What the Rule-Book Cannot See

1. Introduction

Among economists who study how money is actually made, one question is no longer open. Commercial banks create deposits when they lend; they do not lend out pre-existing reserves; the money supply is driven by the demand for credit rather than set as a stock from outside. The post-Keynesian endogenous-money literature argued this for decades — Kaldor against monetarism, Moore (1988), and Lavoie’s (2014) synthesis — and the operating side of the central banking world has now conceded it. Bindseil and König (2013) conclude their survey of operating frameworks with the observation that “we have all become Horizontalists in the last twenty-five years,” and the Bank of England’s 2014 Quarterly Bulletin article states the position in public: banks do not multiply up reserves to lend, and the money-multiplier model “does not describe how money is created in reality” (McLeay, Radia and Thomas 2014). At the operational level, in other words, the constitutive view of money — money as endogenous to the credit system — has won.

This is the conversation the paper joins, and it begins from a puzzle internal to it. If the operating framework is now constitutive, the codified rule-book is not. The Basel capital and liquidity hierarchy, the legal definition of what counts as a bank, the menu of scenarios supervisors stress-test, the statutory mandates that hold monetary policy at arm’s length from the fiscal authority — these were written in the 1990s, on the opposite assumption that money is a veil, a neutral quantity laid over a real economy and managed from outside. They have been amended since, but not rewritten. Crucially, the endogenous understanding and the veil-era rule-book do not sit in the same hands. The understanding lives in the monetary-policy function — the operating desk that must clear the market each day. The rule-book lives in the prudential and supervisory function, which in many jurisdictions is a separate authority, and which is in every case built largely from a supranational template — the Basel framework — that no single national body can rewrite at will; even the central bank’s own founding mandate is veil-leaning. Whether the two functions share an institution, as in the post-2013 Bank of England or the post-2014 euro area, or are formally divided, as in the twin-peaks systems and, in fragmented form, the United States, the knowledge of how money is made and the categories through which it is supervised are not held by one mind. The post-Keynesian literature has documented the first half of this gap — the operating side — exhaustively. It has not analysed the second: what the lag between an endogenous operating understanding and a veil-era rule-book does to what supervisors can actually see.

That is this paper’s question, and the answer it proposes is that codified categories shape the surveillance regime that follows them. A category-system is not a neutral filing scheme. It fixes what can be counted, aggregated, stress-tested, and placed on a committee’s agenda; a risk for which the rule-book has no category is, for practical purposes, not watched — not because anyone chooses to look away, but because the system has nowhere to record it. Codified doctrine therefore works like an immune system: after each crisis the rule-book encodes defences against the kind of risk that crisis exposed, and those defences are exquisitely specialised against the prior pathogen and, for that very reason, blind to risks of a different shape. The blindness is not a failure of vigilance bolted onto the defence; it is the same property as the defence, seen from the other side.

The boundary of the argument is worth stating plainly, because it is easy to mistake for timidity. The claim is about the consequences of codification, not its causes: I do not argue that doctrine decides which settlement is written into the rule-book after a crisis. Material interests and political coalitions do most of that work (Krippner 2011), and settlements then persist for the path-dependent reasons Pierson (2004) describes. But codification is not a neutral relay of those interests — it is how they are made durable, and the specific categories it lays down decide what the resulting regime can later see. Politics picks the rule-book; the rule-book then picks the blind spot, and that second step is this paper’s.

One distinction carries the argument and is best stated at once, since it looks like a contradiction. The categories are, in origin, the frozen victory of a past coalition — fossilised motives, if one likes — but origin and operation come apart over time: a category written to serve an interest becomes, a generation on, an inherited classification that officials apply because it is the rule-book they were handed, not to protect anyone. Conceding that interests write the code is therefore consistent with claiming that the code’s consequences are not read off the present balance of interests. The move places the paper in the sociology of ignorance, which has theorised motivated not-knowing richly but under-theorised the doctrinal not-knowing built into a category-system; separating the two, and supplying a test that tells them apart, is the analytical core.

The argument should interest monetary economists and financial-stability researchers because it does something a retrospective political-economy account cannot: it names, in advance and from the observable, public features of the current rule-book, where the next surveillance gap is most likely to sit. A regime that has changed its mind about money but not its categories will keep being surprised in the same direction, discovering risks only when they detonate; a framework that locates the gap from the rule-book’s own categories, before the loss, is worth more than one that can only explain the loss afterward. The contribution is theoretical — a framework, a typology, and a falsifiable test, evidenced across three centuries of the documentary record and applied to the present — and Section 6 specifies a single archival case, the subject of a companion paper, designed to push the central distinction harder than the historical illustrations can.

The paper proceeds as follows. Section 2 sets out the framework in five steps — the two views as a map of the field, the doctrinal ratchet that codifies a settlement, the epistemic immune system the codification becomes, the discriminating test that makes the mechanism falsifiable, and the dependent variable on which the test runs. Section 3 shows the mechanism in the historical record. Section 4 applies it to the present. Section 5 answers the most serious objections. Section 6 sets out a research programme. Section 7 concludes.

2. The framework: codified doctrine and the shape of institutional blindness

The framework has one load-bearing claim and a careful boundary. The claim is that the codified categories of a monetary settlement shape what the resulting surveillance regime can see. The boundary is that this is a claim about consequences, not causes: it does not require doctrine to defeat material interests, only that the legal codification of whatever settlement does win carries categories whose surveillance effects can be traced. The five steps below build the claim — the field, the ratchet, the immune system, the test, and the variable.

2.1 Two views of money

Monetary thought has organised itself, for two centuries, around two incompatible views.^[The recurrence is itself an old observation. “Economic ideas,” D.H. Henderson remarked, “move in circles: stand in one place long enough, and you will see discarded ideas come round again … And nowhere is this more true than in respect to monetary theory and the associated theory of monetary policy” (quoted in Cramp 1971, 62).] The veil view holds that money is a neutral overlay on real activity. Its modern form combines exogenous money — the quantity of money is a stock determined outside the private sector, by nature in a commodity regime or by the central bank in a fiat regime — with the classical dichotomy: nominal variables do not affect real variables in the long run. On this view the credit system is plumbing, useful but not the substance of allocation. The view is recognisable in the dynamic-stochastic-general-equilibrium tradition, which keeps the classical dichotomy in its long-run properties; in the public reasoning of the 1990s central-bank-independence statutes; and in a Basel baseline that gives sovereign debt a zero risk-weight and treats high-quality liquid assets as a technical category rather than a description of monetary function.

