After decades during which the very concept was taboo and the reality swept under the rug, the US has now rediscovered industrial policy. But this rediscovery remains incomplete, as too much of it is reactive, scattered, and driven by whichever crisis has the attention of the moment. This ad hoc approach can win a few economic skirmishes, but it will not be enough to win Cold War II.

Simply scaling up more and more ad hoc policies, the trajectory we are now on, will never produce a coherent strategy. This is because what counts is not the number of policies, their aggregate budget, or even how widespread they are. What counts is whether they fit together, effectively support each other, and serve rational overall industrial policy goals.

The biggest industrial policy issue now facing the country is therefore converting our scattershot, mostly defensive improvisations into a coherent national industrial strategy.

How to do this is, of course, an ongoing question we will be answering for decades to come. So we will not address it here. Instead, we will review the dangers of not doing so, to generate the required sense of urgency.

Danger #1 of Ad Hoc Industrial Policy: Never Getting Beyond Crisis Response

A huge percentage of our existing policies have derived from a crisis mentality. Fear the collapse of the auto industry? Organize a bailout. Fear a pandemic? Accelerate development of a vaccine. Fear losing the AI race to China? Embargo the most powerful chips.

These policies were necessary and we should continue to respond similarly to similar threats. But this approach suffers from two problems that preclude it from being a template for a successful overall national industrial policy.

First, not every problem meriting an industrial policy response takes the form of a looming catastrophe. Sometimes the danger is just decades of suboptimal performance in a particular industry, such as autos – with all its negative implications for worker incomes, corporate profits, tax revenues, technological progress, productive capacity, the trade balance, and regional decline.

Second, some industrial policy needs do not derive from possible negative outcomes at all, but from opportunities that America would benefit by seizing. Some of most important technologies of the future will require years of government-funded nurturing before they mature enough for the private sector to take over completely.

Here’s an article on key past inventions that could hypothetically have been invented earlier. Whether state support could have helped accelerate their arrival is a case-by-case question, but the point stands: We risk delaying valuable innovations, or waiting for them to appear in another country, which will then reap their commercial benefits.

Danger #2 of Ad Hoc Industrial Policy: Solving Only the Obvious Problems

An ad hoc approach risks addressing only sectors where the problems are obvious. These will get the tariffs, the subsidies, and White House attention. Other sectors will languish.

Consider how in recent years vast industrial policy attention has been lavished on new technologies that everyone is aware of, like electric cars under Biden or artificial intelligence under Trump, while it has not been given to less glamorous technologies that may prove almost as important in coming decades, such as new materials. (If you don’t even know what these are, that’s precisely our point.)

Or consider the machine tool industry, which has been in decline in this country since the early 1980s but has received little policy attention. Machine tools operate behind closed doors in factories, so the average voter is barely aware of their existence. But they are foundational for every other manufacturing industry.

Industrial robotics is another important, underpublicized field where the US has lagged due to a lack of industrial policy attention. Washington spent billions on military and space robotics, but almost nothing on industrial applications. Meanwhile, rivals like Japan, Korea, and Germany allocated huge resources to systematic strategies to build strong positions in the industry.

As a result, there is not a single US-owned and -located firm in the industry’s top 10, and America’s weak robotics industry and limited corporate experience deploying them is now a handicap to the reshoring of other industries.

Danger #3 of Ad Hoc Industrial Policy: Waiting Too Long

Many industrial vulnerabilities are not obvious until it is either too late, or far more difficult and expensive to remedy them than if they had been detected earlier. In many industries, this sort of “frog boiling” is precisely how the US got into its present predicament.

China (and Japan decades earlier) repeatedly exploited this dynamic. These nations’ industrial-policy strategists knew they could rely on the progression of American perceptions from “America’s X industry is too strong to need help” to “America has only lost the low end of the X industry ” to “America’s X industry is collapsing but we should leave it to the market” to “America’s X industry is too far gone to save, so no policy action is justified.”

The core problem here is that the loss of industrial capabilities and their recovery are asymmetric processes. Industries can be killed off very fast if starved of sales revenue by predatory foreign competition. But they will not just spring back up quickly if conditions change.

Most important industries have a large minimum viable scale, which means the investment required to jump-start them from scratch is huge. Also, capital investments, supplier networks, skills bases, customer relationships, and financing channels interlock, often presenting “chicken and egg” problems. Specialized suppliers may have lost their anchor customers and gone broke, leaving nobody in the US with the right skill set and production equipment. Skilled workers will have retired, switched sectors, and never trained their successors. Engineers have moved into other fields, their former expertise has decayed, and they never learned the current state of the art. Customers have redesigned their procurement around foreign suppliers.

Early on, while US production still exists, relatively simple, low-cost measures such as tariffs can suffice to bring an industry back to health. Later on, heavier handed, riskier, and more expensive measures, such as subsidies, governmental equity infusions, and guaranteed purchase agreements, can be required. Rare earths are a good current example of such a package that is actually now being implemented .

