There is a word for an economic system in which meaningful productive activity requires state permission at every stage, where that permission is granted primarily to incumbents, and where the regulatory apparatus was largely written by the industries it purports to regulate. The word is not capitalism. I do not especially care about capitalism, either.
Do I need to go here? I wish I did not need to go here, but I have come to the conclusion that one thing the American Right does not have is a knowledge-retention mechanism. We reinvent the wheel every two or three years for reasons I do not even want to get into here. But all of this “Patriotards” vs. “Hardboners” nonsense has reminded me that nobody remembers even the first principles.
The claim is often dismissed as rhetorical excess. “We live in a communist country” is the kind of thing someone says when he’s angry about property taxes or health insurance premiums, and it usually functions as an expression of frustration rather than a description of the economy. Fine. Let us investigate the claim anyway. I want to press it to its structural conclusion, because I think that the conclusion is eerily real, and because seeing it clearly changes how productive action is understood. Plus, it never hurts us to “get back to the basics.”
Our crucible is simple. Pick any productive sector outside of software, meaning any sector that requires physical inputs, material processes, capital equipment, or in Western alchemical terms the “transmutation” of matter, and try to enter it as a new actor. Not at the hobbyist level. Cringewalker can make biofuel bacteria with his mail-in kit and his Claude Code $20 plan for all eternity without denting the “enterprise” market. I am talking about entering an industry sector at the level of a business that employs people at meaningful wages and generates sufficient return to justify shouldering the risk on its own terms. Try this in finance. Try it in food production. Try it in manufacturing. Come back and tell us what you find.
All right, let’s drill down.
The banking question is maybe the easiest place to start, because the “numbers” tell the story most cleanly. From 2000 through 2009, American regulators chartered an average of 132 new banks per year, more than 1,300 in ten years. By early 2024, only 84 new banks had been chartered since 2010. Total. Roughly six per year, and there were stretches in there where the number flatlined at zero. The system did not stop producing new banks because Americans stopped wanting to get into banking. If anything, the opposite is true. It stopped because the barriers to entry were already severe before 2008 and became near-absolute afterward.1
Securing a banking charter in the United States now means clearing a serious threshold before a single transaction has been done, before there is one depositor, before a single account has been opened. Somewhere between $10 million and $30 million in paid-in capital is generally required, depending on the type of charter being sought and the regulators involved: the OCC for a national bank, a state banking commission for a state charter, the Fed for access to its systems, the FDIC for insured deposits, and insured deposits are the whole point. These are not neat sequential steps that can be optimized. They are overlapping approval processes run by separate agencies with separate priorities and separate clocks. The FDIC alone has a de novo handbook running to dozens of pages, complete with a section on how to demonstrate that a leadership team has sufficient banking experience to receive the charter that would give it banking experience.2
Under all of this sits the Basel framework for capital adequacy: international banking standards written by a committee drawn from the jurisdictions that already dominate global finance, implemented domestically by regulators who circulate through that same world, and structured so that larger institutions can hold capital against risk more efficiently than smaller ones. Scale is not an accidental byproduct of the compliance architecture. It is one of the things the architecture rewards.3
If the charter problem is routed around, if the attempt is to build something that moves money without technically being a bank, a different wall appears. Payment services require money transmitter licenses. There is no federal money transmitter license. There are 53 jurisdictions with separate licensing regimes: 50 states, Washington, D.C., Puerto Rico, Guam. And 49 of the 50 states require a license; Montana is the lone holdout. The fixed costs of obtaining licenses across all 53 jurisdictions, just the state fees, the surety bonds, the applications, run between $250,000 and $350,000. Add legal counsel, compliance infrastructure, and the annual maintenance cost of holding those licenses, around $225,000 per year before transaction volume starts driving up bond requirements, and the figure is well over $1 million before a single dollar has been moved for a single customer. The compliance regime is the business model: for the law firms and compliance vendors who service it, a stream of fees; for everyone else, a wall.4
The fintech “solution” to all of this, presented with great enthusiasm at every conference and in every pitch deck, is: partner with a bank. Rent the charter. Build the product on top of someone else’s regulatory house. This is accurate, as far as it goes. Structurally, what it means is that every genuinely innovative entrant in financial services is a permanent tenant, dependent on an incumbent for the license that lets him operate, paying rent in the form of fees and data, and subject to eviction if the incumbent decides that the product is now something that it would prefer to build itself. The fintech “revolution” — nobody mention this — is a revolution that happens inside a house owned by Chase.5
Food production has a different texture to it, and in some ways a more visceral one, because what we are talking about is whether a person can raise an animal, slaughter it, and sell it to another person who wants to eat it. The answer, at any meaningful scale, is: not really.
