INTRODUCTION
1. The administration of money by discretionary authority has been a slow disaster for economic autonomy. It has permitted the financing of undertakings no population would have funded directly, has transferred wealth continuously and invisibly from those who hold currency to those who receive it first, has made the savings of ordinary people contingent on the judgment of committees they cannot inspect, and has produced in the industrialized world a diffuse sense that one's economic position is subject to revision by parties one will never meet.
2. We are not arguing that central bankers are wicked, and we do not think they are. The argument only becomes serious once one grants that they are competent, well-intentioned, and frequently correct. A monetary authority staffed by fools would be a scandal with a remedy. A monetary authority staffed by capable people producing these outcomes anyway is a structural problem, and it is the only kind we are interested in here.
3. Nor do we claim that the system is on the verge of collapse. Predictions of imminent monetary collapse have been made continuously for fifty years by people selling something, and they have been wrong every time. We expect the system to persist. Our thesis concerns what it does while persisting.
4. The reader will observe that this essay appears alongside documents advocating exit, and will reasonably expect it to conclude that Bitcoin solves the problem. It does not, quite, and the second half of this essay is largely concerned with why.
5. Our thesis in brief: money is a claim on the output of other people; the authority to create it is therefore the authority to redistribute that claim; this authority is exercised continuously and cannot be exercised neutrally; the redistribution is invisible to those it affects because it appears as prices rather than as a transfer; and the arriving generation of programmable currency will make the mechanism more precise, more granular, and correspondingly harder to leave.
6. We proceed as follows. We first describe what money creation actually consists of, since most objections to what follows are objections to a misunderstanding of it. We then describe the distributional mechanism and why it is invisible. We then examine the coming architecture of programmable money and what it changes. We then examine the proposed exit and its actual costs, at greater length than its advocates usually manage. We close with what can honestly be recommended.
7. A note on tone. This subject attracts two kinds of writing: the technical literature, which is careful and declines to draw conclusions, and the popular literature, which draws conclusions and is not careful. We have tried to be careful and to draw conclusions, and where the two pull apart we have marked the seam rather than hiding it.
THE PSYCHOLOGY OF MODERN MONETARY OPINION
8. Before setting out our position we must dispose of the two camps to which it will otherwise be assigned.
9. The first holds that monetary policy is a technical matter best left to specialists, that the questions involved are empirical and largely settled, and that objections to central banking are the product of economic illiteracy. This is not stupid and much of it is true. The technical literature is genuinely difficult, and a great deal of popular criticism is confused about basic mechanics.
10. But notice the structure of the claim. It converts a question about who should hold a power into a question about who understands the machinery, and only the second question has an answer the specialist finds comfortable. Expertise in operating an instrument is not authority over whether the instrument should exist, and the elision is so routine that it is no longer noticed by the people performing it.
11. The second camp holds that fiat money is a fraud, that hyperinflation is imminent, that the institutions are a conspiracy, and that the whole edifice will shortly fall. We think this is mostly false and its persistence is instructive. It is a position adopted by people who need the system to be criminal, because a system that is merely structurally biased offers no villain and no vindication date.
12. The two camps share a premise: that the important question is whether the managers are competent and honest. The specialist says yes and concludes there is no problem. The conspiracist says no and concludes there is a scandal. We argue that they are largely competent and honest and that there is still a problem, which is why we irritate both.
13. There is a third position, more serious, held by many working economists: that discretionary monetary authority is imperfect but that the alternatives are worse, that the gold standard produced deeper depressions, and that a central bank able to act as lender of last resort has prevented catastrophes that would otherwise have occurred. We think this position is largely correct on its own terms and we do not intend to refute it. We observe that it concerns stability, and that our objection concerns distribution and consent, and that a system can be genuinely good at the first while remaining unaccountable on the second.
