Full Text
FROM VALUE ADDED TAX TO VALUE ADDED SUBSIDY: REDESIGNING CONSUMPTION TAXATION FOR DIGITAL PAYMENT SYSTEMS
Value added tax is a consumption tax by legal design but a production burden in practice: producers prepay tax on inputs and wait months for recovery, refund requests attract audits, and compliance costs fall hardest on the smallest producers. Our paper asks whether the credit-invoice architecture, conceived in 1921 and first deployed in France in 1954 to make businesses administer the tax for states unable to observe transactions, remains justified now that digital payments give states direct observation, and asks how removed burdens can become active production support. Because productive capacity compounds through learning effects and supplier networks into economic growth, military endurance, and cultural influence, and because small distributed producers add resilience and innovation, collection machinery is a strategic design problem, not an administrative detail. Our method combines mechanism analysis of how a consumer tax burdens producers; review of empirical evidence on refund delays, registration-threshold bunching, and compliance-cost regressivity; assessment of real-time invoicing programs in Italy, Brazil, and India; and design synthesis of final-point digital collection bundled with constitutional payment-privacy safeguards named Fiscal Secularity. Three successor scenarios, a thin universal transaction levy, a full rate collected only at final consumption, and a hybrid, are evaluated for revenue adequacy, cascading burden, evasion robustness, administration cost, and privacy cost, with tax rates and small-producer exemption thresholds held as named open parameters. Three findings are anticipated: digitized enforcement without redesign deepens regressive burdens; final-consumption collection secured by digital identity defeats the fake-business-purchase fraud that undermined retail sales taxes; and abolished compliance machinery releases resources near two percent of gross domestic product for capability subsidies in machinery, training, standardization, and small producers. Our paper recommends taxing final consumption only, subsidizing productive capability, and sealing transaction records behind judicial gates, so that taxation remains infrastructure serving production and collection efficiency is never purchased with citizen surveillance.
INTRODUCTION
Every economy contains two layers. The productive layer grows food, builds houses, makes machines, produces energy, and cares for people. The instrumental layer consists of money, credit, taxation, accounting, payment systems, and marketplaces, which exist to coordinate the productive layer. Our paper rests on one axiom: production is the terminal value, and an instrumental tool is justified exactly to the degree that the tool serves production. The axiom has a long lineage. Aristotle separated oikonomia, the science of household provisioning, from chrematistics, the acquisitive art, and further separated natural acquisition, which has a limit set by need, from lucrative acquisition aimed at profit for its own sake. Karl Polanyi called the distinction "probably the most prophetic pointer ever made in the realm of the social sciences". Modern scholarship continues to develop the theme (Leshem 2016; Dierksmeier and Pirson 2009). The design standard that follows from the axiom is simple to state: the instrumental layer should resemble good plumbing, cheap, invisible, and unable to mistake itself for the point of the building.
Against that standard, our paper examines value added tax, the dominant consumption tax of the modern world. Our paper never makes the naive claim that value added tax legally taxes production; the statutory design assigns the burden to final consumers. Our claim is narrower and better supported: value added tax is a consumption tax by legal design and a production burden by administration. Inside the collection machinery, the producer holds two unpaid jobs. The producer acts as unpaid banker, financing the state through prepaid input taxes, and as unpaid inspector, policing suppliers through the invoice chain.
Two research questions follow. First, does the credit-invoice collection architecture, conceived in 1921 and first deployed nationally in 1954 so that businesses would administer the tax on behalf of states that could not observe transactions, remain justified once digital payment infrastructure gives states direct observation of transactions? Second, if the architecture is obsolete, how should a successor design convert the removed administrative burden into active support for production, the conversion our title names as the move from value added tax to value added subsidy?
Because the venue of our research paper submission is the Advanced Design Conference, one framing sentence is warranted. Tax collection systems are designed artifacts, built under the informational and technological constraints of their era, and artifacts built under vanished constraints deserve redesign rather than reverence. Our method is accordingly a design method: constraint analysis of the original architecture, mechanism analysis of the current burden, requirement analysis for the productive layer, comparative scenario synthesis for the successor, safeguard co-design for the privacy problem, and conversion of the savings into capability investment, followed by a defined evaluation program with named open parameters and closing design recommendations.
THE ARCHITECTURE OF 1954 AND THE CONSTRAINT THAT PRODUCED IT
The idea of taxing value added has two independent parents. Wilhelm von Siemens argued before the Reichstag in 1918 that Germany should replace its cascading turnover tax, which taxed the full sale price at every stage and therefore punished long supply chains while rewarding vertical integration. Thomas S. Adams, an American economist, proposed a business tax with credits in 1921, and the Adams version carried the specific mechanism that matters for our paper: invoice documentation to substantiate credit claims, creating an audit trail and distributing collection responsibilities across the supply chain (Mehrotra 2022). The German conception relied on direct subtraction of input values; the Adams conception subtracted input tax from output tax through invoices. Neither proposal was adopted at the time.
Implementation waited three decades. The American state of Michigan operated a business activities tax from 1953 to 1967, a form of value added tax computed by the subtraction method rather than through invoices. The modern credit-invoice mechanism arrived with Maurice Lauré, deputy director of the French tax authority, whose law of April 10, 1954 replaced a cascade of turnover and production taxes. The 1954 French tax initially reached only large businesses and the wholesale level. Denmark became the first European country to extend a value added tax to the retail level in July 1967, and France and Germany followed in January 1968, the year the French system was generalized across the whole economy. The European directives of 1967 made the model the common standard, and more than 170 countries eventually adopted a version, with the United States remaining the notable exception among wealthy economies.
The reason for the rapid adoption is the heart of our diagnosis. The Lauré design spread precisely because the design forced taxpayers at every level of the production process to administer and account for the tax themselves, rather than requiring assessment by tax authorities. The mechanism is elegant. Every business pays tax on purchases and charges tax on sales, reclaiming the difference. Because the buyer needs the seller's invoice to claim the credit, every buyer polices every seller, and the state collects the tax in fractions along the chain, so a single vanishing trader cannot cost the state the entire amount. In plain language, the credit-invoice chain is a self-policing surveillance workaround, a brilliant answer to a hard question: how does a state collect a high-rate consumption tax on transactions the state cannot see?
The constraint that justified the workaround has dissolved. Italy routes invoices through a state platform in real time. Brazil and Mexico have operated mandatory electronic invoicing for years. India couples electronic invoices to the tracking of goods in motion. The European Union adopted a digital reporting package in March 2025, with mandatory digital reporting for cross-border business transactions from July 2030 and harmonization of domestic systems by 2035, on the stated grounds that real-time reporting will reduce fraud by up to eleven billion euro per year and compliance costs by more than four billion euro per year. A state that observes transactions directly no longer needs businesses to observe transactions on the state's behalf. Once the informational reason for the chain disappears, the chain's costs, the float, the refund machinery, the filing, and the audits, lose their justification and stand exposed as pure instrumental-layer overhead. The next section measures that overhead.
HOW A CONSUMPTION TAX BECOMES A PRODUCTION BURDEN
A skeptic may object that economic theory settles the incidence question in favor of the consumer, making producer complaints irrelevant. Behavioral public finance has removed that objection. Consumers demonstrably underreact to taxes that are not salient: in a grocery-store field experiment, posting tax-inclusive price tags reduced demand by eight percent, and tax increases hidden inside posted prices change consumption more than equivalent taxes added at the register (Chetty, Looney and Kroft 2009). The theoretical consequence is exactly the license our diagnosis needs: the economic incidence of a tax depends on its statutory incidence, and even policies that induce no behavior change can create efficiency losses. How a tax is administered and experienced is economically real.
Consider first the float, which makes the producer an unpaid banker. Under the credit-invoice system, producers finance input taxes upfront and recover the taxes only after filing, waiting, and often surviving scrutiny. The magnitudes are large. In South Africa, refunds amount to fifty percent of gross value added tax collection, a transfer that must occur at high frequency, often monthly, and delays in refund payments reduce domestic investment, especially by small firms. When the South African revenue authority roughly halved audit rates on refund-claiming returns and sped up processing, investment increased by thirty-one percent and output by twenty-four percent (Brusco, Piek and Velayudhan 2024). Evidence from China matches. More than sixty percent of surveyed companies experienced refund delays, delays compromise firm liquidity, and a ten percent improvement in rebate efficiency raises small and medium enterprise employment by roughly six percent, with financially constrained firms benefiting disproportionately from faster clearance (Lu, Ma, Hu and Xu 2025). The float is a zero-interest loan from producers to the state, and the credit arrangement is regressive, because the state borrows for free from exactly the small producers who themselves borrow at the highest market rates.
Refund friction is institutional rather than accidental, which makes producer fear of claiming rational. Refund requirements and procedures are complex and burdensome, often discouraging taxpayers from claiming legitimate refunds, and taxpayers who do claim frequently experience significant delays or are not paid at all. Many administrations assume taxpayers cannot be trusted, treat the majority of refund claims as fraudulent, and design administrative processes to increase compliance costs and delay payment (International Monetary Fund 2021). In one documented case, the revenue authority explicitly states that first-time refund requesters are automatically subjected to audit. A remedy that punishes the claimant is a deterrent wearing a remedy's clothes.