The constitutive view holds that money is endogenous to the credit system. Money is created in the act of credit extension, principally by banks expanding their balance sheets, and the credit system is not an appendage to the real economy but the form the real economy takes at scale. Its descriptive content is the post-Keynesian endogenous-money literature — Kaldor’s critique of monetarism, Moore’s (1988) Horizontalists and Verticalists, Lavoie’s (2014) synthesis, and the structuralist line of Chick (1992) and Dow (1996). The internal post-Keynesian debate concerns the slope of the credit-supply schedule and the role of liquidity preference; the fullest synthesis runs through the stock-flow-consistent modelling of Godley and Lavoie (2007) and Lavoie’s (2014) treatment, in which the horizontalist position is more developed than its structuralist critics sometimes allow (cf. Palley 2013). What unites the literature is that the central bank sets the price of reserves and accommodates the quantity. The difference between the two views is not academic: it changes what a crisis is. On the veil view a crisis is a friction in the plumbing; on the constitutive view it is the credit system failing to roll its own obligations — the economy itself seizing.

As Section 1 noted, the descriptive question is now settled in favour of endogeneity at the specialist and operating level (Bindseil and König 2013; McLeay, Radia and Thomas 2014). The dyad therefore enters this paper not as a contest over which view is correct but as a map of the field within which successive settlements have been codified. The 1819 Resumption, the 1844 Bank Charter Act, and the 1990s independence wave each carried into the rule-book a category-system that is recognisably veil-leaning, whether or not their architects held the veil view in its strong form; Bretton Woods carried one that is recognisably constitutive. The interest here is in the consequences of those codified categories, not in which view is true.

2.2 The doctrinal ratchet

Each major settlement is, after each crisis, encoded into the institutional rule-book. The encoding is more than legal text — statutes, charter acts, the legal definition of money — though it is that too. It is also operating practice, the architecture of the central bank’s balance sheet, the training of its staff, the supervisory category-system, the design of stress scenarios, the remit of internal committees, and the institution’s self-conception. I call this the doctrinal ratchet: once a position is locked into legal-institutional form it resists movement backward without an external release, each crisis is such a release, and between crises the accumulated weight of the codification holds the settlement in place. Pierson’s (2004) work on path dependence supplies the persistence mechanism — increasing returns, sunk investments, and the cost of coordinating on an alternative all make the inherited form sticky — and Goodhart’s (1988) history of central banks shows how slowly the supervisory function turns in practice. The ratchet adds only a name for the asymmetry: forward at crises, resistant between them.

The asymmetry has a mechanical source. The operating framework and the codified rule-book are disciplined by different forcing functions. The operating desk faces a daily reconciliation with reality — it must clear the market each morning — so it adapts continuously and, because the system is endogenous, drifts toward the constitutive description on its own. The codified rule-book faces no such daily test: a miscategorisation can sit in the supervisory data system or the Basel taxonomy for years, because nothing in routine operation forces a reconciliation, and only a crisis supplies one. That is why the rule-book moves in jumps while the operating framework moves continuously — not because anyone defends the inherited categories, but because nothing routinely tests them.

I concede the causal question to political economy. Krippner’s (2011) account traces the political origins of the rise of finance; Eichengreen’s (1992) Golden Fetters shows how an inherited mentalité, codified in institutions, made reflation unthinkable in the 1930s. Material interests do most of the work in determining which view is codified after a crisis. What the framework adds is downstream: the ratchet is not a causal mechanism that explains why one view wins, but a transmission mechanism that converts a political-economic settlement into a particular legal-institutional form — a particular set of categories, remits, and analytical scopes — and the claim is that the specific content of those categories then matters in a way the political-economy account does not by itself predict. When the 1844 Act fixed “money” as the note issue, it did not merely register the Currency School’s victory; it created a data system, a reporting cadence, and a committee remit organised around note issue. The same victory codified into different operational artefacts would have produced a different surveillance regime. That is the space the framework occupies.

2.3 The epistemic immune system

The codified categories shape what the institution can subsequently see, and here the argument joins an established literature it should not pretend to originate. That a classification system makes some things visible and others invisible is the central thesis of Bowker and Star’s (1999) Sorting Things Out: categories “enable and constrain” at once; they embed the politics of the moment that made them; and, once in place, they sink below notice and operate as infrastructure until something breaks. The method of this paper is what Bowker and Star call “infrastructural inversion”: it treats the categories through which a central bank or supervisor counts and monitors not as a transparent window onto risk but as a historical artefact that determines what risk can be seen.

Bowker and Star already theorise the temporal life of classifications, so the contribution here is not to add time to their account. It is to specify, for one domain, a particular temporal mechanism — the ratchet that ties each category-system to the doctrinal settlement of the crisis that produced it, and so links the moment of a category’s birth to the location of the surveillance gap that follows — and to add a typology of the resulting blindness (Section 2.4) and a falsification test, neither of which the classification literature supplies because neither is its question. The governing image is the epistemic immune system. After each crisis the institutional order develops defences against the risk the prior crisis exposed; those defences are coded into the category-system, the scenario taxonomy, the model coverage, and the committee agendas of the regime that follows. The immune response is hyper-specialised: it hunts the prior pathogen aggressively and is, for exactly that reason, blind to risks that do not resemble it. The specialisation that makes the defence effective is the property that makes it blind.