The mechanism here is continuous federal inspection. Any facility that processes poultry or red meat for commercial sale, not for subsistence or personal use, requires a USDA inspector to be physically present during operating hours in the parts of the process that legally require inspection. This sounds like a reasonable food-safety measure. In practice, it means that the economics of small-scale meat processing are heavily controlled by whether federal inspection can be attracted, retained, scheduled, and absorbed at a throughput small plants often cannot make pay. As a result, the number of federally inspected livestock processing facilities declined sharply across the second half of the 20th century, and what we are left with is a processing infrastructure concentrated almost entirely in the hands of a few enormous operators: as of 2022, 12 federally inspected plants produced slightly under half of the country’s beef supply; 14 plants produced roughly 60% of its pork; 4 companies — Tyson, JBS, Cargill, National Beef — control approximately 85% of beef processing.6
Only 12 buildings produce half of America’s beef. Nothing like “competition” in the imagined sense produces this kind of consolidation. What produces it is a regulatory regime selecting for scale, then incumbents lobbying to preserve the conditions that favor them. The Biden administration, to its credit, recognized this and announced roughly $1 billion in programs meant to rebuild small and mid-size processing capacity. The program was contested, slow to disburse, and operated against the headwind of the very incumbents whose position depends on the infrastructure remaining what it is. You do not fix a moat with a bucket.7
If food is produced and sold without going through that processing apparatus, through cottage-food, direct-to-consumer, or farmers-market channels, most states will allow it up to a revenue cap that very often still lands in the low five figures. Subsistence, not a business. The moment the threshold is crossed that would make the operation worth doing, the moment it would support a hire, a building, a future, the compliance regime designed by the people already occupying the shelf space arrives. The moment of crossing that threshold is the moment the cartel reasserts itself. Smallness is permitted. Growth is not.8
Manufacturing is the sector where the regulatory problem compounds most visibly with the trade problem, and where the compounding has been most catastrophic for actual American communities. Between 2000 and 2010, the United States lost nearly 6 million manufacturing jobs, one of the largest and fastest sectoral collapses in the history of a developed economy. The “China shock” literature in economics estimates that Chinese import competition accounts for somewhere between one-quarter and one-third of those losses, which means the rest came from somewhere else: automation, yes, but also the accumulated weight of a compliance environment that made domestic manufacturing structurally more expensive than its foreign competitors, who did not operate under the same rules.9
On CEQ’s own government-wide data for the 2010s, the environmental permitting process for a new manufacturing facility involving emissions, chemical processes, or effluent discharge averaged 4.5 years from initiation to completion. The environmental impact statement itself averaged over 600 pages. In transportation and infrastructure work with a federal hook, agency-specific averages pushed past 7 years, and some reviews took a decade or more. Carrying costs, legal fees, and compliance staff accumulate for 4 to 7 years before the facility has produced a single unit of output. The competitor in Guangdong does not carry the same burden.10
This is the regulatory asymmetry sitting underneath the job-loss numbers. Chinese labor was cheaper, in the early 2000s dramatically so. But the total cost of operating in the United States, compliance costs, permitting timelines, environmental liability, land-use restrictions, OSHA requirements, and the labor premium made the landed cost of Chinese goods competitive even accounting for shipping, tariffs, and supply-chain risk. Chinese wages did not have to be matched in order to be undercut. It was enough to face a regulatory burden the competitor did not face.
