America Went Back to Hamilton — and It’s Working

Tariffs, subsidies, public credit, and now equity stakes: the U.S. has quietly reassembled its founding economic playbook. Here is the theory, the data, and an index to measure it.

The Hamiltonian Turn — Article I of a new Vector1 Research Program

By Brian Curry · Vector1 Research

This is the opening paper of The Hamiltonian Turn*, a new Vector1 Research program. Its thesis: between 2018 and 2026, across two administrations of opposite parties, the United States dismantled the policy consensus of the past seventy years and reassembled — plank by plank, mostly without naming it — the economic program Alexander Hamilton wrote down in 1791. This series treats that turn the way Vector1 treats everything: as a system to be formalized, measured, and projected. Article I builds the framework, presents the evidence to date, and introduces the* Hamilton Index — a single composite tracking how Hamiltonian American policy has been across 230 years. The index’s shape is the story: a long high plateau, a postwar collapse, and now a vertical recovery.

1. The Pattern Nobody Named

Consider the following events as a single sequence. In 2022, Congress passed the CHIPS and Science Act — $52.7 billion in direct manufacturing subsidies plus a 25% investment tax credit — and the Inflation Reduction Act’s hundreds of billions in production incentives. In 2024–2025, tariffs returned as a central instrument of policy at rates not seen since mid-century. In February 2025, Executive Order 14196 directed the creation of a United States sovereign wealth fund. In June 2025, the government took a golden share in U.S. Steel — veto power over a private company’s strategic decisions. In July, the Defense Department became the largest shareholder of MP Materials, the only American rare-earth producer, with a guaranteed-purchase agreement for 100% of its new magnet output. In August, $8.9 billion of grant money was converted into a 10% equity stake in Intel. By spring 2026 the government had taken equity, warrants, or golden shares in more than twenty companies — $20.9 billion across sixteen deals — alongside revenue-sharing agreements with Nvidia and AMD.

Commentators have processed each event separately: a subsidy debate here, a tariff fight there, a governance controversy over the Intel stake. Viewed together, they are not separate events. They are the five planks of a single, very old program: protective tariffs, manufacturing subsidies (“bounties,” Hamilton called them), public credit directed at industry, federal participation in strategic enterprises, and internal improvements. That is, almost item for item, the Report on the Subject of Manufactures, submitted to Congress by Alexander Hamilton in December 1791.

Figure 1. The Hamilton Index — Vector1’s composite of tariff intensity, industrial subsidies, public equity participation, and directed credit (Equation 6, methodology in §6) — across 230 years. The American System era ran high for a century and a half; the postwar liberal order drove the index to its 2010s floor; the 2018–2026 turn is the steepest sustained rise in the series. America is not adopting a new economic model. It is reverting to its oldest one.

The claim of this series is not merely historical rhyme. It is that the Hamiltonian program is a coherent piece of economics — formalizable, testable, and, on the evidence assembled below, working — and that understanding it as a system, rather than as a pile of controversies, is the only way to project where the American economy goes next.

2. What Hamilton Actually Argued

Strip the 1791 Report of its eighteenth-century prose and its argument is startlingly modern. Against Adam Smith’s case that nations should specialize where they currently hold advantage, Hamilton made four moves. First: advantage is manufactured, not endowed — a nation that only does what it is currently best at will never become best at anything new. Second: manufacturing differs from agriculture because it exhibits what he called “the division of labour… the extension of the use of machinery… and the promoting of emigration of skilled workmen” — in modern terms, increasing returns, learning effects, and knowledge spillovers. Third: private investors will underinvest in infant industries because they cannot capture those spillovers and cannot survive the loss-making years against established foreign incumbents — a coordination failure requiring “the incitement and patronage of government.” Fourth: for a nation surrounded by rivals, the capacity to make essential goods domestically carries a security value above its market price.