The unsold-goods asymmetry supplies the clearest single illustration of the gap between design and practice. When goods are produced and never sold, because the goods spoil, fall out of fashion, or find no demand, no consumption ever occurs, yet the producer financed the input tax for the entire period. On paper the credit exists. In practice the credit arrives late, arrives after an audit, or, in non-deductible and exempt categories, never arrives. A tax that is supposed to touch only consumption has, in the failure case, touched only production.
Compliance costs are large and regressive. The literature review finds value added tax compliance costs significant in absolute terms and relative to revenue in developed and developing countries alike, and highly regressive, disproportionately affecting small businesses (Vishnuhadevi 2021). The regressivity has an arresting arithmetic: for a trader at a typical registration threshold, compliance costs of two percent of turnover combined with a twenty percent profit margin are equivalent to a ten percent tax on income, a burden that largely stays with small traders who lack market power (Ebrill, Keen, Bodin and Summers 2001). Aggregate magnitudes confirm the scale. United States taxpayers spend 7.9 billion hours on tax compliance annually, a total cost of 546 billion dollars, nearly two percent of gross domestic product, with most of the burden borne by businesses. European enterprise compliance costs average 1.9 percent of turnover, and Canadian estimates reported in the same study run between 1.2 and 1.8 percent of gross domestic product.
The machinery visibly brakes small-firm growth. United Kingdom administrative records show bunching in annual turnover just below the registration threshold, and many firms with positive growth deliberately hold sales under the line (Liu, Lockwood, Almunia and Tam 2021). Thai tax returns show the same pattern, and the authors conclude that the threshold acts as "a brake on small business growth", with non-registered firms growing significantly slower than comparable registered firms (Muthitacharoen and co-authors 2021). The threshold theory was formalized by Keen and Mintz (2004), and Japanese evidence adds that bunching intensifies exactly where compliance costs are higher. A tax architecture under which staying small is a rational business strategy prunes the productive layer at the root.
There is also a layer asymmetry: the instrumental layer largely escapes the machine built in its name. European Union law requires member states to exempt the transactions listed in Article 135 of the value added tax directive, a diverse range of which are associated with money and finance, because the margin-based nature of financial services makes the computation complex and potentially unworkable. Exempt suppliers cannot recover input taxes, and the known result is that financial services sold to consumers end up underpriced relative to full taxation (Mirrlees Review 2011). The architecture therefore concedes, in law, an inability to handle the instrumental layer's own core activity. The bakery finances the float while much of banking sits outside the machine.
Administrative burden also consumes something beyond money. Scarcity overloads mental bandwidth, and the strain on cognitive resources degrades job performance, decision-making, and self-control; scarcity reduces not a person's inherent capacity but how much of that capacity is currently available for use (Mullainathan and Shafir 2013; Mani, Mullainathan, Shafir and Zhao 2013). Applied to the present setting, every hour a small producer spends navigating input credits is an hour of managerial cognition diverted from product, process, and customer. Our paper records one honesty clause here: a later replication project reproduced most but not all of the foundational results in the scarcity literature (Shah, Mullainathan and Shafir 2019), so we treat the bandwidth mechanism as strong but contested at the edges.
Finally, the tax hides. The hypothesis that taxpayers systematically underestimate indirect taxes concealed inside prices runs from Mill (1848) through Puviani (1903) to Buchanan (1967), and carries empirical support: governments relying more on invisible indirect taxes have sustained higher spending, other things equal. A democracy is healthier when citizens can see what government costs. Any successor design should therefore make the levy visible on every transaction record, a requirement our design adopts below.
WHY PRODUCTION DESERVES PRIORITY
The mechanisms above would matter little if production were one sector among many. Our paper argues that productive capacity is the compounding root of three forms of national strength, so burdens on producers propagate into weakness everywhere.
The economic chain rests on the fact that production teaches. Wright's study of aircraft manufacturing gave the first academic statement of the learning curve, the stable relationship between cumulative output and falling unit cost (Wright 1936), and Arrow built early endogenous growth theory on the observation that labor hours per airframe decline as a precise function of cumulative production, a relation reliable enough to anchor United States Air Force planning (Arrow 1962). The modern demonstration is renewable energy, where every doubling of cumulative solar panel installation has been associated with roughly a twenty percent cost decline. Beyond the single firm, capabilities pool into what Pisano and Shih call the industrial commons, the collective operational capabilities embodied in workforces, suppliers, competitors, and universities that underpin new product and process development. Their warning is that declining manufacturing capability produces a corresponding loss of innovation, and that a lost industrial commons is nearly impossible to retrieve (Pisano and Shih 2009, 2012). At national scale, the economic complexity literature quantifies the thesis: the Economic Complexity Index, a measure of productive capabilities, predicts growth in income per person more accurately than conventional measures of governance, competitiveness, and educational attainment, including after statistical controls (Hidalgo and Hausmann 2009). The same literature names the failure mode, premature deindustrialization, in which economies that never build manufacturing depth suffer lower growth later. Production also spreads prosperity locally: each additional tradable manufacturing job has been estimated to create roughly 1.6 additional local service jobs, and high-end manufacturing categories show multipliers near five (Moretti 2010), although our paper cites the honest range, since a replication using an alternative instrument found closer to 0.8 additional jobs per manufacturing job (van Dijk 2015).
The military chain follows from a simple law: in extended conflict, military power equals production rate versus attrition rate. Defense institutions state the dependence directly, arguing that security requires a strong military, which is possible only if the underlying economic infrastructure is in place, and that national security rests on industrial capability as much as on military capability. The historical benchmark is stark. The wartime United States produced seventeen aircraft carriers, three hundred thousand airplanes, and roughly fifty thousand tanks in under four years, against a maximum modern production capacity of approximately one hundred fifty F-35 fighters per year. The erosion is documented: United States domestic semiconductor production capacity fell from thirty-seven to twelve percent of the global total, and production struggles now span artillery shells, drones, and submarines, with supporting sectors such as key machine tools essentially nonexistent domestically. The war in Ukraine supplied the live demonstration. Both armies fired ten thousand to twenty thousand artillery rounds on single days; stockpiles were depleted or neared exhaustion, transferring battlefield pressure directly onto the industrial base; Russian shell production reportedly rose from one million per year toward a target of four and a half million, while Western countries struggled to keep pace, exposing the fragility of peacetime production. Two details connect the military chain to the rest of our design. A major defense manufacturer entered talks to convert an automotive plant to defense production, which shows that civilian factories are the convertible reserve of wartime economies, and military analysts now recommend standardizing not only technical specifications but ballistic characteristics of artillery ammunition to achieve true interoperability, an independent restatement of the standardization principle our broader research program advocates for civilian production. One caution accompanies the whole chain: the sources in the present paragraphs are government reports, policy institutes, and defense trade analysis, documentary evidence rather than peer-reviewed economics, and the full research program pairs them with the economic history of wartime mobilization. The connection to tax design is direct. The float penalizes capital-intensive factory building hardest, because input tax on machinery is financed upfront and recovered slowly, and compliance overhead suppresses the small workshop tier that gives an industrial base redundancy and convertibility.
The cultural chain treats products as ambassadors. Nye defined soft power as the ability to attract and persuade, arising from the attractiveness of a country's culture, ideals, and policies rather than from coercion (Nye 1990, 2004). Manufactured goods are among culture's most persistent carriers, and the effect is measurable in prices: experimental studies show that country of origin has a positive causal impact on consumer willingness to pay, with consumers paying measurably more for identical products attributed to countries with favorable images (Koschate-Fischer, Diamantopoulos and Oldenkotte 2012). Our paper states the direction of the loop honestly. The experiments measure the path from image to price premium, while our claim adds the prior link, that decades of production excellence build the image in the first place. The two links together mean that every excellent factory pays a dividend to every other producer under the same flag. A country that only consumes imports culture with every container; a country that produces exports culture with every container. The tax connection is immediate, because the small ateliers, workshops, and designers who carry a nation's design culture are precisely the firms on which fixed compliance costs fall hardest.
All three chains strengthen when production is distributed. Many small producers are harder to destroy than one plant, in war or in supply-chain crisis. Many parallel producers run many parallel experiments, which is how innovation searches its space. Local producers form local capital and local skills. Distribution is engineering redundancy rather than romantic nostalgia, and the evidence above shows that the current machinery selects against exactly the distributed tier.
WHY INACTION IS NOT NEUTRAL
Industrial policy has returned worldwide, with major programs addressing the green transition, geopolitical competition, and supply-chain resilience, and the scholarly assessment has turned: a considerable recent literature provides rigorous evidence on how industrial policies work, improves substantially on the older correlational generation, and on the whole offers a more positive assessment (Juhász, Lane and Rodrik 2024). A state that declines to support domestic production does not receive a neutral market. That state receives a market tilted by foreign treasuries.
The current tax architecture is itself an industrial policy, an accidental one aimed against small producers. The mechanisms documented above tilt the field against long production chains, capital-intensive builders, and small firms, while largely exempting the financial core of the instrumental layer. A state operating such machinery is already picking sides. The honest choice is not between intervention and neutrality but between accidental anti-production policy and deliberate pro-production design.