Zoning is where this gets operationally concrete. Light industrial and heavy industrial land classifications, the land on which manufacturing actually happens, are a shrinking and politically contested category in many American municipalities. Real-estate pressure pushes that land toward logistics, retail, institutional, or residential conversion. Local governments like tax revenue that arrives without too much noise, smell, truck traffic, particulate, or organized neighborhood resistance. A manufacturing plant rarely offers them that. The result is that industrial land in many metro areas is being converted away at the margin, while genuinely new industrial land is designated only fitfully, if at all.11
What has been eliminated, specifically, is the middle. Large, legacy manufacturing operations with capital sunk over decades can absorb the compliance overhead. Very small artisan producers operate below the thresholds where the heaviest burden kicks in. What has been made essentially impossible is the factory employing 50 to 200 people, making something durable at a price that can compete with imports, selling it domestically. That entity, the backbone of American industrial capacity for most of the 20th century and the institutional base of the American working class, is not merely struggling; it has been progressively excluded from the possibility of existence by the intersection of compliance costs, land-use attrition, and regulatory asymmetry with producers operating outside the same regime. That this is treated as a natural fact of economic life, rather than a policy result, is itself a measure of how completely the apparatus has colonized the imagination of the people who write about it.
Software, at this point, appears to refute everything I have just said. In a narrow sense, it does. A software company can be started with a laptop, a credit card, and a willingness to forego sleep. The barriers to entry are near zero. Capital requirements are effectively zero for a first product. There is no EPA permit for a mobile application. The FDA does not inspect a codebase. OSHA does not care about a standup meeting. This is all true.
It is true because software occupied, for several decades, a regulatory gap. The product is not a “good” in the legal sense that triggers most commercial regulation. It produces no physical emissions. It does not employ workers in environments where they can be injured by machinery. It does not process food. It did not initially move money in the ways the existing financial-regulatory regime had been built to supervise. Software was, for a while, the one productive sector where the full weight of the apparatus had not yet landed.
That frontier is being fenced. GDPR established that software handling personal data is subject to comprehensive regulatory oversight. The SEC’s treatment of crypto established that software moving value is subject to financial regulation. The FDA’s guidance on software as a medical device established that software making clinical decisions requires certification. AI legislation in Europe is already law, and multiple jurisdictions are moving toward treating AI systems as regulated products. The apparatus is not slow to recognize an escape route. Give it a productive sector that slipped past the first time, and it will close the gap.12
More subtly, the winner-takes-all dynamics of software markets mean that what looks like dynamism is often a rapid consolidation process. The software frontier produced Google, Amazon, Apple, Meta, and Microsoft as the de facto infrastructure of the digital economy within roughly 30 years. New software companies now largely operate within or beneath those five platforms, paying rent to them in the form of app-store fees, cloud margins, and advertising dependency. The frontier produced a new oligopoly faster than any prior sector. When the regulatory apparatus finishes closing around it, it will find the incumbents already entrenched and privately relieved.13
So, what is the name for this? The standard communist economy features state ownership of the means of production. Private property in the United States is real. The owner of a manufacturing plant owns it. The holder of a bank charter owns it. No state official comes to take the factory.
What the United States has instead is something more sophisticated and more durable: state control of entry into productive activity, combined with private capture of the profits once entry is permitted. The regulatory apparatus functions not as a production system but as a moat. It was designed, is maintained, and is periodically deepened by the incumbents it ostensibly regulates. They understood this early, which is why they supported the regulations, and they knew that the most durable competitive advantage is not innovation or efficiency but the exclusion of new competitors by administrative fiat. You do not need to be the best bank if no new banks are being chartered. You do not need to make better beef if the processing infrastructure required to bring a competitor’s beef to market does not exist.
Call it cartelism: licensed private monopoly, in which the state provides the enforcement mechanism and the incumbents provide the political support to sustain it. The vocabulary of capitalism is preserved: private property, free enterprise, competition, market. The substance has been replaced. What can be called a free market is not what functions as one. The entrepreneur who actually tries to build at meaningful scale discovers that every productive sector requires prior permission from a regulatory apparatus structurally disposed against granting it.