Tariffs, bounties, the Bank, federal participation, internal improvements — each plank maps to one of those four arguments. The program ran, under the name of Clay’s American System and its successors, from 1791 to roughly the Second World War — the precise interval during which the United States went from agrarian periphery to industrial hegemon. Then, as the established incumbent with no rivals left, America did what incumbents rationally do: it preached free trade to everyone else and let its own index collapse. The 2018–2026 turn began, on schedule, exactly when a rival capable of running the Hamiltonian playbook against the United States — China — made the incumbent’s strategy obsolete.

3. The Economics, Formalized

Hamilton’s four arguments compress into one production function. Let output depend on capital and labor through a productivity term — but make productivity endogenous: it grows with cumulative production (learning-by-doing) and with the output of neighboring firms in the cluster (Marshallian spillovers):

The learning term is not a theoretical flourish; it is one of the most robust empirical regularities in industrial economics — Wright’s law, measured first in 1936 aircraft plants and confirmed since across semiconductors, ships, solar panels, and batteries:

For semiconductors and batteries the learning exponent implies cost declines of roughly 20–30% per doubling of cumulative output. This single fact carries Hamilton’s whole argument: whoever produces, learns; whoever learns, leads. Comparative advantage in such industries is not discovered. It is accumulated — and a rival who subsidizes accumulation takes it from you.

When does intervention pass a welfare test? The classical answer — the Mill–Bastable condition — requires the discounted future cost advantage the industry eventually achieves to exceed the upfront cost of protection:

And the optimal intervention is not unlimited: it prices exactly the wedge between private and social returns — the learning externality, the cluster spillover, and (Hamilton’s fourth argument) a security premium for capacity whose value appears only in crises:

This equation is the series’ discipline. Hamiltonianism done right is not “subsidize everything”; it is price the three externalities, and nothing else. Every failure mode of industrial policy — and there are many — is a violation of Equation 4: subsidizing where no learning curve exists, protecting where no spillover operates, or letting the security premium become a blank check. The modern empirical literature (Juhász, Lane, and Rodrik’s 2023 synthesis; Lane’s 2022 study of Korea’s heavy-industry drive) finds industrial policy works precisely where Equation 4 says it should: steep learning curves, strong clustering, and clear externalities.

4. The Evidence: What Four Years Bought

Theory established, what does the data say? Begin with the most direct measure of whether incentives moved real resources: factory construction.

Figure 2. U.S. manufacturing construction, 2005–2026. After averaging ~$6B per year for a decade, electronics-factory construction reached a $135B annual rate by mid-2024 — more capacity begun in 2024 alone than in the prior twenty years combined — driving total manufacturing construction to roughly triple its pre-2022 level. The post-2024 decline is examined honestly in §5.

The legislative timing is visible in the series itself: construction turns exactly along the CHIPS Act’s legislative arc, and the Census data measure realized private spending — concrete poured, not promises. The headline stock: roughly $460 billion in private investment stimulated across semiconductors, batteries, and clean-energy manufacturing against $39 billion of CHIPS manufacturing incentives. That ratio is the modern version of Hamilton’s central design insight — public money should anchor, not replace, private capital:

Figure 3. The crowd-in pattern across programs. CHIPS incentives: ~12 private dollars per public dollar. MP Materials: a $400M defense stake plus guaranteed offtake unlocked $1B from JPMorgan and Goldman within weeks — Wall Street followed because procurement certainty de-risked the project. This is the mechanism, not a coincidence: the state absorbs the coordination risk private capital cannot, and private capital does the rest.

The microeconomic detail of MP Materials deserves emphasis, because it is Equation 4 operating in the wild. Rare-earth magnets sit at the intersection of all three externalities — steep learning curve, single-point-of-failure security exposure, and a Chinese incumbent with 90% market share built by exactly these tools. The Pentagon’s stake plus 100% offtake guarantee converted an unbankable project into an oversubscribed one. The scoreboard so far: the company’s new Texas magnet facility is funded, built ahead of schedule, and fully pre-sold — domestic capacity that did not exist and was not going to exist at any market price.