The obvious objection, that governments should modernize the existing tax rather than replace the architecture, now has an evidence-based answer. Digital compliance tools bolted onto value added tax systems raise substantial revenue yet impose regressive compliance costs that fall most heavily on small firms, and without careful design and sequencing such reforms risk undermining welfare and formalization (International Growth Centre 2025). Electronic invoicing attached to the 1954 chain automates the surveillance while keeping the burden: the state gains real-time sight while the producer keeps the float, the refunds, and the filing. Digitizing the old design is not the reform. Redesigning the collection point is the reform.
Coherence supplies a further argument. Contemporary states subsidize production with one hand and burden production with the other, and both hands carry administrative overhead, one foot on the accelerator and one foot on the brake, with both pedals burning fuel. Agriculture is the established example, permanently subsidized because society needs food and simultaneously taxed and audited. Our paper generalizes the observation to the entire productive layer. A state that stops contradicting a principle the state already holds is not adopting radical policy. That state is achieving coherence.
Self-interest points the same way. Every stream of state revenue is downstream of production, so a state that burdens the productive layer shrinks the base of all future collection. There is also an automation asymmetry: tax administration is bureaucracy, and bureaucracy has no consumers to destroy, so automating collection is the safest possible automation target and attacks one root of permanently rising administrative cost in the sense of Baumol's cost disease, since administration performed by human hours never enjoys manufacturing's productivity growth. On time horizons, productive capability pays back over decades, beyond private investment windows, and the record shows states have long carried that horizon: the state has not merely fixed market failures but actively shaped and created markets, funding the foundational technologies of modern computing and communication (Mazzucato 2013). Mazzucato's further observation, that the present innovation system socializes risks while privatizing rewards, states almost word for word the dysfunction our broader research program identifies in contemporary artificial intelligence economics.
The objection that governments pick winners deserves respect. Skeptics rightly point to the many failures littering development economics, and political incentives, not only market failures, shape which policies are adopted. Our design answers structurally rather than rhetorically. Support is horizontal, directed at capability, standards, machinery, and the small-producer tier, never at named champion firms. Allocation is plural, spread across multiple state banks, private banks, and small local banks with overlapping mandates, so no single allocator's failure is fatal. Verification is algorithmic on the transaction ledger, which knows who genuinely buys inputs, ships goods, and sells to real customers.
THE SUCCESSOR DESIGN SPACE
The history of consumption taxation teaches one law: collection friction is never deleted, only relocated. Under value added tax, friction lives in the producer's cash flow, as float, refunds, and filing. Under the American retail sales tax, friction lives on the retailer's honesty, because the entire levy concentrates at one point. Under naive digital collection, friction moves into the citizen's privacy, because the taxing authority sees every purchase in real time. The design task is not to abolish friction but to relocate friction to where friction does the least damage, and the following section argues that a fourth location exists.
The retail sales tax record deserves precision, because our design must beat that record. Analysts have observed that at rates above twelve percent sales taxes become too easy to evade, and that no historical precedent exists for a high-rate enforceable retail sales tax; replacement-level rates for a national retail sales tax in the United States were estimated at thirty-four to eighty-nine percent on a tax-exclusive basis once evasion and base erosion were considered. Even at Florida's low single-digit rates, one study found five percent of business exemption certificate purchases misused for personal consumption. American state rates cluster in single digits for a reason. The classic killer is the identity question: is a given buyer a business or a consumer? Under paper administration, that question is a fog in which evasion hides. In an all-digital payment system with verified identities, the classification becomes a graph property. Real-time analytics can flag the construction firm whose purchase pattern consists of groceries and television sets, and deliberate misrepresentation is fraud, a category whose consequences the next section defines. Digital enforcement therefore attacks both incumbent designs at their weakest points, provided the privacy cost is answered.
The first successor scenario is a thin universal transaction levy. The lineage is Feige's automated payment transaction tax, which proposed replacing personal and corporate income, sales, excise, capital gains, import and export duty, gift and estate taxes with a single flat tax on all transactions, automatically assessed and collected when transactions settle in the payment system, with lower administrative and compliance costs (Feige 2000). The proposal received national attention in the United States, with a major newspaper describing the design as "fair, simple, and efficient", and Brazil operated a broadly similar transaction tax, reportedly with reasonable success, although layered on top of existing taxes rather than replacing them. Because transaction volumes are a large multiple of gross domestic product, the rate can sit in the low single digits, for illustration a rate consistent with historically persistent flat levies. The thin levy abolishes the reclaim machinery completely: nothing is reclaimed because the small charge simply sticks at each hop, so the float, the refunds, and the audit fear vanish by construction. Our paper openly acknowledges the cost. A turnover levy cascades, so a product passing five taxed stages embeds a cumulative burden on the order of twelve percent while a vertically integrated giant pays once, a structural tilt toward consolidation. The critique is historically grounded, since the cascading German turnover tax of the early twentieth century demonstrably encouraged vertical integration, which is exactly why Siemens proposed reform. Two counterweights answer the tilt: the small-producer exemption described below, and the compliance savings, which for small firms dominate the cascading cost. Financial-sector legs receive separate treatment, a micro-rate for hedging contracts held by parties with verifiable real exposure to the underlying good and the full rate for unanchored churn; the distinction has independent scholarly precedent, since business-ethics research already applies the Aristotelian boundary between natural and unnatural acquisition to assess high-frequency trading, and a companion paper in our research program develops the exposure test in full.
The second scenario collects a full rate only at the final consumption event, identified digitally by the buyer's status, with nothing taxed between businesses. The advantages mirror the defects of value added tax exactly: perfect stage neutrality, which serves distributed domestic supply chains; zero float; zero refunds; zero business-side filing. The risk is the concentration of temptation, because replacing value-added-tax revenue at a single point requires a rate near current European rates, historically the regime where single-point collection collapses into evasion. The second scenario therefore stakes everything on the digital identity argument, a bet our paper judges increasingly plausible but not yet proven at national scale.
The third scenario is a hybrid, a thin universal levy plus a modest final-consumption top-up. The thin component keeps every transaction lightly contributing, while the top-up restores revenue without pushing the final-point rate into the historical collapse zone, and cascading stays mild because the universal component is small. In design terms the hybrid is a tuning dial between the first two scenarios rather than a third philosophy.
The three scenarios are evaluated against five criteria: revenue adequacy, cascading burden, evasion robustness, administration cost, and privacy cost. The anticipated pattern is as follows. On revenue adequacy, the thin levy and the hybrid score high because of the enormous base, while the final-point design is rate-limited by evasion risk. On cascading, the final-point design is perfectly clean, the hybrid nearly clean, and the thin levy carries the acknowledged mild tilt toward integration. On evasion robustness, the thin levy is strongest because no single hop is worth cheating, the hybrid is strong, and the final-point design carries the open bet on digital identity. On administration cost, all three are minimal, because all three abolish the reclaim machinery, which the mechanism analysis identified as the true source of the production burden; the choice among scenarios is secondary to that shared architectural move. On privacy cost, all three share one unpaid bill, total transactional visibility. No scenario survives politically without an answer to the question of who guards the ledger. That question, we argue next, is what killed the Feige proposal, and our program answers the question by construction.
FISCAL SECULARITY
Feige himself noted that automated recording of all payments creates a degree of transparency, transparency of citizens to the state. The automated payment transaction tax was technically sound and politically dead, because a design in which the taxing authority sees every transaction of every citizen is, without safeguards, a surveillance machine with a fiscal feature. The central thesis of our research program follows: the automated transaction tax becomes viable only when bundled with Fiscal Secularity, the constitutional and technical separation of money, taxation, and governance, doing for money what the separation of church and state did for belief. The bundle, not either component alone, is the contribution.
The records architecture is sealed, not secret, with three properties required simultaneously. Records are immutable: no person and no institution can alter, delete, or roll back an entry, because without immutability the access log proves nothing. Records are sealed: not readable without legal cause. Records are conditionally accessible: readable with legal cause through a judicial gate, and every act of access is itself logged, with the log entry, stating who read, when, and for what stated reason, instantly and permanently visible to the citizen whose records were opened. The working precedent is Estonia, where any citizen can log into the state portal and see exactly which officials queried their personal data, when, from which database, and for what purpose, and where unauthorized access to citizen data is a criminal offense punishable by imprisonment, with the visible trace itself acting as a deterrent. The security of the design must never depend on secrecy of design; the architecture is fully public and auditable. Sealed data, transparent design.
The state may act upon transaction records only for theft of funds and fraud, including obtaining funds by deception. For lesser matters, citizens receive de facto privacy: the records exist, but no state action may be built upon them. The action threshold must be constitutionally entrenched and deliberately difficult to amend, because historical exceptions always expand, from terrorism to money laundering to tax to speech-adjacent offenses. The design goal is to prevent threshold creep structurally rather than by goodwill. An evidence rule completes the gate: evidence obtained from transaction records outside the judicial process, including through covert or classified programs, is inadmissible in court. Reading the ledger without cause yields legally useless information, which removes the incentive to peek rather than merely the permission.
Nothing on the ledger is ever technically reversed, and the state has no capability to edit any entry. Corrections of error, theft, and fraud occur as new transactions: a court orders the obligated party to initiate a compensating transaction, and the party complies or faces ordinary physical enforcement. Where buyer and seller both consent, as in the routine refund or the duplicate charge, reversal by new transaction plus automatic tax clawback runs with no human involvement, and dispute resolution is algorithmic in the first instance with human arbitration only on appeal, preventing the dispute system from growing into a new bureaucracy. The principle: reversal is a judicial act performed through people, never an administrative act performed on the ledger. The enforcement monopoly compels humans; the enforcement monopoly never touches the ledger.