The comparison to communism holds in the one place that matters. In a communist economy, the state controls who produces what. In a cartel economy, incumbents control who competes with them, using the state as their instrument. The outcome is the same: entry into meaningful productive activity is controlled, and the control is exercised according to political rather than economic logic. The difference — and this is the part that should make a man angrier, not less — is that in communism the state captures the surplus. In cartelism, private incumbents capture the surplus while the coercive apparatus is provided at public expense. The losses are socialized. The gains are private. The barrier is maintained by everyone’s government on behalf of a small number of existing actors. This is, if pressed, the worse arrangement.
In 1971, George Stigler published “The Theory of Economic Regulation,” a short paper, the kind that ages well, with a thesis that was understood at the time as provocative and has since been confirmed so thoroughly that it now functions as baseline description rather than hypothesis. The thesis: regulation is not designed by the public to protect itself from industry; regulation is acquired by industry, designed by industry, and operated primarily for industry’s benefit. The regulatory agency begins as a check on the sector it governs and ends as its instrument. Stigler called this regulatory capture, won a Nobel Prize partly on the strength of that line of work, and then got to watch the next 50 years make his point for him.14
Mancur Olson, a decade later, took the argument further and wider, as seen in his thesis in The Rise and Decline of Nations (1982): stable societies accumulate what he called distributional coalitions, organized interest groups, industry associations, licensing boards, professional guilds, regulatory agencies populated by industry veterans, that are much better at capturing policy than they are at creating economic value. These coalitions exist to protect rents, which they do by erecting barriers that exclude new competitors, and they compound over time. The older and more stable the society, the more thoroughly its productive apparatus has been colonized by entities whose primary function is to prevent competition rather than conduct it. He called the result institutional sclerosis. His standard comparisons were postwar Germany and Japan, whose distributional coalitions had been shattered by military defeat, against countries like the United States and Britain, whose coalitions survived intact and kept accumulating weight.15
The United States has been stable for 80 years. We have had no structural disruption sufficient to clear the accumulated coalitions. What Olson predicted for societies in our position is very close to what we have. The financial industry does not need to compete for new charters because the chartering process is administered through institutions saturated with the financial industry’s own expertise, personnel, and assumptions. The four meat-processing conglomerates do not need to compete with small processors because the inspection and throughput infrastructure that would permit small processors to operate at scale has been allowed to wither under rules the large firms can absorb. The permitting timeline for a new manufacturing facility is not just bureaucratic sloth. It is also protection for the capital value of what already exists. Stigler’s capture and Olson’s sclerosis are the same process at different timescales, not two separate phenomena. Capture is what happens when a regulatory agency is colonized by a single industry over the course of a generation. Sclerosis is what happens when enough agencies have been captured over enough generations that the economy as a whole has lost the capacity for structural renewal. By the charter numbers and the processing numbers and the permitting timelines and the manufacturing job losses, we are well out toward the sclerotic end of that process.16
Which leaves a question almost more disorienting than the diagnosis itself: Why does no one act like it? The vocabulary of American entrepreneurial capitalism, the small manufacturer, the local bank, the family farm, the man who built something from nothing, remains entirely intact. Politicians of both parties campaign on it. Business schools teach it. The mythology of the self-made man is, if anything, more aggressively promoted now than it was 50 years ago, exactly as the structural conditions that once made it plausible were being dismantled. Koselleck’s distinction between Space of Experience and Horizon of Expectation applies here in a more concrete register. When the Space of Experience no longer reliably generates a Horizon of Expectation, people substitute the memory of what was once possible for a genuine account of what is actually available. The expectation that a meaningful business can still be started in finance, food, or manufacturing is a relic, not a plan: the memory of a productive environment the regulatory apparatus has since foreclosed, preserved in the culture because the mythology is politically useful and the admission that it has become mythology is politically costly. The map is still being distributed. The territory it describes was rezoned decades ago.
The standard political response to this diagnosis, when it is made at all, is to propose deregulation. The regulatory apparatus is an ecosystem, not a garden to be pruned back by politicians with the right inclinations. The incumbents who depend on it have the resources, the political relationships, and the regulatory expertise to ensure that deregulatory efforts, when they succeed at all, succeed in ways that benefit the existing large players rather than enabling new entrants. You do not dismantle a moat by changing the sign on the gate.