And the deepest evidence is in what physically got built: 97 new fabs launched between 2023 and 2025, Micron’s $200 billion program including the largest private investment in New York’s history, TSMC’s Arizona complex, and a credible path for the U.S. share of global semiconductor manufacturing from 12% toward the 20% target by 2030 — reversing a fifty-year offshoring trend in roughly fifty months.

5. The Honest Ledger

A research program that only counts wins is advocacy. Three complications, stated plainly.

The construction cooldown. Electronics-factory construction has fallen ~44% from its July 2024 peak, and total manufacturing construction is down about a fifth. Some of this is benign — buildings finish, and spending naturally rotates from construction to equipment installation, which the construction series doesn’t capture. But not all of it: the 2025 tariff broadening did not trigger a measurable second reshoring wave (ex-electronics construction rose only ~5.6% after the tariffs), and policy uncertainty around CHIPS and IRA continuity is itself a tax on long-horizon investment. The lesson is Equation 4’s: the subsidy planks targeted learning-curve industries and moved billions; the broad tariff plank, applied indiscriminately across industries with no learning curve, has so far mostly moved prices. Hamiltonianism is a scalpel that is currently being used, in places, as a hammer — and the data can tell the difference.

The governance gap. A de facto sovereign portfolio assembled by executive action — sixteen deals across three agencies with inconsistent terms, no statutory structure, no independent board, no transparency rule — concentrates exactly the discretion that makes industrial policy degenerate into clientelism. Norway’s fund works because of its governance, not despite it. This series will argue (Article III) that the portfolio’s returns are real but its structure is the program’s largest unpriced risk.

The fiscal cost. Bounties and credits are not free, and the Mill–Bastable integral is an empirical claim, not a guarantee. The test for each program is whether the learning-curve and security returns it purchased exceed its cost — a question Article II answers industry by industry, with the semiconductor program as the cleanest test case.

6. The Hamilton Index

To track the turn as a single object, we define the Hamilton Index: a weighted composite of the four measurable planks — effective tariff intensity, industrial subsidies as a share of GDP, public equity participation in strategic enterprises, and directed credit:

Figure 1 plotted its 230-year arc. Three readings matter. First, the level: at roughly 0.74, the 2026 index has returned to territory last occupied in the 1930s — but its composition is different, lighter on tariffs and far heavier on equity participation than any prior American configuration, which makes the current program in one respect more Hamiltonian than the nineteenth century: Hamilton’s original design included direct federal shareholding (the Bank of the United States was 20% government-owned), a plank the American System never fully used and Washington has now revived at scale. Second, the speed: the 2018–2026 rise is the steepest in the series, compressing a generational policy reversal into eight years. Third, the bipartisanship: the index rose under both parties — CHIPS and IRA under one administration, tariffs, the sovereign fund order, and the equity portfolio under the next — which historically is the signature of turns that persist.

Figure 4. The newest plank, quantified: $20.9 billion of public equity across sixteen deals in sixteen months, led by Defense and Commerce, plus golden shares and revenue-sharing agreements. The Bank of the United States was 20% federally owned in 1791; the instinct is not new — only the tickers are.

7. What This Series Will Do

The program is three papers, applying the Vector1 method — formalize, measure, project — to the turn as a whole:

Article II — The Scorecard. Grading the turn’s instruments against Equation 4, industry by industry. The learning-curve dividend: how much cost decline, capability, and market share the CHIPS portfolio actually purchased, tested against Mill–Bastable. The tariff ledger: which tariffs price a real externality and which merely tax consumers, in the revenue, price, and reshoring data. And the security premium: pricing ρ — what domestic capacity in chips, rare earths, pharmaceuticals, and energy is worth as insurance, using option-value methods. The cleanest experiment in industrial-policy history is running now; we grade it.