The mechanism resembles the folk maxim that generals fight the last war, but says something more specific and more testable: the next blind spot is not located at random but maps onto the categories the prior codification omitted, so its shape is partly predictable from the content of the last settlement. Three cases make the pattern concrete. The 1844 Act codified the Currency School’s category-system, in which “money” meant notes and deposits were derivative; the regime watched note issue closely and credit-system risk poorly, and the panics of 1847, 1857, and 1866 each materialised in the unwatched category. The Bretton Woods architecture codified categories built around the capital flight of the 1920s and 1930s; the 1960s eurodollar build-up materialised in the unwatched category — and the path by which it was finally seen is itself legible: the offshore market grew for a decade before the central banks gave it a standing committee in 1971, the regulation then debated was abandoned rather than codified (Braun, Krampf and Murau 2021), and the rule-book moved only when the failures of 1974 released the ratchet into the Basel Committee (Schenk 2014). The post-2008 stack codified categories built around the mortgage-securitisation chain of 2007–08; the non-bank credit system — private credit, direct lending — materialised in the unwatched category, the contemporary gap of Section 4. The contribution is therefore distinct from the path-dependence claim it shares with political economy: path dependence explains why a codified settlement persists; the framework explains the specific content of the surveillance failure that results, in a way that maps onto the prior pathogen and is testable against the documentary record of any surveillance regime.

2.4 The discriminating test: doctrinal versus motivated blindness

A surveillance gap can have either of two structures, and telling them apart is the framework’s central problem — and the point at which it contributes to a literature beyond monetary economics. A gap is doctrinal when a risk is simply unrepresented in the codified category-system: no committee owns it, no scenario contains it, no model can quantify it — not because anyone decided to look away, but because the rule-book does not name the risk and the regime is built around the rule-book. A gap is motivated when a risk is represented somewhere but no one is given the resources or incentive to monitor it, because the dominant coalition benefits from the absence of monitoring.

The motivated pole is already well theorised, and the paper concedes it rather than claiming it. In the sociology of ignorance, McGoey (2012) names “strategic ignorance” — ignorance cultivated because not-knowing is a productive asset, an “institutional alibi” that helps actors command resources and deny liability after the fact; Davies and McGoey (2012) apply the idea directly to the 2008 crisis. The doctrinal pole is the under-theorised one. Its closest neighbour is Best’s (2022) account of “uncomfortable knowledge” in central banking — knowledge an organisation must keep at the edge of vision to keep functioning — and her later analysis of central banks’ “contested failures,” in which the externalised risk returns “because it can no longer reasonably be ignored” (Best 2024). That image of return is, in this paper’s vocabulary, the blowback signature. McGoey theorises ignorance that is chosen; Best describes how central banks manage the ignorance built into their frames; this paper adds the typology that separates the two and a test that tells them apart in evidence. The primary discriminator turns on the form of the archival trace. A motivated gap should leave traces of contest: someone proposed monitoring the risk, an interested party resisted, the resistance prevailed — a suppressed live category. A doctrinal gap should leave an absence of a different kind: the risk does not appear as a live category at all, not because it was argued down but because the category-system had no slot for it — a missing category. That contrast, traced across the four proxies and the periods before and after a codification, is what the test rests on.

The honest difficulty must be stated, not buried, because it is the strongest objection to any documentary method of this kind. Power has a third face (Lukes 2005, after Bachrach and Baratz): its most effective form keeps a risk off the agenda entirely, so that a successful motivated suppression leaves no trace of contest — the same archival silence as a genuinely missing category. Absence of recorded resistance does not, by itself, distinguish “no one thought of it” from “no one could afford to say it.” The framework does not claim to dissolve this; it claims two partial answers that, together, do real work. The first is geometry: agenda-denial by a coalition would have to reproduce, silently and across all four independent proxies, exactly the contours of the previous settlement’s omissions, which is a far less parsimonious account of a consistent multi-layer pattern than a doctrinal lag that predicts that pattern directly. The second is blowback: a silently suppressed category, if the suppression were truly motivated, should benefit the coalition that achieved it; a missing category that detonates on the very coalition that built the rule-book is hard to read as successful hegemony and easy to read as a category nobody — including the powerful — was watching. Neither is decisive alone; jointly, the missing-category proxy, the cross-proxy geometry, and the direction of harm are far harder to satisfy under silent suppression than under a doctrinal lag. The test is probabilistic, not algorithmic, and the paper claims no more.

Blowback must be handled carefully, because it can mislead. A self-interested coalition can lobby for a blind spot, misjudge the tail risk, and be harmed by its own miscalculation — blowback under bounded rationality, with motivation fully intact. Blowback therefore cannot prove doctrinal blindness; it raises the cost of the motivated reading without settling it, and the weight is carried by the missing-versus-suppressed-category contrast, with blowback as support. The 2008 securitisation gap illustrates the limit as much as the signature: the regime left the securitisation chain outside the perimeter and was damaged by it, which is suggestive, but a determined motivated account can read it as rent-seeking that misfired — which is exactly why the archival-trace contrast must do the discriminating work. The test requires only that the consequence of codification is partly determined by the codified doctrine rather than entirely by the balance of interests prevailing while the regime operates. That is the modest claim, and the discriminating case of Section 6 is designed to put it under the sharpest available pressure.

A boundary belongs here too, because the doctrinal/motivated pair concerns what a regime fails to watch, and not every consequential effect is of that kind. Some effects are seen and measured exhaustively yet lie outside the objective the mandate sets — the distributional or climate footprint of policy is the obvious case. These are not gaps in the category-system but boundaries of the remit, drawn by the veil view: if money is neutral, what it does to allocation, distribution, or the climate is “not really monetary.” Section 4.2 marks where that edge falls. The point here is that such cases sit outside the typology, on the normative question of what the mandate ought to contain — a question the framework does not try to settle.

2.5 The dependent variable: the distribution of supervisory attention

The testable claim is that codified doctrine shapes the distribution of supervisory attention. The dependent variable is therefore attention allocation, and naming it that way places it in a recognised tradition: the attention-based view of the organisation holds that behaviour follows how an organisation’s rules, resources, and structures channel the attention of its decision-makers (Ocasio 1997). What the framework adds is that the attention structure here is a codified doctrine with a datable origin, read historically rather than cross-sectionally. The variable has four observable proxies in the documentary record of any modern central bank.

The first is the supervisory category-system: the formal taxonomy by which the data system measures, classifies, and reports the regulated population — a risk with no category cannot be counted, and what cannot be counted is, for practical purposes, not watched. The second is the scenario taxonomy: the stress scenarios the supervisor designs and runs, with the transmission channels they assume — a map of the futures an institution treats as worth imagining. The third is the distribution of supervisory staff time, visible in committee remits, divisional structure, working-paper output, and the topic distribution of the public record. The fourth is the agenda of the responsible committees: what the Monetary Policy Committee, the Financial Policy Committee, the Discount Office, or the relevant predecessor body actually met to discuss, as visible in minutes, agendas, and papers.

