The Envelope Clause: Britain’s Constitution by Correspondence

Parliament still meets. Ministers still answer questions. The Commons still divides, occasionally, and the lobbies still fill. But the machinery of consent in 2026 Britain often turns somewhere else: in an exchange of letters between a departmental official and a regulator, a Treasury minute and a funding assurance, a ministerial note and a quiet reply. The side letter is not a constitutional curiosity. It is a working part of the constitution, and it is almost entirely unpublished.

The habit is easy to miss because it looks like administration rather than power. A minister writes to the chair of an arm’s-length body to set out ‘expectations’. The Treasury offers a ‘letter of comfort’ to a stretched local authority or a national body. A devolution deal is accompanied by side letters on funding flexibility. An academy trust receives a funding agreement plus correspondence that interprets it. A regulator receives an assurance that its budget will be protected if it behaves. None of this is statute. All of it shapes behaviour.

The Letter as Instrument

The first thing to understand about the side letter is that it is not necessarily sinister. Governments have always needed to communicate intent without legislating every detail. A letter can clarify a funding settlement, record a shared understanding, or smooth an operational transition. It can be more flexible than a statutory instrument and faster than a bill. In a state that already governs through guidance, codes and framework documents, correspondence is simply one more layer of soft law.

But soft law with hard consequences is still a governing instrument. A letter of comfort from the Treasury may not create a legal right to money, yet recipients treat it as a signal of what will happen if they comply — and what will not happen if they do not. A ministerial letter to a regulator may not amount to a formal direction, yet it can set the tone for an entire regulatory year. A side letter attached to a devolution deal may not appear in the statutory text, yet it can determine how a mayor’s budget is actually spent. The letter is not law, but it is leverage.

This is where the constitutional question begins. Parliament authorises broad powers. Ministers fill in the detail. Normally that detail is at least published: regulations are laid, guidance is issued, directions are recorded. The side letter is the delegated legislation of the private office. It is not laid before Parliament, not subject to prayer, not routinely disclosed, and not easily challenged. It is the constitutional equivalent of the back staircase.

Who Holds the Pen?

If the side letter is an instrument of government, then the question of who writes it matters. In practice, the pen is held by ministers, special advisers and senior officials. The recipients are chairs, chief executives and finance directors of bodies that depend on government for funding, appointments or statutory goodwill. The exchange is usually polite, often technical, and frequently invisible.

Freedom of information law does not make these letters reliably public. Exemptions for policy formulation, commercial confidence and information given in confidence can be stretched across correspondence that is really about public money and public authority. Departments can release summaries, or disclose letters years later after the decision has hardened. Select committees can ask for them, but asking is not the same as receiving. The Public Accounts Committee can summon accounting officers, but it cannot audit a bargain it cannot see. The National Audit Office can follow the money, but it cannot easily follow the promise.

The result is a familiar asymmetry. The body that receives the letter knows what it says. The department that sent it knows what it says. The regulator or arm’s-length body may build its behaviour around it. Parliament, the public and most journalists know only that a letter exists because a written answer mentions ‘ongoing correspondence’ or because a freedom of information request is refused on grounds of confidentiality.

The Enforcement Question

A side letter is only as strong as the relationship it encodes. In most cases, it is not legally enforceable in the ordinary sense. A court will not usually compel the Treasury to honour a letter of comfort, nor will it force a minister to act consistently with an informal expectation. The sanction is political and financial: future funding, future appointments, future goodwill. That is not nothing. In a centralised state, it can be everything.

But enforcement by leverage is not accountability. If a minister promises a regulator that its budget will be protected, and then changes the funding formula, the regulator may have no remedy beyond a public complaint that few will hear. If a department assures a local authority that a grant will continue, and then does not, the authority may have to cut services while lawyers argue about whether a letter created a legitimate expectation. If a devolution side letter shapes spending, and the main deal is published without it, the public sees the architecture but not the wiring.

The deeper problem is not that letters are secret. It is that they are unreadable as a system. A citizen can find a statute. A reasonably determined researcher can find a statutory instrument. A side letter may exist in a departmental file, a recipient’s inbox, or nowhere at all. A promise that cannot be produced is a promise that cannot be tested. A constitution that relies on such promises is not transparent; it is merely well-mannered.

A Constitution of Carbon Copies

Britain has spent decades moving power from the chamber to the corridor. The side letter completes that journey. Parliament still legislates, but it increasingly legislates at a level of abstraction that leaves the operative decisions to correspondence. Ministers still answer, but they answer about the letter rather than with it. Regulators still regulate, but they do so in the shadow of assurances they cannot publish. The location of power has quietly shifted from the division lobby to the filing cabinet.

This is not a conspiracy. It is an administrative grammar. The side letter is useful precisely because it is informal, flexible and deniable. It allows government to coordinate without legislating, to reassure without committing, and to manage dependencies without creating rights. But useful habits become constitutional facts. When the hardest questions — funding, regulatory expectation, local government finance, devolution — are settled in correspondence, the public settlement is not the real settlement.

What Would Change

The answer is not to ban letters. Government would collapse into formalism, and the informal would simply move somewhere else. The answer is to treat side letters as public instruments when they do public work. A register of ministerial correspondence with regulators and arm’s-length bodies, published by default within a month, would not solve every problem, but it would make the system legible. Letters that alter funding, regulatory expectation or devolved spending should be published alongside the decision they support, with narrow and time-limited redactions. Parliament should be notified when a side letter is used to secure a commitment that would otherwise require a direction. The NAO should have a right of access to such correspondence.

None of this is radical. It is the minimum required for argument. A polity can disagree about a published promise. It cannot disagree about a promise it cannot see. The side letter is not the only place where Britain governs by private assurance, but it is the clearest. If the answer to ‘where is it written?’ is ‘in a letter you cannot read’, then the answer to ‘who governs?’ has already been given.

To unhive this habit is not to demand that every note between officials be published. It is to insist that the instruments which bind public bodies, allocate public money and shape public services are not treated as private correspondence. Britain’s constitution has long been unwritten. It should not become unreadable as well.