Freezing or blocking a citizen's payments must be simultaneously illegal, by constitution and statute, and expensive, requiring a judicial order that is logged, visible to the affected citizen, appealable, and deliberately slow. The motivation is contemporary and concrete: accounts of protesters frozen, individuals debanked, payments blocked over opinions. Money is stored life energy. Citizens who fear that money can be taken for the wrong words self-censor, and a democracy in which citizens fear their own payment system carries a silent veto on speech. Fiscal Secularity exists so that no such veto can be built.
The theoretical core of the design can now be stated. The iron law says friction relocates; Fiscal Secularity does not delete the privacy friction either, but relocates friction to a fourth location, constitutional maintenance, and the relocation is decisive for a reason public finance has always understood: marginal wedges distort behavior, while fixed costs do not. The float scales with every production cycle. Retailer evasion risk scales with every sale and with the rate. Naked surveillance chills at every purchase. The judicial gate, the visible access log, the inadmissibility rule, and the entrenched threshold cost approximately the same to maintain whether the economy performs one million or one billion transactions per year. The bundle therefore converts taxation's friction from a distortion at the margin of every productive act into a fixed institutional overhead, which is the translation into public finance of the opening axiom that the instrumental layer must be thin at the point of contact with production.
Who guards the ledger? The answer is structural, in three parts. Nobody can edit the ledger, because of immutability. Citizens guard the readers, because every reading leaves fingerprints visible to the person read. The constitution guards the thresholds, through entrenchment and inadmissibility. Guarding is mutual and permanent rather than promissory.
Three honest residuals remain and are stated rather than hidden. First, mandatory digital payment is a civilizational precondition, not a software update. Shadow economies averaged 31.9 percent of gross domestic product across 158 countries between 1991 and 2015, with advanced economies below twenty percent (Medina and Schneider 2018), and a payment mandate is legitimate only alongside three guaranteed decencies: decent conversion of assets into the digital currency, decent ability to earn within open marketplaces, and decent ability to spend without foreign or administrative shutdown. Second, the trust bootstrap: the fixed-cost conversion works only to the degree citizens believe the seal, and belief is built by visible structure, logged access and real prosecutions for unauthorized access, not by declarations. Third, seal durability under future regimes: append-only records and public design raise the cost of abuse but cannot make abuse impossible, and the ultimate backstop is constitutional culture. Our paper says so plainly rather than overclaim.
THE VALUE ADDED SUBSIDY
The conversion promised in our title proceeds in two steps, and the fiscal story closes without new money. Step one is the negative subsidy: stop the friction. Abolishing the credit-invoice machinery costs no real resources and releases compliance capacity documented near two percent of gross domestic product. The cheapest production subsidy available is the removed burden. Step two is the positive subsidy: redirect a fraction of the released resources into productive capability. The subsidy is prepaid by the abolished paperwork.
The target must be capability, never gross output, because subsidizing output invites production for the subsidy's sake. Four channels carry the support: machinery, attacking exactly the capital-formation margin the float suppressed; training, the skills component of the industrial commons; standardization, the compatibility and repairability of mechanical and electronic products, whose military value the war evidence documented independently; and the small-producer tier, the resilience and innovation reserve. Verification is algorithmic on the ledger, which knows who buys inputs, ships goods, and sells to real customers, so the subsidy needs no new inspection bureaucracy.
The small-producer exemption is engineered against gaming. Producers employing fewer than ten people and selling into the domestic market pay no tax of any kind, with the exemption tapering smoothly to zero at twenty-five employees. Smooth tapering matters because hard thresholds teach firms to stay small. The French case is the canonical demonstration: administrative data show a sharp fall in the number of firms at exactly fifty employees compared to forty-nine, where labor regulations begin to bind, with regulation costs equivalent to a 2.3 percent labor tax and welfare costs estimated at 3.4 percent of gross domestic product, and the same study analyzes a parallel threshold at ten employees, directly relevant to our boundary (Garicano, Lelarge and Van Reenen 2016). A smooth taper makes splitting a company into fake small pieces gain almost nothing. Where fragmentation is attempted anyway, detection is a graph query rather than an audit, since related entities reveal themselves through shared beneficial owners, shared addresses, and transaction flows circulating among themselves. Whistleblower bounties supplement the graph, on the American model, where the tax whistleblower program has recovered nearly seven billion dollars since 2007 while paying whistleblowers over one billion, with mandatory awards of fifteen to thirty percent of proceeds. Structured evasion through fake fragmentation is classified as fraud, which places the conduct above the judicial action threshold defined earlier. The governing sentence of the whole design: your money is inviolable; lying to the system about who you are is not.
The exempt are watched. Exempt small producers are checked automatically once per year at a randomized moment, and each checked producer sees the check in the personal access log. Randomized timing follows a variable-interval schedule, which behavioral research since Ferster and Skinner (1957) associates with steady, persistent behavior that is highly resistant to extinction. Visible auditing of the untaxed maintains legitimacy for everyone else. Exempt, but watched.
The subsidy half of our title stands in an old line, from Hamilton's Report on Manufactures through List's national system to the contemporary industrial policy revival. What our design adds to the lineage is the funding source, abolished compliance machinery rather than new taxation, and the verification method, the ledger rather than the inspectorate.
ANTICIPATED FINDINGS AND EVALUATION PROGRAM
Three findings are anticipated, stated as testable expectations. First, digitizing enforcement without redesigning the collection point deepens regressive burdens on small firms, as early evidence already indicates. Second, final-consumption collection secured by digital identity defeats the fake-business-purchase fraud that historically limited retail sales taxes, because purchase-pattern classification in a verified-identity ledger is algorithmically tractable. Third, abolishing the compliance machinery releases resources near two percent of gross domestic product, sufficient to fund the capability subsidies without new taxation.
The named open parameters are the universal levy rate; the final-point top-up rate; the exemption boundary and taper slope, illustratively ten and twenty-five employees; the micro-rate for exposure-verified hedging instruments; the refund-latency guarantee for consensual reversals; the randomization interval for exempt-sector checks; and the privacy metrics, including access-log volume per capita, gate-refusal rates, and prosecution rates for unauthorized access.
The evaluation procedure uses five instruments in order of increasing commitment: microsimulation of the three scenarios on national transaction datasets, testing revenue adequacy and distributional incidence; cascading measurement through input-output tables, quantifying the consolidation tilt of the thin levy; adversarial stress-testing of the identity layer, so the final-point bet is red-teamed before being trusted; privacy auditing against the metrics above under simulated hostile administrations; and staged regional or sectoral pilots with pre-registered success criteria, because a tax architecture, like any designed artifact, earns trust through prototypes rather than proclamations.
The limitations of the supporting evidence are recorded without embarrassment. The refund-burden findings are heterogeneous: one careful Honduran study of a withholding reform estimated null average effects on growth and investment, challenging the premise that unrefunded credits bind in all settings (Pineda Pinto, Bermudez and Scot 2024). The null result sharpens rather than weakens our claim, since the burden concentrates on financially constrained small producers, exactly where our design directs relief, and we present the South African, Chinese, and Honduran results together as a heterogeneity finding. The cognitive-bandwidth mechanism carries a replication caveat. The local-multiplier magnitudes are contested within a factor of two. The military-chain sources are documentary rather than peer-reviewed. The distributional profile of flat transaction levies requires modeling rather than assertion; Feige argued that the base itself introduces progressivity, because the wealthy conduct a disproportionate share of transactions, and our microsimulation instrument is designed to test that argument rather than assume the argument. Finally, the cashless precondition and the trust bootstrap are political constructions whose timescale no simulation can compress.
IMPLICATIONS AND DESIGN RECOMMENDATIONS
Seven recommendations follow for treasuries, central banks, and the design community. Treat tax collection architecture as a designable public artifact, subject to prototyping, evaluation, and replacement when founding constraints vanish; the credit-invoice chain answered an observability constraint that no longer exists, and the chain took thirty-three years to travel from concept to national implementation and another fourteen to full deployment, a reminder that redesign should begin before the old design's constraints have been obsolete for a generation. Do not confuse digitizing the old design with redesign, because electronic invoicing bolted onto the chain automates surveillance while keeping the burden. Collect consumption tax at the final consumption event, or through a thin universal levy, or through the hybrid of the two, but in every scenario abolish the reclaim machinery, which is the true engine of the production burden. Make the levy visible on every transaction record, ending the fiscal illusion that hides the cost of the state inside prices. Never deploy total transactional visibility without the constitutional counterpart: sealed-by-default records, a judicial action threshold restricted to theft and fraud and entrenched against creep, citizen-visible access logs, an inadmissibility rule for evidence gathered outside the gate, and an append-only ledger the state can compel people to correct but can never itself rewrite, the doctrine our research program names Fiscal Secularity. Convert the released compliance resources into capability subsidies, for machinery, training, standardization, and the small-producer tier, verified algorithmically on the ledger rather than through a new inspectorate. Judge every element of the design against the single axiom from which our paper began: the instrumental layer of an economy exists to serve the productive layer and should be as thin as possible, cheap, invisible, and unable to mistake itself for the point, because food, housing, machinery, and care are the purpose, and money, taxation, and ledgers are only the plumbing through which a nation's real wealth flows.