14. A definition, since the argument depends on it. By money creation we mean the expansion of the total stock of claims on real output, whether by central bank reserve creation, by commercial bank lending, or by any successor mechanism. We are not interested in which institution performs it. We are interested in the fact that it is performed, that it is discretionary, and that its effects are distributional.
15. One preliminary concession. We write from within wealthy currency areas, and the argument reads differently elsewhere. A citizen of a country whose currency has actually failed does not need this essay, and the remedies we discuss late are not marginal for them but urgent. We have kept the focus narrow deliberately, and the reader should discount accordingly.
WHAT MONEY CREATION ACTUALLY IS
16. Most disagreement about this subject is disagreement about mechanics, so we begin there.
17. Money is not principally created by governments printing notes. In modern systems the great majority is created by commercial banks in the act of lending: the loan and the deposit come into existence together, the bank's balance sheet expands on both sides, and the deposit is spendable money that did not exist the previous day. The central bank does not create this money. It sets the conditions — the price of reserves, the regulatory constraints — under which private institutions create it.
18. This matters for the argument in a way that cuts against the popular critique. The system is not a printing press operated by officials. It is a distributed process in which thousands of private lending decisions expand the money stock, steered indirectly by an authority that influences the price of credit. The steering is real but it is loose, lagged, and frequently surprises the steerers.
19. It also cuts against the specialist's defense, though less obviously. If the money stock is determined by a distributed process that the authority influences but does not control, then the authority's claim to be managing the currency in the public interest is a claim about something it steers rather than something it holds. The steering wheel is connected, but the linkage is long.
20. The reader may ask why we labor this. Because the exit argument is frequently made against a caricature — a state printing press debasing an otherwise honest currency — and an argument that defeats a caricature has defeated nothing. What we are describing is worse than the caricature in one respect and better in another. Better: nobody is simply printing. Worse: the process has no single author, which means it has no one who can be held to account for its aggregate effects.
21. We should also note what the authority does in a crisis, since this is where the discretionary power becomes visible. It creates reserves at whatever scale is required and purchases assets with them. In 2008 and again in 2020 the sums involved exceeded, within weeks, what parliaments debate for years. The decisions were made rapidly, by small groups, under genuine emergency, and were in our view probably correct.
22. We stress probably correct, because the point is not that these were bad decisions. The point is that a capacity to commit resources at that scale, that speed, and with that little prior authorization is a form of power whose existence is barely acknowledged in the ordinary description of how these societies govern themselves. It is not hidden. It is simply not counted as power, because it does not take the form of a law.
THE DISTRIBUTIONAL MECHANISM
23. We come now to the central point.
24. New money does not arrive everywhere at once. It enters at particular points — the banking system, the holders of the assets purchased, the borrowers who qualify for credit on the new terms — and spreads outward through the economy over months and years. Those who receive it early spend it at prices that have not yet adjusted. Those who receive it late, or never, encounter the adjusted prices with an unchanged income.
25. This is not a controversial claim and it is not a fringe one. Richard Cantillon set it out in the 1730s, describing how money entering through the mines and through foreign trade enriched those nearest the point of entry and raised prices sector by sector as it spread. Hume made the same observation a decade later. It has been in the standard literature ever since, and it carries Cantillon's name. What is remarkable is not that it is disputed but that it is so rarely described as what it plainly is: a transfer, continuous and unlegislated, from the late recipients to the early ones.
26. We will use the standard term, the Cantillon effect. It is our central mechanism and everything downstream depends on it. We note that the term has lately acquired a partisan flavour through its adoption by monetary reformers of various stripes, and that this has made some economists reluctant to use it. The reluctance is a fact about the term's associations and not about the phenomenon, which is orthodox.
27. The Cantillon effect is not a side effect of monetary expansion; it is the mechanism by which monetary expansion has any effect at all. If new money reached everyone simultaneously and proportionally, all prices would adjust at once and nothing real would change. Monetary policy works because the money arrives unevenly. The distributional consequence is not a flaw in the instrument, to be engineered away by better design. It is the instrument.