Article III — The American Fund and the Hamiltonian Century. The portfolio exists; the institution doesn’t. What the U.S. should copy from Norway, Singapore, and its own In-Q-Tel — the governance design that determines whether public equity compounds or corrodes — and then the projection: where the index, the learning curves, and the portfolio put American manufacturing, growth, and strategic position by 2050, with the three ways the turn could fail.

Two and a half centuries ago, the first Treasury Secretary argued that markets build wealth but do not, unaided, build the capacity on which wealth and sovereignty rest — and the country that took his advice became the industrial center of the world. The country that forgot it spent fifty years watching that capacity migrate. The wager of the current turn — and of this series — is that the oldest American playbook still works. The early data says the wager is paying.

References and Sources

1. Hamilton, A. (1791). Report on the Subject of Manufactures. The five-plank program: tariffs, bounties, public credit, federal participation, internal improvements.
2. Wright, T. P. (1936). “Factors Affecting the Cost of Airplanes.” Journal of the Aeronautical Sciences. The original learning curve.
3. Juhász, R., Lane, N., & Rodrik, D. (2023). “The New Economics of Industrial Policy.” Annual Review of Economics. The modern evidence synthesis.
4. Lane, N. (2022). “Manufacturing Revolutions: Industrial Policy and Industrialization in South Korea.” The HCI natural experiment.
5. US Census Bureau, Value of Construction Put in Place survey; PIIE (2024), “US chip construction spending skyrocketed after the CHIPS Act”: electronics construction from ~$6B/yr (2011–2020) to a $135B annual rate (mid-2024).
6. IoT Analytics (May 2026), “US manufacturing reshoring boom: what the data says”: the post-2024 cooldown — total manufacturing construction −21% from the June 2024 $239B peak; electronics −44%; ex-electronics +5.6% post-tariffs.
7. Factory Settings / Employ America (March 2026), “Did the CHIPS Act Trigger the Manufacturing Construction Boom?”: construction timing along the legislative arc; 30+ companies, 45+ project locations.
8. Semiconductor Industry Association / Octagon (2026): ~$460B private investment stimulated; CHIPS $52.7B appropriation, $39B manufacturing incentives; 12%→20% production-share goal.
9. CFR Investment Tracker (April 2026): $20.9B across 16 federal equity deals since January 2025; agency breakdown; U.S. Steel golden share; Nvidia/AMD revenue-sharing.
10. OMFIF (August 2025): the Intel equity-for-grants transaction ($8.9B → 10%); MP Materials ($400M DoD stake, 100% offtake, $1B private crowd-in); EO 14196 (Feb 3, 2025).
11. Westside Construction Group / Manufacturing Dive (2026): 97 fabs launched 2023–2025; Micron’s $200B program and $100B Clay, NY megafab groundbreaking (Jan 2026).

Methodological note: Figure 1’s Hamilton Index is a stylized composite pending the full historical construction in Article III; weights and component series will be published with code. Figure 2 uses approximate annual rates from the Census VIP survey as reported by the cited analyses. The crowd-in ratio ψ attributes stimulated investment per the SIA tally and should be read as an upper bound on causal leverage; Article II addresses attribution formally. The MP Materials operational status reflects reporting through early 2026. Dollar figures nominal. Code: github.com/vector1-ai (forthcoming).

Brian Curry is a Kansas City–based data scientist, AI researcher, and founder of Vector1 Research. The Hamiltonian Turn program develops the economic theory and measurement of America’s return to developmental statecraft. Vector1’s companion series, Returns to Data*, covers the economics of data ownership in the age of AI.*

Contact: brian at vector1.ai · License: CC BY 4.0

#HamiltonianEconomics #IndustrialPolicy #CHIPSAct #AmericanSystem #Manufacturing #SovereignWealthFund #EconomicHistory #Reshoring #LearningCurves #Vector1Research