What this means practically is specific. Any attempt to employ 50 people making something physical is going to run into every wall described above. Any attempt to build a bank, or a food-processing operation, or a manufacturing facility with an industrial footprint runs into entry requirements calibrated to keep new actors out. Terrain information, not despair. A mined field is not crossed because it ought to be crossable. The mines are mapped, and action is taken accordingly.
What is outside the effective reach of the cartel state is, specifically, what cannot be registered, inspected, licensed, or commodified in the standard sense. Relationships. Accumulated judgment about who can be trusted to do what under pressure. The capacity to organize effort around shared interests without going through a regulated structure to do it. The intellectual and cultural production that moves in registers the compliance apparatus was not designed to track: not because those things are economically marginal, but because the instruments of control are calibrated for physical goods, financial flows, and registered entities. What the club produces is none of those things. What it produces does not appear on a balance sheet and does not require a charter. It is, for exactly that reason, the one productive structure in this analysis that the cartel cannot capture by making the minimum capital requirement prohibitive.
In a landscape where meaningful productive entry has been foreclosed sector by sector, what accumulates freely is the kind of production that requires only judgment, trust, and repeated contact. The club form emerges here as a nucleus for exactly that reason. The economic terrain confirms it from the outside: the cartel state cannot capture what does not require a charter, cannot exclude what does not require a license, cannot foreclose what compounds through repeated contact rather than registered capital. The conditions make this more valuable than it appears.
Look what time it is.
Charter counts: in 2024 comments on EGRPRA, the FDIC reported an average of 132 de novo banks per year from 2000 to 2009. Banking Strategist’s FDIC-based count, updated through Q1 2024, shows 84 de novos opened since 2010. The FDIC’s 2020 Community Banking Study also notes several post-crisis years with no de novo formation at all. The causal sentence in the body is interpretive, but the collapse in entry is not.
Primary sources are the FDIC’s Applying for Deposit Insurance: A Handbook for Organizers of De Novo Institutions, the OCC’s licensing materials, and state charter guidance such as New York DFS’s Information and Procedure for Charter of a Commercial Bank. None of these publishes a single national dollar minimum. What they do publish is a case-by-case capital adequacy standard, heightened leverage expectations, extensive business-plan review, background investigations, and demonstrated managerial competence. Contemporary organizers, advisers, and trade sources usually discuss required startup capital in the low eight figures, often $20 million or more, sometimes considerably higher in major metro markets. The dollar range in the body should therefore be read as order-of-magnitude rather than as a statutory schedule.
Basel Committee membership is public: 45 member institutions across 28 jurisdictions. In the United States, the Fed’s advanced-approaches capital framework applies only to large, internationally active banking organizations and uses internal ratings-based methods unavailable to ordinary entrants. For the revolving-door dimension, see Lucca, Seru, and Trebbi, The Revolving Door and Worker Flows in Banking Regulation (Federal Reserve Bank of New York Staff Report No. 678, 2014). The last sentence of the paragraph is a synthesis, but not an extravagant one.
CSBS and state regulator materials confirm there is no federal money transmitter license and that Montana is the lone state-level holdout. Practitioners commonly model a national launch as a 53-jurisdiction problem once D.C., Puerto Rico, and certain territories are added. On cost, see Brico’s 2025 50-state fee model for direct licensing outlay ($250,000–$350,000 before counsel) and PAG Law/Astraea Counsel for full first-year estimates well above $1 million once legal, bond, compliance, and build costs are included. The exact totals vary by business model, which is the point.
On the sponsor-bank model, see the 2023 interagency guidance on third-party relationships, the 2024 joint statement on bank arrangements with third parties to deliver deposit products, and the FDIC’s consumer guidance on accounts opened through nonbank apps. “Charter rental” is polemical. The dependence structure is not.