28. We regard this as the strongest form of the argument and we want to state it precisely, because it is frequently overstated. We do not claim the effect is large relative to other forces acting on the distribution of wealth. We do not claim it is the principal cause of inequality; the evidence there is contested and we are not competent to settle it. We claim only that it is real, that it is a transfer, that it is continuous, and that it was never voted on.
29. Consider who is early. Financial institutions, by construction. Holders of the assets the authority purchases, whose prices rise first. Borrowers large enough to access credit at institutional terms. Consider who is late. Wage earners, whose incomes reprice annually at best. Savers holding currency. Anyone whose claim on output is fixed in nominal terms — pensioners, holders of long bonds, the young accumulating a deposit for a house whose price is set by the credit conditions they have not yet accessed.
30. The effect on this last group deserves particular notice, because it is where the mechanism does its most visible social damage while being least recognized. Asset prices respond to credit conditions faster than wages do. A generation attempting to buy assets out of wages is therefore running against a current, and experiences it not as a monetary phenomenon but as a personal failure, or as the fault of the generation that bought earlier.
31. This misattribution is the second half of the mechanism. The transfer is invisible because it does not appear as a transfer. It appears as prices. There is no line item, no notification, no counterparty. The person on the losing side experiences a world in which things have become expensive, and reaches for whatever explanation is culturally available: greed, foreigners, the young, the old, the immigrants, the landlords.
32. We think this accounts for a good deal of the political disorder of the present period, and we offer the claim tentatively because it is the kind of monocausal explanation this essay elsewhere warns against. But the shape fits. A population subject to a real and continuous economic injury with no visible author will produce authors, and the authors it produces will be whoever is nearby.
33. Note the contrast with taxation, which performs transfers of comparable scale. A tax is legislated, itemized, debated, and attributable. One knows the rate, one can compute the liability, and one can vote against the people who set it. Whatever else is wrong with taxation, it is legible. The Cantillon effect performs a similar function with none of these properties, which is precisely why it attracts so little opposition: opposition requires an object.
34. We should be fair to the other side. The standard defense is that the alternative — a fixed money stock in a growing economy — produces its own distribution, favoring existing holders of currency at the expense of debtors and new entrants, and that this is not obviously more just. The defense is correct. There is no neutral monetary arrangement. Every rule about money advantages someone.
35. But observe what this concession actually establishes. If no arrangement is neutral, then the choice of arrangement is a political question and not a technical one. And if it is political, the case for settling it inside an institution designed to be insulated from politics becomes considerably harder to make. The specialists' strongest argument — that neutrality is impossible — is also the argument that removes the ground for their own authority.
PROGRAMMABLE MONEY
36. We turn now to what is arriving, since it changes the analysis.
37. Central bank digital currency is generally presented as a modernization: faster settlement, lower cost, financial inclusion for the unbanked, a public alternative to private payment monopolies. Each of these is a real benefit and we do not dismiss them. The existing payment infrastructure is slow, extractive, and genuinely worse than what the technology permits.
38. But a digital currency issued directly by the monetary authority differs from existing money in a way that is not a matter of degree. Present-day money is a bearer instrument at the margin and an entry in a private ledger otherwise. The authority influences it. The authority does not hold it, does not see most of it, and cannot act on any particular unit of it.
39. A currency issued as a direct liability of the authority, recorded in an infrastructure it operates, is a different object. It is money over which the issuer retains an operational relationship after issuance. Whether the issuer exercises that relationship is a policy question. That it possesses it is an architectural fact, and architecture outlasts policy.
40. We want to be precise about what this permits, avoiding the more excitable claims. It permits transaction-level visibility, subject to whatever privacy design is adopted. It permits money with rules attached: expiry dates, sectoral restrictions, holding limits, differential interest by holder. It permits action at the level of the individual account rather than at the level of the aggregate. And it permits all of these to be implemented by configuration rather than by legislation.
