of the differences between instruments issued under delegated powers and whether these are legislative or administrative in nature. This article has also shown that significant changes to legislation are being made through innominate delegated legislation and that there is a greater reliance on delegated legislation in the UK than has previously been understood. As a result, the findings of this article have significant implications for the parliamentary scrutiny and judicial reviewability of such innominate delegated legislation, and thus in turn their democratic oversight and constitutional legitimacy. These questions will be explored by the authors elsewhere.
Open access