Value added tax is a consumption tax by legal design but a production burden in practice: producers prepay tax on inputs and wait months for recovery, refund requests attract audits, and compliance costs fall hardest on the smallest producers. Our paper asks whether the credit-invoice architecture, conceived in 1921 and first deployed in France in 1954 to make businesses administer the tax for states unable to observe transactions, remains justified now that digital payments give states direct observation, and asks how removed burdens can become active production support. Because productive capacity compounds through learning effects and supplier networks into economic growth, military endurance, and cultural influence, and because small distributed producers add resilience and innovation, collection machinery is a strategic design problem, not an administrative detail. Our method combines mechanism analysis of how a consumer tax burdens producers; review of empirical evidence on refund delays, registration-threshold bunching, and compliance-cost regressivity; assessment of real-time invoicing programs in Italy, Brazil, and India; and design synthesis of final-point digital collection bundled with constitutional payment-privacy safeguards named Fiscal Secularity. Three successor scenarios, a thin universal transaction levy, a full rate collected only at final consumption, and a hybrid, are evaluated for revenue adequacy, cascading burden, evasion robustness, administration cost, and privacy cost, with tax rates and small-producer exemption thresholds held as named open parameters. Three findings are anticipated: digitized enforcement without redesign deepens regressive burdens; final-consumption collection secured by digital identity defeats the fake-business-purchase fraud that undermined retail sales taxes; and abolished compliance machinery releases resources near two percent of gross domestic product for capability subsidies in machinery, training, standardization, and small producers. Our paper recommends taxing final consumption only, subsidizing productive capability, and sealing transaction records behind judicial gates, so that taxation remains infrastructure serving production and collection efficiency is never purchased with citizen surveillance.
INTRODUCTION
Every economy contains two layers. The productive layer grows food, builds houses, makes machines, produces energy, and cares for people. The instrumental layer consists of money, credit, taxation, accounting, payment systems, and marketplaces, which exist to coordinate the productive layer. Our paper rests on one axiom: production is the terminal value, and an instrumental tool is justified exactly to the degree that the tool serves production. The axiom has a long lineage. Aristotle separated oikonomia, the science of household provisioning, from chrematistics, the acquisitive art, and further separated natural acquisition, which has a limit set by need, from lucrative acquisition aimed at profit for its own sake. Karl Polanyi called the distinction "probably the most prophetic pointer ever made in the realm of the social sciences". Modern scholarship continues to develop the theme (Leshem 2016; Dierksmeier and Pirson 2009). The design standard that follows from the axiom is simple to state: the instrumental layer should resemble good plumbing, cheap, invisible, and unable to mistake itself for the point of the building.
Against that standard, our paper examines value added tax, the dominant consumption tax of the modern world. Our paper never makes the naive claim that value added tax legally taxes production; the statutory design assigns the burden to final consumers. Our claim is narrower and better supported: value added tax is a consumption tax by legal design and a production burden by administration. Inside the collection machinery, the producer holds two unpaid jobs. The producer acts as unpaid banker, financing the state through prepaid input taxes, and as unpaid inspector, policing suppliers through the invoice chain.
Two research questions follow. First, does the credit-invoice collection architecture, conceived in 1921 and first deployed nationally in 1954 so that businesses would administer the tax on behalf of states that could not observe transactions, remain justified once digital payment infrastructure gives states direct observation of transactions? Second, if the architecture is obsolete, how should a successor design convert the removed administrative burden into active support for production, the conversion our title names as the move from value added tax to value added subsidy?
Because the venue of our research paper submission is the Advanced Design Conference, one framing sentence is warranted. Tax collection systems are designed artifacts, built under the informational and technological constraints of their era, and artifacts built under vanished constraints deserve redesign rather than reverence. Our method is accordingly a design method: constraint analysis of the original architecture, mechanism analysis of the current burden, requirement analysis for the productive layer, comparative scenario synthesis for the successor, safeguard co-design for the privacy problem, and conversion of the savings into capability investment, followed by a defined evaluation program with named open parameters and closing design recommendations.
THE ARCHITECTURE OF 1954 AND THE CONSTRAINT THAT PRODUCED IT
The idea of taxing value added has two independent parents. Wilhelm von Siemens argued before the Reichstag in 1918 that Germany should replace its cascading turnover tax, which taxed the full sale price at every stage and therefore punished long supply chains while rewarding vertical integration. Thomas S. Adams, an American economist, proposed a business tax with credits in 1921, and the Adams version carried the specific mechanism that matters for our paper: invoice documentation to substantiate credit claims, creating an audit trail and distributing collection responsibilities across the supply chain (Mehrotra 2022). The German conception relied on direct subtraction of input values; the Adams conception subtracted input tax from output tax through invoices. Neither proposal was adopted at the time.
Implementation waited three decades. The American state of Michigan operated a business activities tax from 1953 to 1967, a form of value added tax computed by the subtraction method rather than through invoices. The modern credit-invoice mechanism arrived with Maurice Lauré, deputy director of the French tax authority, whose law of April 10, 1954 replaced a cascade of turnover and production taxes. The 1954 French tax initially reached only large businesses and the wholesale level. Denmark became the first European country to extend a value added tax to the retail level in July 1967, and France and Germany followed in January 1968, the year the French system was generalized across the whole economy. The European directives of 1967 made the model the common standard, and more than 170 countries eventually adopted a version, with the United States remaining the notable exception among wealthy economies.
The reason for the rapid adoption is the heart of our diagnosis. The Lauré design spread precisely because the design forced taxpayers at every level of the production process to administer and account for the tax themselves, rather than requiring assessment by tax authorities. The mechanism is elegant. Every business pays tax on purchases and charges tax on sales, reclaiming the difference. Because the buyer needs the seller's invoice to claim the credit, every buyer polices every seller, and the state collects the tax in fractions along the chain, so a single vanishing trader cannot cost the state the entire amount. In plain language, the credit-invoice chain is a self-policing surveillance workaround, a brilliant answer to a hard question: how does a state collect a high-rate consumption tax on transactions the state cannot see?
The constraint that justified the workaround has dissolved. Italy routes invoices through a state platform in real time. Brazil and Mexico have operated mandatory electronic invoicing for years. India couples electronic invoices to the tracking of goods in motion. The European Union adopted a digital reporting package in March 2025, with mandatory digital reporting for cross-border business transactions from July 2030 and harmonization of domestic systems by 2035, on the stated grounds that real-time reporting will reduce fraud by up to eleven billion euro per year and compliance costs by more than four billion euro per year. A state that observes transactions directly no longer needs businesses to observe transactions on the state's behalf. Once the informational reason for the chain disappears, the chain's costs, the float, the refund machinery, the filing, and the audits, lose their justification and stand exposed as pure instrumental-layer overhead. The next section measures that overhead.
HOW A CONSUMPTION TAX BECOMES A PRODUCTION BURDEN
A skeptic may object that economic theory settles the incidence question in favor of the consumer, making producer complaints irrelevant. Behavioral public finance has removed that objection. Consumers demonstrably underreact to taxes that are not salient: in a grocery-store field experiment, posting tax-inclusive price tags reduced demand by eight percent, and tax increases hidden inside posted prices change consumption more than equivalent taxes added at the register (Chetty, Looney and Kroft 2009). The theoretical consequence is exactly the license our diagnosis needs: the economic incidence of a tax depends on its statutory incidence, and even policies that induce no behavior change can create efficiency losses. How a tax is administered and experienced is economically real.
Consider first the float, which makes the producer an unpaid banker. Under the credit-invoice system, producers finance input taxes upfront and recover the taxes only after filing, waiting, and often surviving scrutiny. The magnitudes are large. In South Africa, refunds amount to fifty percent of gross value added tax collection, a transfer that must occur at high frequency, often monthly, and delays in refund payments reduce domestic investment, especially by small firms. When the South African revenue authority roughly halved audit rates on refund-claiming returns and sped up processing, investment increased by thirty-one percent and output by twenty-four percent (Brusco, Piek and Velayudhan 2024). Evidence from China matches. More than sixty percent of surveyed companies experienced refund delays, delays compromise firm liquidity, and a ten percent improvement in rebate efficiency raises small and medium enterprise employment by roughly six percent, with financially constrained firms benefiting disproportionately from faster clearance (Lu, Ma, Hu and Xu 2025). The float is a zero-interest loan from producers to the state, and the credit arrangement is regressive, because the state borrows for free from exactly the small producers who themselves borrow at the highest market rates.
Refund friction is institutional rather than accidental, which makes producer fear of claiming rational. Refund requirements and procedures are complex and burdensome, often discouraging taxpayers from claiming legitimate refunds, and taxpayers who do claim frequently experience significant delays or are not paid at all. Many administrations assume taxpayers cannot be trusted, treat the majority of refund claims as fraudulent, and design administrative processes to increase compliance costs and delay payment (International Monetary Fund 2021). In one documented case, the revenue authority explicitly states that first-time refund requesters are automatically subjected to audit. A remedy that punishes the claimant is a deterrent wearing a remedy's clothes.