Two properties make these proxies suitable. They are functional rather than verbal: one can measure where attention went without finding anyone who wrote down a theory of where it should go, because officials write about the operational questions in front of them — reserve ratios, eligible liabilities, deposit-rate competition — not in the framework’s vocabulary. And the four are independent measurements of the same underlying variable, which permits triangulation: a finding on one proxy is weak, a finding consistent across all four is strong, and divergence is itself informative. The prediction is that, after a codification, the four cluster on the prior pathogen and leave structurally identifiable gaps where the omitted categories would otherwise have demanded attention.

2.6 What the framework does not claim

Three boundaries fix the argument. First, it does not claim that doctrine determines which settlement wins each codification round; that is conceded to political economy, and the framework speaks only to consequence. Second, it does not claim that the monetary system’s operation depends on the public holding a veil-leaning view. The 2014 Bulletin article shows that constitutive operational practice can be stated publicly without destabilising confidence — and that once a central bank of such standing had said so, others felt licensed to describe in public what they had long done in practice: the Magyar Nemzeti Bank’s own economists set out the endogenous-money account in the bank’s journal, citing the Bulletin approvingly (Ábel, Lehmann and Tapaszti 2016). The Dang–Gorton–Holmström literature on information-insensitive debt (Dang, Gorton and Holmström 2020; Dang et al. 2017) does not support the macroeconomic version of that supposition — as Holmström (2015) notes, information-insensitivity is a property of particular collateral, not of public belief in money-neutrality. What that literature does supply is a micro-foundation for the immune-system mechanism: safe assets work precisely because no one examines them, so a category that codes an instrument as safe is, by construction, a category that directs attention away from it. Third, the dyad is not a claim about individual thinkers, who held mixed and developing positions — Thornton had a more constitutive sensibility than the bullionist position that won, and Bagehot wrote with constitutive instincts inside a Currency-School institution. The framework operates at the level of the codified rule-book, not of any individual mind.

3. The mechanism in the historical record

This section is the paper’s first body of evidence: three episodes in which the mechanism is visible in the documentary record. They are not the decisive archival test — that is reserved for the single discriminating case of Section 6, where the doctrinal and motivated readings can be made to compete directly — but three settlements spanning three centuries, each leaving its surveillance gap in the category the prior codification omitted, are a real evidential pattern, not a set of anecdotes. In each episode the political-economy account explains why the settlement took the form it did, and the framework adds an account of how the codified categories shaped the surveillance regime that followed.

3.1 The Currency–Banking School debate and the 1844 Act

The clearest historical instance is the 1844 Bank Charter Act, because the codified rule and its repeated failure are both unusually legible. The Currency School held that only banknotes were properly money, with deposits and bills derivative, and that the remedy for panics was to tie note issue tightly to gold; the Banking School held that deposits and bills performed monetary functions in their own right, so a regime controlling notes alone would fail. A precise point is owed here, and it requires conceding something the framework’s critics rightly press. The Currency School did not fail to notice deposits or the credit system: the Banking Department’s reserve was watched closely, and the Chancellor’s Letter that suspended the Act in each panic was, in part, an understood emergency release — a clumsy proto-lender-of-last-resort, not pure incomprehension. The claim here is narrower and survives that concession. The Act codified a category-system in which money meant the note issue, and it gave the Bank no lawful instrument for acting on credit-system distress short of overriding its own founding statute. The risk was visible; the rule-book had no routine, in-category way to act on it. That is the doctrinal blindness this paper means — not ignorance of a phenomenon, but a binding codified category that forced the institution either to ignore the credit system within the rules or to step outside them by suspension. The panics of 1847, 1857, and 1866 each required the note-issue rule to be suspended by Treasury letter. Three suspensions in twenty-two years, each at the same pressure point, are not an occasional safety valve working as designed; they are the signature of a category that never fit the system it governed, which is why the Act was not amended in substance until 1928. The Banking School view was not lost intellectually — Bagehot’s (1873) Lombard Street reads as the post-hoc rationalisation of an arrangement that kept suspending its own founding rule in order to work — but it had been displaced from the rule-book, reached for each crisis as an improvisation the official doctrine could not absorb.

3.2 The interwar collapse and the Bretton Woods settlement

The interwar episode is the one the existing literature has worked out most fully, and I take it from Eichengreen (1992) rather than claim it. The classical gold standard functioned, broadly, between 1880 and 1914, amid smaller and less complex credit systems than the economies that emerged from the First World War. The attempt to restore it — most consequentially Britain’s 1925 return to gold at the pre-war parity — tested the veil view under conditions in which the credit system had become large, internationally interconnected, and operationally central. The test failed in a sequence the inherited doctrine could not absorb: the 1929 crash, the United States bank failures of 1930–31, the Creditanstalt collapse of May 1931, sterling’s exit from gold in September 1931, the 1933 American bank holiday. Keynes’s writings of the 1930s, culminating in the General Theory (1936), are best read in this context: the decisive move is the rejection of the classical dichotomy. The political codification of that constitutive moment came at Bretton Woods — managed exchange rates, capital controls, and a public credit architecture that accepted the state’s role in stabilising the credit system, taking national forms in Glass–Steagall and Regulation Q in the United States and in the sterling-area controls and governed clearing-bank system in Britain (Howson 1993).

Eichengreen’s argument is that an inherited mentalité, codified in the gold-standard architecture and in operating practice, made reflation politically unthinkable and constrained the responses available in 1929–33. That is precisely the consequential mechanism this paper proposes — a prior codification determining the shape of the response regime when the next crisis arrives — and the paper’s contribution is to generalise it beyond the gold-standard case and give it the discriminating test of Section 2.4. The interwar episode is not the framework’s novelty; it is the case in which the mechanism is already most fully worked out, and saying so places the present contribution accurately.