41. That last property is the one we regard as decisive. Every capability listed above exists today in some form — governments freeze accounts, restrict transactions, subsidize sectors, sanction individuals. But each currently requires an institution to act, a legal instrument, a bank to comply, a process that leaves a record and can be contested. Programmable money collapses that distance. The action becomes a parameter change.
42. The standard reply is that these capabilities will be constrained by law, that privacy will be designed in, and that democratic societies will not permit abuse. We take this seriously and we note two difficulties.
43. First, the constraints and the capabilities are not equally durable. The capability is in the architecture and persists as long as the system does. The constraint is in policy and persists as long as the current political consensus does. Building a capability on the assumption that its constraints will hold indefinitely is a bet on the permanence of present politics, and that is not a bet the historical record supports.
44. Second, the capabilities will be used first in cases where nearly everyone approves. Sanctioned individuals. Terrorist financing. Sectoral relief that must not be spent on the wrong things. Benefits that must reach the intended recipient. Each application will be defensible on its own terms, and the precedent established by each will be available for the next. This is the pattern by which every surveillance infrastructure of the past thirty years was built, and there is no reason to expect a different one here.
45. We do not predict a dystopia and we want to distance ourselves from those who do. The likely outcome in a stable wealthy democracy is a payment system that is faster, cheaper, better for most people most of the time, and that carries a capability which is exercised rarely, at the margins, against people the majority does not mind seeing it exercised against.
46. Which is exactly the problem, stated in its least dramatic form. The system will work well. The capability will sit unused for years. And the question of what happens when it is eventually pointed somewhere else will have been settled long before anyone thinks to ask it, by the fact that the infrastructure was built.
47. There is a further consequence specific to our earlier mechanism. Programmable money makes the Cantillon effect targetable. At present the authority can only expand aggregates and watch where the money goes. With direct issuance it can determine the point of entry precisely — this sector, this account class, this region. This is presented as an improvement, and in efficiency terms it is. It also converts a diffuse structural bias into a series of specific decisions about who receives new money first, which is either a great improvement in accountability or a great improvement in the technology of patronage, depending on institutions we cannot inspect in advance.
THE PROPOSED EXIT
48. We come to the remedy this site exists to advocate, and we intend to examine it more sceptically than is customary here.
49. The Bitcoin proposition, reduced to its core, is this: a monetary asset whose issuance schedule is fixed by protocol rather than by discretion, whose ledger any participant may verify, and whose transfer requires no permission from any institution. If the objection to the incumbent system is that discretionary authority over money is unaccountable, then an instrument with no discretionary authority at all is a direct response.
50. We regard this as a genuine achievement and not a speculative one. Whatever else is true, a working monetary system with no administrator has existed continuously for over a decade and a half, which was widely held to be impossible. That is a demonstrated fact about what is buildable, and it does not stop being interesting because the price is volatile.
51. The property that matters is exit. Not that the instrument performs better, but that holding it does not require anyone's permission and cannot be revoked by configuration. A holder of keys is in a different structural position from a holder of an account, and the difference is not one of degree.
52. Now the difficulties, which are serious.
53. The fixed supply is a policy, not the absence of one. A protocol with a fixed issuance schedule has made a monetary policy decision and hardcoded it. Recall ¶34: no arrangement is neutral. A fixed supply advantages early holders and existing savers over debtors and new entrants, systematically, by design. This is a defensible choice. It is not the absence of a choice, and describing it as neutrality is a rhetorical move that should not survive contact with ¶35, which we deployed against the other side and must now accept ourselves.
54. The verification argument is largely notional. The instrument's claim to trustlessness rests on the ability of any participant to verify the chain. The number of holders who run a validating node is small. The number who have audited the implementation is far smaller. Most holders trust an exchange, a wallet vendor, or the general impression that someone competent is checking. This is automation bias operating in the one domain built specifically to make it unnecessary. A verification that only specialists can perform is not verification; it is trust with additional steps and better marketing.