The unsold-goods asymmetry supplies the clearest single illustration of the gap between design and practice. When goods are produced and never sold, because the goods spoil, fall out of fashion, or find no demand, no consumption ever occurs, yet the producer financed the input tax for the entire period. On paper the credit exists. In practice the credit arrives late, arrives after an audit, or, in non-deductible and exempt categories, never arrives. A tax that is supposed to touch only consumption has, in the failure case, touched only production.
Compliance costs are large and regressive. The literature review finds value added tax compliance costs significant in absolute terms and relative to revenue in developed and developing countries alike, and highly regressive, disproportionately affecting small businesses (Vishnuhadevi 2021). The regressivity has an arresting arithmetic: for a trader at a typical registration threshold, compliance costs of two percent of turnover combined with a twenty percent profit margin are equivalent to a ten percent tax on income, a burden that largely stays with small traders who lack market power (Ebrill, Keen, Bodin and Summers 2001). Aggregate magnitudes confirm the scale. United States taxpayers spend 7.9 billion hours on tax compliance annually, a total cost of 546 billion dollars, nearly two percent of gross domestic product, with most of the burden borne by businesses. European enterprise compliance costs average 1.9 percent of turnover, and Canadian estimates reported in the same study run between 1.2 and 1.8 percent of gross domestic product.
The machinery visibly brakes small-firm growth. United Kingdom administrative records show bunching in annual turnover just below the registration threshold, and many firms with positive growth deliberately hold sales under the line (Liu, Lockwood, Almunia and Tam 2021). Thai tax returns show the same pattern, and the authors conclude that the threshold acts as "a brake on small business growth", with non-registered firms growing significantly slower than comparable registered firms (Muthitacharoen and co-authors 2021). The threshold theory was formalized by Keen and Mintz (2004), and Japanese evidence adds that bunching intensifies exactly where compliance costs are higher. A tax architecture under which staying small is a rational business strategy prunes the productive layer at the root.
There is also a layer asymmetry: the instrumental layer largely escapes the machine built in its name. European Union law requires member states to exempt the transactions listed in Article 135 of the value added tax directive, a diverse range of which are associated with money and finance, because the margin-based nature of financial services makes the computation complex and potentially unworkable. Exempt suppliers cannot recover input taxes, and the known result is that financial services sold to consumers end up underpriced relative to full taxation (Mirrlees Review 2011). The architecture therefore concedes, in law, an inability to handle the instrumental layer's own core activity. The bakery finances the float while much of banking sits outside the machine.
Administrative burden also consumes something beyond money. Scarcity overloads mental bandwidth, and the strain on cognitive resources degrades job performance, decision-making, and self-control; scarcity reduces not a person's inherent capacity but how much of that capacity is currently available for use (Mullainathan and Shafir 2013; Mani, Mullainathan, Shafir and Zhao 2013). Applied to the present setting, every hour a small producer spends navigating input credits is an hour of managerial cognition diverted from product, process, and customer. Our paper records one honesty clause here: a later replication project reproduced most but not all of the foundational results in the scarcity literature (Shah, Mullainathan and Shafir 2019), so we treat the bandwidth mechanism as strong but contested at the edges.
Finally, the tax hides. The hypothesis that taxpayers systematically underestimate indirect taxes concealed inside prices runs from Mill (1848) through Puviani (1903) to Buchanan (1967), and carries empirical support: governments relying more on invisible indirect taxes have sustained higher spending, other things equal. A democracy is healthier when citizens can see what government costs. Any successor design should therefore make the levy visible on every transaction record, a requirement our design adopts below.
WHY PRODUCTION DESERVES PRIORITY
The mechanisms above would matter little if production were one sector among many. Our paper argues that productive capacity is the compounding root of three forms of national strength, so burdens on producers propagate into weakness everywhere.
The economic chain rests on the fact that production teaches. Wright's study of aircraft manufacturing gave the first academic statement of the learning curve, the stable relationship between cumulative output and falling unit cost (Wright 1936), and Arrow built early endogenous growth theory on the observation that labor hours per airframe decline as a precise function of cumulative production, a relation reliable enough to anchor United States Air Force planning (Arrow 1962). The modern demonstration is renewable energy, where every doubling of cumulative solar panel installation has been associated with roughly a twenty percent cost decline. Beyond the single firm, capabilities pool into what Pisano and Shih call the industrial commons, the collective operational capabilities embodied in workforces, suppliers, competitors, and universities that underpin new product and process development. Their warning is that declining manufacturing capability produces a corresponding loss of innovation, and that a lost industrial commons is nearly impossible to retrieve (Pisano and Shih 2009, 2012). At national scale, the economic complexity literature quantifies the thesis: the Economic Complexity Index, a measure of productive capabilities, predicts growth in income per person more accurately than conventional measures of governance, competitiveness, and educational attainment, including after statistical controls (Hidalgo and Hausmann 2009). The same literature names the failure mode, premature deindustrialization, in which economies that never build manufacturing depth suffer lower growth later. Production also spreads prosperity locally: each additional tradable manufacturing job has been estimated to create roughly 1.6 additional local service jobs, and high-end manufacturing categories show multipliers near five (Moretti 2010), although our paper cites the honest range, since a replication using an alternative instrument found closer to 0.8 additional jobs per manufacturing job (van Dijk 2015).
The military chain follows from a simple law: in extended conflict, military power equals production rate versus attrition rate. Defense institutions state the dependence directly, arguing that security requires a strong military, which is possible only if the underlying economic infrastructure is in place, and that national security rests on industrial capability as much as on military capability. The historical benchmark is stark. The wartime United States produced seventeen aircraft carriers, three hundred thousand airplanes, and roughly fifty thousand tanks in under four years, against a maximum modern production capacity of approximately one hundred fifty F-35 fighters per year. The erosion is documented: United States domestic semiconductor production capacity fell from thirty-seven to twelve percent of the global total, and production struggles now span artillery shells, drones, and submarines, with supporting sectors such as key machine tools essentially nonexistent domestically. The war in Ukraine supplied the live demonstration. Both armies fired ten thousand to twenty thousand artillery rounds on single days; stockpiles were depleted or neared exhaustion, transferring battlefield pressure directly onto the industrial base; Russian shell production reportedly rose from one million per year toward a target of four and a half million, while Western countries struggled to keep pace, exposing the fragility of peacetime production. Two details connect the military chain to the rest of our design. A major defense manufacturer entered talks to convert an automotive plant to defense production, which shows that civilian factories are the convertible reserve of wartime economies, and military analysts now recommend standardizing not only technical specifications but ballistic characteristics of artillery ammunition to achieve true interoperability, an independent restatement of the standardization principle our broader research program advocates for civilian production. One caution accompanies the whole chain: the sources in the present paragraphs are government reports, policy institutes, and defense trade analysis, documentary evidence rather than peer-reviewed economics, and the full research program pairs them with the economic history of wartime mobilization. The connection to tax design is direct. The float penalizes capital-intensive factory building hardest, because input tax on machinery is financed upfront and recovered slowly, and compliance overhead suppresses the small workshop tier that gives an industrial base redundancy and convertibility.
The cultural chain treats products as ambassadors. Nye defined soft power as the ability to attract and persuade, arising from the attractiveness of a country's culture, ideals, and policies rather than from coercion (Nye 1990, 2004). Manufactured goods are among culture's most persistent carriers, and the effect is measurable in prices: experimental studies show that country of origin has a positive causal impact on consumer willingness to pay, with consumers paying measurably more for identical products attributed to countries with favorable images (Koschate-Fischer, Diamantopoulos and Oldenkotte 2012). Our paper states the direction of the loop honestly. The experiments measure the path from image to price premium, while our claim adds the prior link, that decades of production excellence build the image in the first place. The two links together mean that every excellent factory pays a dividend to every other producer under the same flag. A country that only consumes imports culture with every container; a country that produces exports culture with every container. The tax connection is immediate, because the small ateliers, workshops, and designers who carry a nation's design culture are precisely the firms on which fixed compliance costs fall hardest.
All three chains strengthen when production is distributed. Many small producers are harder to destroy than one plant, in war or in supply-chain crisis. Many parallel producers run many parallel experiments, which is how innovation searches its space. Local producers form local capital and local skills. Distribution is engineering redundancy rather than romantic nostalgia, and the evidence above shows that the current machinery selects against exactly the distributed tier.
WHY INACTION IS NOT NEUTRAL
Industrial policy has returned worldwide, with major programs addressing the green transition, geopolitical competition, and supply-chain resilience, and the scholarly assessment has turned: a considerable recent literature provides rigorous evidence on how industrial policies work, improves substantially on the older correlational generation, and on the whole offers a more positive assessment (Juhász, Lane and Rodrik 2024). A state that declines to support domestic production does not receive a neutral market. That state receives a market tilted by foreign treasuries.
The current tax architecture is itself an industrial policy, an accidental one aimed against small producers. The mechanisms documented above tilt the field against long production chains, capital-intensive builders, and small firms, while largely exempting the financial core of the instrumental layer. A state operating such machinery is already picking sides. The honest choice is not between intervention and neutrality but between accidental anti-production policy and deliberate pro-production design.
The obvious objection, that governments should modernize the existing tax rather than replace the architecture, now has an evidence-based answer. Digital compliance tools bolted onto value added tax systems raise substantial revenue yet impose regressive compliance costs that fall most heavily on small firms, and without careful design and sequencing such reforms risk undermining welfare and formalization (International Growth Centre 2025). Electronic invoicing attached to the 1954 chain automates the surveillance while keeping the burden: the state gains real-time sight while the producer keeps the float, the refunds, and the filing. Digitizing the old design is not the reform. Redesigning the collection point is the reform.