3.3 Monetarism, the Volcker shock, and the 1990s veil baseline

The third episode shows a codification surviving the failure of the operating rule it was built to carry. Bretton Woods broke under two stresses: the strain on the gold-dollar peg, resolved by closing the gold window in August 1971, and the stagflation the late-Keynesian Phillips curve could not explain. The same year saw the United Kingdom’s Competition and Credit Control reform — a liberalising, veil-leaning domestic reform whose surveillance consequences are the subject of Section 6. Friedman’s monetarism gained traction by recovering the classical dichotomy in a new form, and although its distinctive operating prediction — that controlling a monetary aggregate would control inflation — broke down within a decade as velocity relationships proved unstable, the political institutionalisation of the monetarist victory survived the failure of its operating proposal. The 1979 Volcker shock, the parallel disinflations, and the wave of central-bank reform statutes through the 1980s and 1990s established a baseline in which the central bank’s role was understood in veil terms even after the specific monetarist rule had been abandoned. The replacement frameworks — the Taylor rule, inflation targeting, New Keynesian modelling — preserved the veil reading, and the 1990s independence statutes codified it in law, justified by the claim that the central bank’s task is the management of nominal variables, separable from credit and fiscal conditions. This is the baseline the 2008 crisis tested and that, lightly amended, governs the present.

4. Contemporary blind spots

The framework, applied to the present, identifies where the surveillance regime is now structurally weak. It does not forecast when stress will occur; it identifies where the codified categories of the post-1990s rule-book leave gaps that a constitutive codification would not have left.

4.1 The operating framework has moved; the rule-book has not

The crucial distinction for the present is between the operating framework and the codified rule-book, and the two now point in different directions. The operating framework is already horizontalist — the policy rate is set and the reserve quantity accommodated, as Bindseil and König (2013) and McLeay, Radia and Thomas (2014) describe. The veil baseline at issue is the rule-book: the Basel hierarchy, the supervisory category-system, the scenario taxonomy, the legal definition of money, the formal mandate. That baseline is the residue of the 1990s independence wave, lightly amended after 2008. The amendments are real — resolution regimes, central-counterparty oversight, the standing repo facility, term-funding schemes — but their cumulative effect has been to absorb constitutive insights into a veil-style baseline rather than to rewrite it. This is the structure Baker (2013) identified in the macroprudential turn: an ideational shift radical in content but minimal in institutional consequence, layered onto the existing settlement rather than replacing it (Stellinga 2019).

This distinction is also, increasingly, an institutional one, and saying so sharpens the mechanism. The constitutive understanding and the veil rule-book are not two faces of a single mind that could simply reconcile them; they are lodged in different functions, and often different bodies. The endogenous understanding sits in the monetary-policy and market-operations function — the desk that accommodates reserves. The rule-book sits in the prudential and supervisory function, which is a separate authority in the twin-peaks systems and, in fragmented form, across the United States, and which even under one roof with the monetary function — the Bank of England’s Prudential Regulation Authority since 2013, the European Central Bank’s Single Supervisory Mechanism since 2014 — works from a category-system set multilaterally at Basel rather than chosen at home. A third split runs inside the central bank itself: its operating practice is constitutive, but its founding mandate, the 1990s independence statutes, is veil-leaning. The lag is therefore not a puzzle about why one actor fails to update its own beliefs; it is the predictable result of a division of labour in which the body that holds the knowledge does not own the categories, and the body that owns the categories cannot rewrite them alone — and moves slowly when it does, in the multilateral, crisis-punctuated cycles of Basel I (1988), II (2004), and III (negotiated from 2010 and implemented across the following decade), each a response to the prior crisis rather than an anticipation of the next.

The veil cast of the rule-book is clearest in its risk-measurement core. The capital regime assumes — in Value-at-Risk and the Basel risk-weights — that risk is exogenous: a statistical property of assets, estimated from history and managed from outside. Daníelsson’s long critique is a cousin of the present argument on a different axis: risk, he shows, is endogenous, generated within the system as participants react to each other and to the rules they are handed, so that “risk modelling is not an appropriate foundation for regulatory design” (Daníelsson 2002; Daníelsson and Shin 2003). His mechanism is risk dynamics where this paper’s is classification, but the two rhyme, and his adds a twist the classification account does not: a uniform rule does not merely fail to see a risk, it can manufacture one, as common models drive synchronised selling and fire sales — a regulatory monoculture fragile exactly where diversity would be safe (Daníelsson, Shin and Zigrand 2004; Daníelsson 2022). Haldane and Madouros (2012) reach the same place from the other side: simple leverage metrics out-predict the elaborate risk-weighted apparatus, so complexity in the rule-book can buy less sight, not more.

4.2 The contemporary gaps

The gaps the present rule-book carries are not, for the most part, unnoticed: each is discussed, often at length, in the narrative commentary of financial-stability reports. The claim is not that supervisors fail to see them in prose, but that they remain outside the binding codified categories — the capital hierarchy, the stress-test perimeter, the standing committee remit, the data taxonomy that fixes what is aggregated and monitored as a matter of routine. A worried paragraph in a Financial Stability Report and a line in the supervisory category-system are different objects, and the gap between them is the whole of the argument. The precise word for the present is therefore non-codification, not blindness: “blindness” is apt for the historical episodes, where the risk had no place in the rule-book and went unseen until it detonated; for the present it is shorthand for non-codification despite awareness — and it is the routine allocation of attention that the dependent variable tracks.

The clearest gap is private credit and the non-bank financial system. The post-2008 codification built its taxonomy around the mortgage-securitisation chain of the prior crisis. The fastest-growing site of credit creation since 2010, private credit and direct lending, is flagged repeatedly in stability commentary yet sits substantially outside the codified apparatus: it is not aggregated in supervisory data the way bank credit is, it is not in the principal stress scenarios, and it is the standing remit of no central-bank committee. Thiemann’s (2018) study of how regulators tolerated the bank/shadow-bank perimeter is the empirical authority here: supervisors remained wedded to formal models that exclude observable phenomena they cannot accommodate. When the credit cycle turns and assets never priced through a default cycle face one, the dislocation will materialise in a category the regime is poorly equipped to monitor in real time.