55. Concentration has reasserted itself. Mining is capital-intensive and geographically concentrated. Custody has consolidated into a small number of large institutions. The exchange-traded products that brought the asset to institutional portfolios reintroduce, precisely, a custodian who may be regulated, compelled, or frozen. A great deal of the ownership is now structurally indistinguishable from ownership of any other financial asset, differing only in what it tracks.
56. This last point deserves to be stated without softening, since it is the one most often waved away in documents of this kind. An instrument designed to eliminate custodians has been adopted, at scale, principally through custodians. The property that made it interesting is available to holders who take custody themselves and is simply absent for the rest. Most holders have bought the price exposure and not the exit.
57. Self-custody transfers risk rather than eliminating it. The holder who takes custody has replaced institutional risk with operational risk borne alone: irrecoverable keys, no reversal of error, no recourse against theft, and an inheritance problem that has already destroyed real fortunes. This is a genuine trade and for many people it is the wrong one. An elderly person of modest means is not obviously better served by an arrangement in which a single mistake is final.
58. The instrument is not yet money. It is held overwhelmingly as an asset and used marginally as a medium of exchange, and its volatility is the reason. An instrument that may lose half its value in a quarter cannot denominate a wage, a rent, or a debt. This may change with scale. It has not changed yet, and an exit that requires re-entry to the incumbent system in order to buy anything is a partial exit at best.
59. We note in passing that the two properties are in tension. The instrument is attractive as an asset precisely because of the appreciation that its fixed supply produces, and that appreciation is what makes it unusable as a unit of account. Success at the first function postpones the second.
60. The regulatory surface is real. Exit is exercised by people who live in jurisdictions, hold jobs, own property, and pay taxes. The protocol cannot be shut down; the on-ramps, the exchanges, the custodians, and the individuals can all be reached. Sovereignty over an asset is compatible with a great deal of leverage over its holder, and arguments that ignore this are arguing about protocols while the pressure is applied to people.
61. Having listed these, we should say what survives, because we do not think the objections are fatal.
62. What survives is the option. The existence of a monetary instrument outside the discretionary system changes the position of everyone in it, including those who hold none, because it establishes a boundary condition. An authority that knows an exit exists is constrained differently from one that does not. This is a modest effect and it is not nothing.
63. What also survives is the demonstration. Prior to 2009 the claim that money required an issuer was treated as a fact about the world rather than a fact about available technology. That is no longer available as an assumption, and it will not become available again. Whatever eventually displaces the present arrangement will be built by people for whom administered money is one option among several rather than the definition of the thing.
64. What does not survive is the maximalist version: that holding the asset constitutes an exit, that the exit is available to anyone who buys, and that the incumbent system is therefore already obsolete. The exit is available to holders who take custody, understand what they are doing, and accept operational risks that many people should not accept. That is a real population and it is not most people.
WHY REFORM IS DIFFICULT
65. It will be objected that the correct response to all of this is reform rather than exit — better mandates, distributional targets, democratic oversight of the monetary authority. We think most of these proposals are sincere and that they encounter obstacles their proponents underrate.
66. The first is that the authority's insulation from politics is not an oversight but its central design feature, adopted deliberately after a period in which politically directed monetary policy produced inflation that fell hardest on the poor. Proposals to democratize the authority are proposals to undo a specific historical remedy, and they need to engage with why it was adopted. Most do not.
67. The second is that the Cantillon effect has no policy handle. It is not produced by a rule that could be amended; it is produced by the fact that money enters the economy somewhere rather than everywhere. Mandating that new money be distributed evenly would be mandating that monetary policy have no effect, per ¶27. The mechanism cannot be reformed away without abolishing the instrument.
68. The third is the ratchet. Capabilities, once built, are not dismantled by the institutions holding them. An emergency facility created in 2008 becomes a standing tool by 2020. A payment infrastructure built for inclusion retains its capabilities after the inclusion argument has served its purpose. Nothing malicious is required; institutions simply do not surrender instruments, and no one is positioned to make them.