Coherence supplies a further argument. Contemporary states subsidize production with one hand and burden production with the other, and both hands carry administrative overhead, one foot on the accelerator and one foot on the brake, with both pedals burning fuel. Agriculture is the established example, permanently subsidized because society needs food and simultaneously taxed and audited. Our paper generalizes the observation to the entire productive layer. A state that stops contradicting a principle the state already holds is not adopting radical policy. That state is achieving coherence.
Self-interest points the same way. Every stream of state revenue is downstream of production, so a state that burdens the productive layer shrinks the base of all future collection. There is also an automation asymmetry: tax administration is bureaucracy, and bureaucracy has no consumers to destroy, so automating collection is the safest possible automation target and attacks one root of permanently rising administrative cost in the sense of Baumol's cost disease, since administration performed by human hours never enjoys manufacturing's productivity growth. On time horizons, productive capability pays back over decades, beyond private investment windows, and the record shows states have long carried that horizon: the state has not merely fixed market failures but actively shaped and created markets, funding the foundational technologies of modern computing and communication (Mazzucato 2013). Mazzucato's further observation, that the present innovation system socializes risks while privatizing rewards, states almost word for word the dysfunction our broader research program identifies in contemporary artificial intelligence economics.
The objection that governments pick winners deserves respect. Skeptics rightly point to the many failures littering development economics, and political incentives, not only market failures, shape which policies are adopted. Our design answers structurally rather than rhetorically. Support is horizontal, directed at capability, standards, machinery, and the small-producer tier, never at named champion firms. Allocation is plural, spread across multiple state banks, private banks, and small local banks with overlapping mandates, so no single allocator's failure is fatal. Verification is algorithmic on the transaction ledger, which knows who genuinely buys inputs, ships goods, and sells to real customers.
THE SUCCESSOR DESIGN SPACE
The history of consumption taxation teaches one law: collection friction is never deleted, only relocated. Under value added tax, friction lives in the producer's cash flow, as float, refunds, and filing. Under the American retail sales tax, friction lives on the retailer's honesty, because the entire levy concentrates at one point. Under naive digital collection, friction moves into the citizen's privacy, because the taxing authority sees every purchase in real time. The design task is not to abolish friction but to relocate friction to where friction does the least damage, and the following section argues that a fourth location exists.
The retail sales tax record deserves precision, because our design must beat that record. Analysts have observed that at rates above twelve percent sales taxes become too easy to evade, and that no historical precedent exists for a high-rate enforceable retail sales tax; replacement-level rates for a national retail sales tax in the United States were estimated at thirty-four to eighty-nine percent on a tax-exclusive basis once evasion and base erosion were considered. Even at Florida's low single-digit rates, one study found five percent of business exemption certificate purchases misused for personal consumption. American state rates cluster in single digits for a reason. The classic killer is the identity question: is a given buyer a business or a consumer? Under paper administration, that question is a fog in which evasion hides. In an all-digital payment system with verified identities, the classification becomes a graph property. Real-time analytics can flag the construction firm whose purchase pattern consists of groceries and television sets, and deliberate misrepresentation is fraud, a category whose consequences the next section defines. Digital enforcement therefore attacks both incumbent designs at their weakest points, provided the privacy cost is answered.
The first successor scenario is a thin universal transaction levy. The lineage is Feige's automated payment transaction tax, which proposed replacing personal and corporate income, sales, excise, capital gains, import and export duty, gift and estate taxes with a single flat tax on all transactions, automatically assessed and collected when transactions settle in the payment system, with lower administrative and compliance costs (Feige 2000). The proposal received national attention in the United States, with a major newspaper describing the design as "fair, simple, and efficient", and Brazil operated a broadly similar transaction tax, reportedly with reasonable success, although layered on top of existing taxes rather than replacing them. Because transaction volumes are a large multiple of gross domestic product, the rate can sit in the low single digits, for illustration a rate consistent with historically persistent flat levies. The thin levy abolishes the reclaim machinery completely: nothing is reclaimed because the small charge simply sticks at each hop, so the float, the refunds, and the audit fear vanish by construction. Our paper openly acknowledges the cost. A turnover levy cascades, so a product passing five taxed stages embeds a cumulative burden on the order of twelve percent while a vertically integrated giant pays once, a structural tilt toward consolidation. The critique is historically grounded, since the cascading German turnover tax of the early twentieth century demonstrably encouraged vertical integration, which is exactly why Siemens proposed reform. Two counterweights answer the tilt: the small-producer exemption described below, and the compliance savings, which for small firms dominate the cascading cost. Financial-sector legs receive separate treatment, a micro-rate for hedging contracts held by parties with verifiable real exposure to the underlying good and the full rate for unanchored churn; the distinction has independent scholarly precedent, since business-ethics research already applies the Aristotelian boundary between natural and unnatural acquisition to assess high-frequency trading, and a companion paper in our research program develops the exposure test in full.
The second scenario collects a full rate only at the final consumption event, identified digitally by the buyer's status, with nothing taxed between businesses. The advantages mirror the defects of value added tax exactly: perfect stage neutrality, which serves distributed domestic supply chains; zero float; zero refunds; zero business-side filing. The risk is the concentration of temptation, because replacing value-added-tax revenue at a single point requires a rate near current European rates, historically the regime where single-point collection collapses into evasion. The second scenario therefore stakes everything on the digital identity argument, a bet our paper judges increasingly plausible but not yet proven at national scale.
The third scenario is a hybrid, a thin universal levy plus a modest final-consumption top-up. The thin component keeps every transaction lightly contributing, while the top-up restores revenue without pushing the final-point rate into the historical collapse zone, and cascading stays mild because the universal component is small. In design terms the hybrid is a tuning dial between the first two scenarios rather than a third philosophy.
The three scenarios are evaluated against five criteria: revenue adequacy, cascading burden, evasion robustness, administration cost, and privacy cost. The anticipated pattern is as follows. On revenue adequacy, the thin levy and the hybrid score high because of the enormous base, while the final-point design is rate-limited by evasion risk. On cascading, the final-point design is perfectly clean, the hybrid nearly clean, and the thin levy carries the acknowledged mild tilt toward integration. On evasion robustness, the thin levy is strongest because no single hop is worth cheating, the hybrid is strong, and the final-point design carries the open bet on digital identity. On administration cost, all three are minimal, because all three abolish the reclaim machinery, which the mechanism analysis identified as the true source of the production burden; the choice among scenarios is secondary to that shared architectural move. On privacy cost, all three share one unpaid bill, total transactional visibility. No scenario survives politically without an answer to the question of who guards the ledger. That question, we argue next, is what killed the Feige proposal, and our program answers the question by construction.
FISCAL SECULARITY
Feige himself noted that automated recording of all payments creates a degree of transparency, transparency of citizens to the state. The automated payment transaction tax was technically sound and politically dead, because a design in which the taxing authority sees every transaction of every citizen is, without safeguards, a surveillance machine with a fiscal feature. The central thesis of our research program follows: the automated transaction tax becomes viable only when bundled with Fiscal Secularity, the constitutional and technical separation of money, taxation, and governance, doing for money what the separation of church and state did for belief. The bundle, not either component alone, is the contribution.
The records architecture is sealed, not secret, with three properties required simultaneously. Records are immutable: no person and no institution can alter, delete, or roll back an entry, because without immutability the access log proves nothing. Records are sealed: not readable without legal cause. Records are conditionally accessible: readable with legal cause through a judicial gate, and every act of access is itself logged, with the log entry, stating who read, when, and for what stated reason, instantly and permanently visible to the citizen whose records were opened. The working precedent is Estonia, where any citizen can log into the state portal and see exactly which officials queried their personal data, when, from which database, and for what purpose, and where unauthorized access to citizen data is a criminal offense punishable by imprisonment, with the visible trace itself acting as a deterrent. The security of the design must never depend on secrecy of design; the architecture is fully public and auditable. Sealed data, transparent design.
The state may act upon transaction records only for theft of funds and fraud, including obtaining funds by deception. For lesser matters, citizens receive de facto privacy: the records exist, but no state action may be built upon them. The action threshold must be constitutionally entrenched and deliberately difficult to amend, because historical exceptions always expand, from terrorism to money laundering to tax to speech-adjacent offenses. The design goal is to prevent threshold creep structurally rather than by goodwill. An evidence rule completes the gate: evidence obtained from transaction records outside the judicial process, including through covert or classified programs, is inadmissible in court. Reading the ledger without cause yields legally useless information, which removes the incentive to peek rather than merely the permission.
Nothing on the ledger is ever technically reversed, and the state has no capability to edit any entry. Corrections of error, theft, and fraud occur as new transactions: a court orders the obligated party to initiate a compensating transaction, and the party complies or faces ordinary physical enforcement. Where buyer and seller both consent, as in the routine refund or the duplicate charge, reversal by new transaction plus automatic tax clawback runs with no human involvement, and dispute resolution is algorithmic in the first instance with human arbitration only on appeal, preventing the dispute system from growing into a new bureaucracy. The principle: reversal is a judicial act performed through people, never an administrative act performed on the ledger. The enforcement monopoly compels humans; the enforcement monopoly never touches the ledger.