Stablecoins show the same perimeter being drawn in real time, and the live drawing makes the case instructive rather than settled. Stablecoins and tokenised deposits are runnable private money — they carry the systemic risks of credit money: runs, fire-sales of reserve assets, propagation into short-term dollar funding (Gorton and Zhang 2021) — yet the current codifications frame them predominantly as payment rails: the United States GENIUS Act, the Bank of England’s systemic-stablecoin regime, the EU’s MiCA. Because the codification is happening now and in plain view, this is not yet a settled doctrinal category but an active contest, with the traces of contest the archive of a contest leaves; that the legal definition does the work is Pistor’s (2013) point, and the design question is Ricks’s (2016). Whether such runnable claims belong in the money supply at all is contested: Nersisyan and Dantas (2017) would broaden the money concept to take in non-bank liabilities, and Michell (2017) traces shadow money through the monetary circuit, but the stronger case is that they do not count, because banks remain special — uniquely backstopped and supervised (Bouguelli 2020) — which is why moving credit-money risk outside the bank perimeter is the danger. The framework’s prediction is therefore about how the contest settles: it expects the payment-rail framing to harden into the routine category while the credit-money risks are left outside it, until a par failure forces the ratchet. The prediction is directional, and that is what makes it falsifiable: were supervisors to recodify stablecoins as credit money pre-emptively, absent a failure, that would be the rule-book updating without a crisis release — evidence against the ratchet. This is the one prospective, datable test the contemporary section offers (Section 5.1).

A third gap is implicit fiscal dominance, located in the mandate and committee-agenda proxies. The veil reading embedded in the 1990s settlement requires the central bank to be detachable from the fiscal authority, and the rule-book still carries that requirement; the substantive operations of large central banks increasingly do not match it — extended balance-sheet operations, purchases of long-duration sovereign debt, the central bank as market-maker of last resort whose mandate, as Gabor (2021) shows, has been extended through often-clandestine steps since 2008. The joint fiscal-monetary decision-space has no category, so attention is not allocated to it even as the substantive coordination deepens.

The framework has an edge worth marking, because not every consequential effect is a gap of this kind. Some effects are seen and measured exhaustively yet lie outside the objective the mandate sets: the distributional footprint of monetary policy, which is a channel of transmission rather than a side effect (Auclert 2019); the impact of finance on the climate, as against the risk climate poses to financial stability; the international spillovers large central banks generate (Rajan 2016; Rey 2015). These are not missing categories but bounded objectives, and the veil view is what draws the boundary. Where the boundary is held by a doctrine of market neutrality that codes a real allocative choice as neutral plumbing (Braun 2020; Thiemann, Büttner and Kessler 2023), the framework’s reading shades into the political economy of depoliticisation and hands off to it. The paper claims only that the veil view is what makes the boundary feel natural, and that monetary policy is in any case a blunt instrument for objectives it was not built to pursue (McKay and Wolf 2023).

4.3 The bureaucratic mechanism of structural power

This framework does not sit downstream of political economy in a separate vacuum; it supplies one of its missing mechanisms. Krippner (2011) reads the modern regime as a depoliticisation strategy — the state evading accountability for credit allocation by delegating to a technocratic central bank. The codification of doctrine is how such a settlement makes its power durable and cheap to maintain: a victorious coalition’s preferences, once written into categories, remits, and data systems, no longer need defending, because they have become the unexamined background against which everyone works. What the framework adds to the depoliticisation account is the content of that background — which risks the resulting regime can and cannot see. Depoliticisation alone does not predict that private credit will be under-monitored or stablecoins left at the perimeter; the framework predicts those specific gaps because the categories carry the doctrinal moment at which they were written. Political economy explains the structure of authority; this paper explains the structure of attention — the bureaucratic micro-mechanism by which a political settlement reproduces itself as a blind spot.

4.4 Where the framework could be wrong about the present

The diagnosis is conditional. It over-predicts failure in these categories if the rule-book can be quietly upgraded without a crisis-induced release of the ratchet — and several recent reforms are exactly such upgrades, which may cumulatively update the category-system before the next stress arrives. A second possibility is that the discriminating test of Section 2.4, run on the case set out in Section 6, will not cleanly separate doctrinal from motivated blindness; in that event the claim about the specific content of gaps would need substantial reconsideration. Both possibilities are open. The diagnosis stands only so long as the codified categories retain their grip in the absence of crisis-driven renegotiation — which the evidence so far supports but does not guarantee.

5. Objections and replies

A framework that proposes a consequential role for codified doctrine will draw objections from several quarters. The seven most serious are set out below.

5.1 Falsifiability. Any surveillance gap can be redescribed, after the fact, as a doctrinal blind spot; what would refute the framework? The consequential claim is operationalised through four observable proxies (Section 2.5). For any codification, the framework predicts that the proxies cluster on the prior pathogen and leave gaps in the omitted categories. A codification whose regime does not cluster on the prior pathogen, or whose later gaps do not map onto the omitted categories, would refute the claim directly. The stablecoin prediction of Section 4.2 supplies a live, dated instance: a pre-emptive recodification as credit money, absent a failure, would count against the ratchet. And the discriminating test supplies a second path: if gaps consistently track motivated rather than doctrinal blindness, the framework’s distinctive contribution dissolves and political economy is the better account on its own. All three are live possibilities.

5.2 The post-Keynesian objection. The constitutive view just is the post-Keynesian endogenous-money position, developed over forty years and now accepted operationally; you cannot claim novelty for it. Conceded in substance. The paper claims no novelty for the endogenous-money description; that has been settled by the post-Keynesian literature and conceded by the institutions it criticised. The contribution is downstream: the operating framework is endogenous, the codified rule-book has not been updated to match it, and the resulting gap shapes the surveillance regime in ways the post-Keynesian literature has not analysed in detail — the critical-macrofinance strand (Gabor 2021; Sissoko 2019) is the partial exception, alert to shadow banking’s dangers, if focused on the plumbing rather than on the supervisory category-system. The structuralist strand (Chick 1992; Dow 1996; Palley 2013) is not a rival but the engine. Structuralists hold that banks continually innovate new credit forms to evade existing constraints as demand expands — liability management, off-balance-sheet vehicles, the shadow-money chain — which is precisely the private sector generating categories faster than the rule-book codifies them. The doctrinal blind spot is therefore not a static remainder but is actively and continuously produced: structuralist evasion on the market side meets a codified taxonomy that updates only at crises on the regulatory side (the clock asymmetry of Sections 2.2 and 4.1). So the horizontalist–structuralist debate about the slope of the credit-supply schedule is not the issue here; the structuralist account of endogenous category-creation is, because it supplies the mechanism by which the rule-book is kept perpetually behind. That engine has a regulatory mirror: a single codified rule-book homogenises behaviour across institutions, so the uniform categories do not merely lag the market but can themselves generate instability — which is why the corrective is diversity as much as speed.