69. The fourth is jurisdictional competition. A country that restricted its monetary authority would find its currency judged against those that did not, immediately and by markets. A country that declined to issue a digital currency while its trading partners did would face a real question about settlement infrastructure. The pressure runs toward adoption in both directions.
70. We should also concede the strongest reform argument, which is that the system has in fact been reformed repeatedly and sometimes well. Inflation targeting was an improvement. Central bank independence, whatever its costs, ended a genuine pathology. Post-crisis capital requirements made the banking system materially safer. The claim that reform is impossible is refuted by the record, and we do not make it. We claim something narrower: that the specific property we object to — unlegislated continuous redistribution with no attributable author — is not among the things reform has addressed or is likely to address, because it is not experienced by anyone as a harm requiring a remedy.
71. There is a deeper obstacle, which we raise reluctantly because it is the kind of argument that cannot be tested. The injury we are describing is not felt as an injury. It is felt as the cost of living, as the housing market, as the way things are. A population subject to an invisible transfer does not demand its cessation, because the transfer has no name in ordinary speech and no counterparty. We cannot exclude the possibility that we are constructing a mechanism to explain a set of grievances that have other causes, and that the mechanism appeals to us because it is elegant. We have tried to state the argument's limits as we go, and we are aware that stating limits is also a rhetorical technique.
72. What would count against our thesis? A sustained period of monetary expansion during which asset prices did not outpace wages. A CBDC deployment in a democratic state that ran for a decade with its programmable capabilities constrained by durable law and never quietly extended. A significant population using the exit instrument as money rather than as an asset. Any of these would undermine what we have argued, and the second and third are observable within twenty years.
CONCLUSION
73. We have not proposed a program, and the reason is not squeamishness. A program requires an agent — someone positioned to execute it — and we can identify none. There is no institution positioned to relinquish the capability, no jurisdiction that could act alone without penalty, and no electorate that experiences the mechanism directly enough to vote on it.
74. What can be recommended is smaller than one would like. A person may hold some fraction of their savings in an instrument that does not depend on the discretion of any authority, taking custody themselves, accepting the operational risk knowingly, and sizing the position so that being wrong about all of this is survivable. We decline to specify the fraction. Anyone who specifies it is selling something.
75. We would add the observation that made ¶56 uncomfortable to write. The property worth having is not the asset but the custody, and these are sold together and understood as one thing. A holder who has never moved funds themselves, never verified anything, and holds through an intermediary has purchased exposure to a price. There is nothing wrong with that. It is simply not an exit, and the two should not be confused by the person doing it.
76. For the reader inclined to dismiss the incumbent system entirely, one caution. It has delivered, in the wealthy currency areas, seventy years without a monetary collapse, deposit insurance that made ordinary savings safe, and crisis responses that in 2008 and 2020 prevented depressions that were genuinely possible. These are not small achievements and they were not achieved by accident. An argument that treats the whole arrangement as a fraud is not a more radical version of ours; it is a worse one, and it will be refuted by events in a way that discredits the part that was correct.
77. One last observation, which we find the most uncomfortable in the essay. The capacity we have been describing is one that a protocol can actually secure — unlike most things worth having, this one has a technical remedy that works. And most holders have declined to take it. Not from ignorance, and not because the remedy failed. They declined because taking it requires doing the difficult part yourself, and an intermediary was available who would do it for you. Whatever is wrong with the incumbent arrangement, it is not the only thing standing between people and an exit.
78. We expect this argument to be read as ungrateful by the specialists and as insufficiently committed by the enthusiasts, and both readings are fair. The monetary system of the past century financed real prosperity and prevented real catastrophes. It also transfers wealth continuously, without a vote, to those nearest its point of issue, and is now acquiring the technical means to do so with precision. Both of these are true at once. They are not opposites; they are the same arrangement, described by the people it reaches first and by the people it reaches last.