Freezing or blocking a citizen's payments must be simultaneously illegal, by constitution and statute, and expensive, requiring a judicial order that is logged, visible to the affected citizen, appealable, and deliberately slow. The motivation is contemporary and concrete: accounts of protesters frozen, individuals debanked, payments blocked over opinions. Money is stored life energy. Citizens who fear that money can be taken for the wrong words self-censor, and a democracy in which citizens fear their own payment system carries a silent veto on speech. Fiscal Secularity exists so that no such veto can be built.
The theoretical core of the design can now be stated. The iron law says friction relocates; Fiscal Secularity does not delete the privacy friction either, but relocates friction to a fourth location, constitutional maintenance, and the relocation is decisive for a reason public finance has always understood: marginal wedges distort behavior, while fixed costs do not. The float scales with every production cycle. Retailer evasion risk scales with every sale and with the rate. Naked surveillance chills at every purchase. The judicial gate, the visible access log, the inadmissibility rule, and the entrenched threshold cost approximately the same to maintain whether the economy performs one million or one billion transactions per year. The bundle therefore converts taxation's friction from a distortion at the margin of every productive act into a fixed institutional overhead, which is the translation into public finance of the opening axiom that the instrumental layer must be thin at the point of contact with production.
Who guards the ledger? The answer is structural, in three parts. Nobody can edit the ledger, because of immutability. Citizens guard the readers, because every reading leaves fingerprints visible to the person read. The constitution guards the thresholds, through entrenchment and inadmissibility. Guarding is mutual and permanent rather than promissory.
Three honest residuals remain and are stated rather than hidden. First, mandatory digital payment is a civilizational precondition, not a software update. Shadow economies averaged 31.9 percent of gross domestic product across 158 countries between 1991 and 2015, with advanced economies below twenty percent (Medina and Schneider 2018), and a payment mandate is legitimate only alongside three guaranteed decencies: decent conversion of assets into the digital currency, decent ability to earn within open marketplaces, and decent ability to spend without foreign or administrative shutdown. Second, the trust bootstrap: the fixed-cost conversion works only to the degree citizens believe the seal, and belief is built by visible structure, logged access and real prosecutions for unauthorized access, not by declarations. Third, seal durability under future regimes: append-only records and public design raise the cost of abuse but cannot make abuse impossible, and the ultimate backstop is constitutional culture. Our paper says so plainly rather than overclaim.
THE VALUE ADDED SUBSIDY
The conversion promised in our title proceeds in two steps, and the fiscal story closes without new money. Step one is the negative subsidy: stop the friction. Abolishing the credit-invoice machinery costs no real resources and releases compliance capacity documented near two percent of gross domestic product. The cheapest production subsidy available is the removed burden. Step two is the positive subsidy: redirect a fraction of the released resources into productive capability. The subsidy is prepaid by the abolished paperwork.
The target must be capability, never gross output, because subsidizing output invites production for the subsidy's sake. Four channels carry the support: machinery, attacking exactly the capital-formation margin the float suppressed; training, the skills component of the industrial commons; standardization, the compatibility and repairability of mechanical and electronic products, whose military value the war evidence documented independently; and the small-producer tier, the resilience and innovation reserve. Verification is algorithmic on the ledger, which knows who buys inputs, ships goods, and sells to real customers, so the subsidy needs no new inspection bureaucracy.
The small-producer exemption is engineered against gaming. Producers employing fewer than ten people and selling into the domestic market pay no tax of any kind, with the exemption tapering smoothly to zero at twenty-five employees. Smooth tapering matters because hard thresholds teach firms to stay small. The French case is the canonical demonstration: administrative data show a sharp fall in the number of firms at exactly fifty employees compared to forty-nine, where labor regulations begin to bind, with regulation costs equivalent to a 2.3 percent labor tax and welfare costs estimated at 3.4 percent of gross domestic product, and the same study analyzes a parallel threshold at ten employees, directly relevant to our boundary (Garicano, Lelarge and Van Reenen 2016). A smooth taper makes splitting a company into fake small pieces gain almost nothing. Where fragmentation is attempted anyway, detection is a graph query rather than an audit, since related entities reveal themselves through shared beneficial owners, shared addresses, and transaction flows circulating among themselves. Whistleblower bounties supplement the graph, on the American model, where the tax whistleblower program has recovered nearly seven billion dollars since 2007 while paying whistleblowers over one billion, with mandatory awards of fifteen to thirty percent of proceeds. Structured evasion through fake fragmentation is classified as fraud, which places the conduct above the judicial action threshold defined earlier. The governing sentence of the whole design: your money is inviolable; lying to the system about who you are is not.
The exempt are watched. Exempt small producers are checked automatically once per year at a randomized moment, and each checked producer sees the check in the personal access log. Randomized timing follows a variable-interval schedule, which behavioral research since Ferster and Skinner (1957) associates with steady, persistent behavior that is highly resistant to extinction. Visible auditing of the untaxed maintains legitimacy for everyone else. Exempt, but watched.
The subsidy half of our title stands in an old line, from Hamilton's Report on Manufactures through List's national system to the contemporary industrial policy revival. What our design adds to the lineage is the funding source, abolished compliance machinery rather than new taxation, and the verification method, the ledger rather than the inspectorate.
ANTICIPATED FINDINGS AND EVALUATION PROGRAM
Three findings are anticipated, stated as testable expectations. First, digitizing enforcement without redesigning the collection point deepens regressive burdens on small firms, as early evidence already indicates. Second, final-consumption collection secured by digital identity defeats the fake-business-purchase fraud that historically limited retail sales taxes, because purchase-pattern classification in a verified-identity ledger is algorithmically tractable. Third, abolishing the compliance machinery releases resources near two percent of gross domestic product, sufficient to fund the capability subsidies without new taxation.
The named open parameters are the universal levy rate; the final-point top-up rate; the exemption boundary and taper slope, illustratively ten and twenty-five employees; the micro-rate for exposure-verified hedging instruments; the refund-latency guarantee for consensual reversals; the randomization interval for exempt-sector checks; and the privacy metrics, including access-log volume per capita, gate-refusal rates, and prosecution rates for unauthorized access.
The evaluation procedure uses five instruments in order of increasing commitment: microsimulation of the three scenarios on national transaction datasets, testing revenue adequacy and distributional incidence; cascading measurement through input-output tables, quantifying the consolidation tilt of the thin levy; adversarial stress-testing of the identity layer, so the final-point bet is red-teamed before being trusted; privacy auditing against the metrics above under simulated hostile administrations; and staged regional or sectoral pilots with pre-registered success criteria, because a tax architecture, like any designed artifact, earns trust through prototypes rather than proclamations.
The limitations of the supporting evidence are recorded without embarrassment. The refund-burden findings are heterogeneous: one careful Honduran study of a withholding reform estimated null average effects on growth and investment, challenging the premise that unrefunded credits bind in all settings (Pineda Pinto, Bermudez and Scot 2024). The null result sharpens rather than weakens our claim, since the burden concentrates on financially constrained small producers, exactly where our design directs relief, and we present the South African, Chinese, and Honduran results together as a heterogeneity finding. The cognitive-bandwidth mechanism carries a replication caveat. The local-multiplier magnitudes are contested within a factor of two. The military-chain sources are documentary rather than peer-reviewed. The distributional profile of flat transaction levies requires modeling rather than assertion; Feige argued that the base itself introduces progressivity, because the wealthy conduct a disproportionate share of transactions, and our microsimulation instrument is designed to test that argument rather than assume the argument. Finally, the cashless precondition and the trust bootstrap are political constructions whose timescale no simulation can compress.
IMPLICATIONS AND DESIGN RECOMMENDATIONS
Seven recommendations follow for treasuries, central banks, and the design community. Treat tax collection architecture as a designable public artifact, subject to prototyping, evaluation, and replacement when founding constraints vanish; the credit-invoice chain answered an observability constraint that no longer exists, and the chain took thirty-three years to travel from concept to national implementation and another fourteen to full deployment, a reminder that redesign should begin before the old design's constraints have been obsolete for a generation. Do not confuse digitizing the old design with redesign, because electronic invoicing bolted onto the chain automates surveillance while keeping the burden. Collect consumption tax at the final consumption event, or through a thin universal levy, or through the hybrid of the two, but in every scenario abolish the reclaim machinery, which is the true engine of the production burden. Make the levy visible on every transaction record, ending the fiscal illusion that hides the cost of the state inside prices. Never deploy total transactional visibility without the constitutional counterpart: sealed-by-default records, a judicial action threshold restricted to theft and fraud and entrenched against creep, citizen-visible access logs, an inadmissibility rule for evidence gathered outside the gate, and an append-only ledger the state can compel people to correct but can never itself rewrite, the doctrine our research program names Fiscal Secularity. Convert the released compliance resources into capability subsidies, for machinery, training, standardization, and the small-producer tier, verified algorithmically on the ledger rather than through a new inspectorate. Judge every element of the design against the single axiom from which our paper began: the instrumental layer of an economy exists to serve the productive layer and should be as thin as possible, cheap, invisible, and unable to mistake itself for the point, because food, housing, machinery, and care are the purpose, and money, taxation, and ledgers are only the plumbing through which a nation's real wealth flows.