5.3 The intellectual-historian’s objection. A dyad applied across two centuries imposes a present categorisation on past thought. Acknowledged in part. Individual thinkers held mixed positions, and the dyad is not a faithful description of any single body of thought taken whole. But the consequential claim does not require the dyad to be true at the level of thinkers; it requires only that the codified rule-book carries categories mapping onto one side of the dyad in a given period, which is the level at which the historical claims operate. The dyad lives in the codified categories, not in the heads of the economists — and the exogenous/endogenous distinction is, in any case, the working axis of the contemporary systemic-risk literature (Daníelsson and Shin 2003), not a historian’s anachronism.

5.4 The political-economy objection. Monetary doctrine is ideology in the service of interests; having conceded co-constitution, the framework has nothing distinctive to add — temporal survival is path dependence, the compositional channel is Eichengreen. The challenge is taken seriously, and the present framing reflects a substantial concession to it. The paper no longer claims that doctrine determines which settlement is codified; interests, coalitions, and path dependence do that work, and Golden Fetters is accepted as the canonical worked example for the interwar period. What the paper adds is generalisation, a test, and a mechanism: the doctrinal/motivated distinction with its blowback signature, applied beyond the gold-standard case and located within the sociology of ignorance (Section 2.4); and the account, in Section 4.3, of codification as the bureaucratic means by which a settlement reproduces itself as a blind spot. The 2008 securitisation gap is the candidate exhibit — a gap that damaged the very banks the settlement had served, which a motivated account strains to explain and a doctrinal account accommodates directly. The framework, the typology, the three historical episodes, and the contemporary diagnosis are the contribution offered here, with the Section 6 case as the sharpest future test.

5.5 The “fossilised motives” objection. If political economy decides the original codification, then the categories are fossilised motives, and “doctrinal” blindness is merely motivated blindness with a time delay; the distinction dissolves once you stop freezing time. This is the deepest objection the framework faces. I grant the premise entirely: the categories are, in origin, the residue of a motivated settlement. The reply is that origin and operation come apart over time, and the framework is a claim about operation. A category written to serve a coalition becomes, a generation later, a constraint that officials neither chose, defend, nor can explain — they apply the rule-book because it is the rule-book. The mechanism of the blindness has changed form: from active, conscious suppression by an interested party to passive, structural omission by a bureaucracy following an inherited classification. Those are different sociological states, they leave different archival traces (a suppressed live category versus a missing one), and — decisively — they call for different cures (Section 7). A statute passed for a lobby still hardens into an unquestioned constraint; that the constraint began as a motive does not make its later operation motivated, any more than a vestigial organ is still performing the function it was selected for. The distinction is not a denial of the political origin; it is the claim that the origin stops doing the explanatory work once the category has gone infrastructural, and that what then governs attention is the category itself — which is precisely what neither pure political economy (which collapses origin into operation) nor classification theory (which ignores origin) supplies.

5.6 The performativity objection. This is performativity, or framing-and-overflowing, under another name. The objection has force, because the framework shares vocabulary with that family — Best’s (2022; 2024) use of Callon’s framing and overflowing is a close cousin of the blowback signature. Performativity theorises overflow as a general property of all framing: every frame leaves a remainder. The present framework makes a more specific claim — that a particular codified frame leaves an asymmetric and datable blind spot whose location is partly predictable from the content of the prior settlement — and it supplies a typology and a falsification test that the performativity programme does not. The risk the performativity literature runs, which this paper avoids, is conflating the model with the reality it formats (Curran 2018); the paper is about exactly the gap between an operating reality that is constitutive and a rule-book that is veil-leaning. Performativity is a literature the paper extends at a specific point, not one it leans on or dismisses.

5.7 The methodological objection. Even if conceptually clean, the test may be impossible to run, because officials operate inside both the category-system and the interest structure at once. This is the most important objection to the empirical programme, and the one the Section 6 case is chosen to meet. That case is tractable with unusual clarity because the reform was lobbied for by one coalition and the crisis detonated in a sector that coalition had limited interest in but had left outside the perimeter. Whether the gap was doctrinal or motivated is a question on which the contemporaneous record can be expected to discriminate, because the two readings predict different things about what the archive contains: a suppressed live category versus a missing one. The distinction is not always clean in every historical case; the claim is that this case is clean enough to give a meaningful first test.

5.8 The illusion-of-control objection. If risk is endogenous and crisis-time risk cannot be measured in advance, then the prescriptive move — add the missing category, update the taxonomy — is itself an instance of the illusion of control the systemic-risk literature warns against (Daníelsson 2022). The framework concedes most of it. It does not claim that a better category, or a better model, would let the regime measure endogenous risk; the paper has disclaimed point-precision prediction, and no ex-ante classification captures a feedback generated as it runs. But the concession does not reach the dependent variable. The claim is about the routine allocation of attention, not the measurement of tail risk: a risk with no codified category is not aggregated, not stress-tested, and not on any committee’s standing agenda, whatever its measurability. The corrective is correspondingly modest — represent the risk so a skilled eye is on it, widen the perimeter so it falls inside some remit, and preserve the diversity that stops a monoculture forming — which is not a rival to Daníelsson’s prescription but nearly a restatement of it. Putting a risk where someone is bound to look is not the same as promising to measure it; the illusion to be avoided is the belief that the existing rule-book already sees what matters, and naming the gap is the opposite of that illusion.

6. A research programme: extensions and a discriminating case

The framework stands on what precedes this section. This section looks outward, to the work it opens up. The most important item is a single archival case designed to push the doctrinal/motivated distinction harder than the illustrative episodes can.