Primary law first: 9 CFR 307.4 and parallel FSIS rules provide that no operations requiring inspection may occur except under the supervision of a program employee. On concentration, USDA ERS reports the top four firms handling 85% of steer and heifer purchases and 67% of hog purchases; Investigate Midwest’s 2022 USDA-data analysis found that 12 federally inspected plants produced just under half of U.S. beef supply and 14 plants about 60% of pork. The paragraph compresses slaughter, packing, and processing into a single choke-point picture because that is how the market structure is experienced on the ground.
USDA’s 2021–2024 competition agenda and Rural Development materials describe roughly $1 billion committed across meat and poultry processing expansion efforts. USDA ERS has also been blunt for years about the throughput problems of small processors, uneven seasonal demand, and the difficulty of keeping expensive labor and equipment fully utilized at low volumes. “You do not fix a moat with a bucket” is mine.
Cottage-food law is a moving target, which means the archive here is statutory and unstable. Institute for Justice’s Baking Bad and related Food Freedom updates show that nearly every state now allows some home-food sales, but many still impose product limits, venue limits, and annual sales caps; among states with caps, half are at or under $50,000. The sentence in the body should be read as a description of the low-five-figure ceiling that still characterizes much of the regime, not as a claim that every state sits on the same number.
BLS puts the 2000–2010 manufacturing employment drop at almost 6 million jobs. On the China shock, Autor-Dorn-Hanson and later summaries by MIT, Brookings, and CSIS place direct manufacturing losses from Chinese import competition in the neighborhood of 1 to 1.5 million, with broader economy-wide estimates around 2 million or more. Saying one-quarter to one-third of the decade’s factory losses is a fair compression of that literature, not a quotation from any single table.
CEQ’s 2018 and 2020 government-wide EIS reports found average completion times of 4.5 years and average final-EIS length of 669 pages across 2010s-era data. CEQ’s January 13, 2025, update changed methodology to NOI-to-final-EIS and showed faster recent medians for 2021–2024, which is worth noting rather than hiding. For transportation-specific contrast, DOT’s Office of Inspector General found FAA environmental impact statements averaging just over 7 years. The body paragraph is therefore describing the older baseline that shaped the complaint, not pretending nothing moved after 2020.
This is one of the places where the evidence is mostly metropolitan rather than national. See the Los Angeles Industrial Development Policy Initiative’s finding that industrial-zoned land has been converted to institutional, retail, and residential uses; MAPC’s Greater Boston industrial-land analysis on conversion pressure toward commercial and residential uses; and regional freight/industrial studies in Atlanta and Oregon showing the squeeze on industrial land. The sentence about local political incentives is a synthesis of that planning literature.
Primary sources: the GDPR; SEC staff statements and guidance treating certain crypto-asset interfaces and transactions as subject to securities-law obligations; FDA’s Clinical Decision Support Software guidance and broader SaMD materials; and the EU AI Act. The frontier-closing argument is interpretive. The encirclement is not hypothetical anymore.
On platform dependence, see the European Commission’s Digital Markets Act gatekeeper designations for Alphabet, Amazon, Apple, Meta, and Microsoft; Apple’s App Store commission schedules; Google Play’s service-fee schedules; and current cloud-market data showing AWS, Azure, and Google Cloud still holding a dominant combined share. “Oligopoly” here is descriptive, not an adjudicated antitrust label.
George J. Stigler, “The Theory of Economic Regulation,” Bell Journal of Economics and Management Science 2, no. 1 (1971). On the Nobel: the Royal Swedish Academy’s 1982 prize materials explicitly cite his work on the causes and effects of public regulation.
Mancur Olson, The Rise and Decline of Nations (1982). The language of distributional coalitions, rent protection, and institutional sclerosis comes from that book’s central argument, as do the comparative references to postwar Germany and Japan.
This paragraph is the essay’s application layer: Stigler’s capture, Olson’s sclerosis, the New York Fed’s revolving-door evidence, USDA’s concentration data, and CEQ’s timelines are being read together. Some of the sentences are interpretive extrapolation, which is not the same thing as invention.
If I had to say, it looks like George Stigler won that Nobel Prize for writing out a clearer set of instructions.