A discriminating case: Competition and Credit Control 1971. The historical episodes of Section 3 show the mechanism operating, but in each the doctrinal and motivated readings are hard to separate cleanly because the documentary record is distant or thin. A sharper test requires a case recent enough to be densely archived, bounded enough to trace in full, and structured so that the two readings make genuinely competing predictions. The United Kingdom’s Competition and Credit Control reform of 1971 and the Secondary Banking Crisis of 1973–75 is such a case — the subject of a companion empirical paper now in preparation. CCC replaced direct quantitative ceilings with a market-conforming reserve-asset regime applied uniformly to the listed banks; the reform’s text drew no line between clearing and secondary banks. The line that mattered was the one CCC inherited and left intact: there was no formal definition of which institutions counted as banks for prudential purposes, supervision remained relational and informal — conducted through the Discount Office, whose remit ran historically to the accepting houses and the discount market rather than to the new deposit-takers — and the “fringe” or secondary banks, many trading on section 123 certificates, stood outside the principal supervisory categories altogether. The reform thus released a far more competitive credit market through a perimeter nobody had formally drawn. The crisis then materialised in exactly that unrepresented zone: from late 1973 a sequence of secondary banks failed after lending into commercial property on wholesale funding, and the Bank organised a “lifeboat” — clearing banks supplying roughly ninety per cent of the support against the Bank’s ten — that ran through 1974–75 and fed into the next turn of the ratchet, the Banking Supervision Division (1974) and the Banking Act 1979. Reid (1982) reads the episode as a credit-cycle event amplified by the post-CCC expansion; Capie (2010) treats reform and rescue in detail.

What makes the case discriminating is that it is a hard case for the framework and an easy one for political economy, so a positive result is informative. CCC was lobbied for by the clearing banks, yet the blowback fell partly on them: the lifeboat required their participation and exposed them to losses. A pure motivated account must explain why a self-interested coalition secured a regime that would, within two years, expose it to losses it had not designed to bear; a doctrinal account says only that the category-system CCC inherited and entrenched named no place for the systemic risk the secondary sector would run up. The blowback is partial — the clearing banks’ core business survived — but partial blowback is still discriminating, because a fully motivated account predicts no self-harm at all. The empirical question is then specific: across the four proxies of Section 2.5, measured before and after the 1971 codification, do the contemporaneous documents show the secondary sector as a suppressed live category (someone proposed supervising it and an interested party prevailed) or as a missing one (a category-system artefact nobody was attending to)? The companion paper draws on the Bank of England and Treasury archives for 1969–76; the present paper specifies the test and its logic, and offers the framework on the evidence already assembled here.

Other extensions. The euro is a half-constitutive settlement — constitutive at the level of state money, veil-leaning in its operating mandate, its no-bailout clause, and the architecture of TARGET2 — and the episodes of 2010–12 and 2020 can be read as the tension between the two. The digital-money frontier is more live: the choice now being made between a fully reserved central-bank-digital-currency architecture (a veil codification) and an open-access architecture that backstops private tokenised liabilities (a constitutive one) will be hard to reverse and will shape the surveillance regime that follows it. Methodologically, the dependent-variable definition invites a matched-pair design — comparing jurisdictions with similar material interests but divergent codifications, such as the Bank of England and the Federal Reserve on private credit. The eurodollar episode invites the same treatment in historical form: whether the offshore dollar was, inside the Federal Reserve of 1960–74, a missing category or a suppressed one is a question the FOMC records and the 1971 reappraisal of the discount mechanism can answer, and the answer would calibrate the stablecoin diagnosis of Section 4.2 against its closest historical precedent.

7. Conclusion

The descriptive question about money is settled: the operating framework of the major central banks is endogenous, and the post-Keynesian literature established this long before the institutions admitted it. This paper has asked what follows from the fact that the codified rule-book has not kept pace, and has argued that the answer is a specific, traceable pattern of institutional blindness. Codified categories shape what the surveillance regime can see; a category-system built for the last crisis is, by the same token that makes it effective against that crisis, blind to risks of a different shape; and the resulting gaps map onto the doctrinal moment at which the categories were written.

The claim is narrow by design, and the narrowness is what makes it testable. It does not explain which settlement gets codified — political economy does that — and it does not forecast when stress will come. It explains the structure of attention a codification leaves behind, and locates that explanation within the sociology of ignorance, distinguishing the doctrinal blindness built into a category-system from the motivated blindness a coalition cultivates, with a falsification test that tells them apart. The dependent variable is operationalised through four observable proxies; the mechanism is shown in three historical episodes and the contemporary diagnosis of Section 4; and a discriminating archival case — Competition and Credit Control and the Secondary Banking Crisis — is specified for a companion paper, designed so that its result will be informative however it resolves.

Two modest things follow that a purely retrospective political-economy account does not provide. Both concern supervisory practice, not macroeconomic management. The first is prescriptive about the opening move. Where a gap is doctrinal, the first obstacle is bureaucratic inertia and intellectual path dependence, and the first move is technical — represent the risk in the taxonomy, re-scope the committee, widen the data system so a skilled eye is bound to fall on it. The aim is to put the risk where someone is looking, not to promise it can be measured: endogenous risk outruns any model, so the corrective is representation and diversity, not the illusion of control (Daníelsson 2022). Where the gap is motivated, the first obstacle is a beneficiary, and the first move is political. A determined critic will note that the two converge downstream: the moment a doctrinal cure threatens an interest, the interest mobilises and the epistemic fix becomes a political fight. That is true, and it relocates the distinction’s value to the start of the process, where misdiagnosing the initial obstacle guarantees a wasted first move. The second thing is prospective: because doctrinal gaps are legible in the observable, public features of the current rule-book, the framework can point to where the next gap is likely to sit before it is realised, whereas an account that waits for the lobbying to leave a footprint can only work after the fact.

In the present period the codified rule-book is veil-leaning even though the operating framework has converged on endogenous money, and the framework identifies the gaps the rule-book carries — private credit and the non-bank system, stablecoins at the perimeter, the unrepresented fiscal-monetary space — and marks the edge beyond which effects are seen but lie outside the mandate. It predicts that, until a crisis large enough to release the ratchet forces a renegotiation of the categories, the gap between the operating framework and the rule-book will widen. The paper is offered to monetary economists and financial-stability researchers as a frame for reading where supervision is structurally weak — not by hunting for villains, but by finding the widest gap between how the system works and how its categories are written, and watching there.