It follows that even skeptics of industrial policy should support a proactive “nip industrial problems in the bud” approach. But this requires three things:

1. Systematic governmental surveillance of the technological and financial health of American industries.

2. Systematic understanding of how industrial policy works, what it can do, and how it should be implemented.

3. Consensus in favor of taking action sooner, rather than later.

Danger #4 of Ad Hoc Industrial Policy: A Lack of Systems Thinking

Successful industrial policies, even if they don’t intervene in every part of an industry, must still reflect the fact that these parts do not exist in isolation. American industries generally thrive when either of the following is true:

A: Their entire supply chain is intact in this country and in good health, with (usually multiple) competitive firms able to produce at viable scale and with sufficient profitability to pay good wages and attract sufficient capital to stay technologically current.

B: It is feasible for the US to “filet out” the most desirable parts of the industry’s supply chain. This generally means the parts with the highest value-added per man hour and return per dollar invested, and must usually include the “chokepoint” steps, i.e., those that can potentially block all the rest.

Industries not only have internal structures, but exist in larger ecosystems of suppliers and customers outside the industry itself. They are often critically dependent upon resources shared with and supported by other industries. For example, in the words of Gary Pisano and Willy Shih of Harvard Business School, who call this phenomenon the “industrial commons,”

Software knowledge and skills, for instance, are vital to an extremely wide range of industries (machine tools, medical devices, earth-moving equipment, automobiles, aircraft, computers, consumer electronics, defense). Similarly, capabilities related to thin-film deposition processes are crucial to sophisticated optics; to such electronic products as semiconductors and disk drives; and to industrial tools, packaging, solar panels, and advanced displays. The knowledge, skills, and equipment related to the development and production of advanced materials are a commons for such diverse industries as aerospace, automobiles, medical devices, and consumer products. Biotechnology is a commons not just for drugs but also for agriculture and the emerging alternative-fuels industry.

Industries depend not only on their production ecosystems, but also on their innovation ecosystems. These are the often-long pipelines, whose structure varies by industry, that bring innovations from scientific discovery to marketable product.

Gaps in these pipelines can result in America producing the underlying science while other nations reap the commercial benefits, as happened, for example, with liquid-crystal displays, photovoltaic cells, lithium-ion batteries, DRAM chips, rare-earth magnets, LED lighting, OLED displays, and semiconductor lasers. Or American firms can pioneer the small-batch, high-cost, cutting-edge initial version of a product, while for lack of scale-up capability, the mass production migrates abroad, as happened with VCRs and many of the aforementioned items.

Danger #5 of Ad Hoc Industrial Policy: Failing to Use All Required Tools

Industrial policy is often caricatured as just tariffs and subsidies. Mis-framed debates that evaluate the merits of tariffs in isolation (inevitably concluding against them) are a dime a dozen. But the industrial policy toolbox is much larger, and industrial policy can only be rightly evaluated as the success or failure of an attempt to deploy a package of tools.

This includes tariffs, quotas, local-content rules, stage-differential tariffs, procurement requirements, export controls, investment screening, publicly funded R&D, tax credits, loan guarantees, federal equity stakes, technological standards-setting, workforce development, infrastructure investment, local and federal permitting reform, regional cluster cultivation, technology extension services, currency management, and demand guarantees.

These tools are most effective in combination. Many are useless or even counterproductive when used alone, because they depend upon other conditions obtaining, or create incentives that can be responded to in both good and bad ways. For example:

1. A tariff without antitrust enforcement can produce a complacent industry.

2. An investment subsidy without tariff protection can finance capacity that gets undercut by imports.

3. Technology development without domestic production incentives can generate inventions that get commercialized abroad.

4. Workforce training without support for the employing industry can train people for jobs that don’t exist.

5. Government procurement in isolation can buy overpriced “hothouse flowers” without rebuilding an industry capable of mass production for the non-governmental market.

6. Local and federal permitting reform without strategic targeting can accelerate low-value projects while leaving critical bottlenecks unresolved (examples here, here, here, and here).

Consider aluminum. A tariff has improved the economics of existing domestic producers, but because they compete with foreign producers enjoying subsidized electricity, this has not resulted in any new primary smelting capacity in the US. The closest the US has gotten is one unbuilt project in Oklahoma that is still pending a solution to the power problem.

We don’t necessarily need great depth of policy intervention in every industry. But we do need, in every industry, to systematically decide what is and is not needed.

Danger #6 of Ad Hoc Industrial Policy: Failing to Coordinate Tools

Constructing America’s industrial policy package on ad hoc basis just begs for multiple failures to coordinate policy tools. Such failures have already occurred.

For example, the US now has tariffs in a wide range of industries, for an average calculated as between 8 and 10 percent. But we also have a currency overvalued at estimates ranging from 12 to 17 to 19 to 20 percent. It follows that, whatever the precise numbers, we do not on net and on average have a tariff at all, despite three successive administrations that embraced tariffs. (And, of course, there’s also the economic headwind for America’s exports created by the overvalued currency.)

Our currency problem has been acknowledged by Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, Chairman of the Council of Economic Advisers Steve Miran – and even by President Trump himself here, here, here, here, and here.

The solution, a moderate, variable tax on foreign capital flows into the US known as the Market Access Charge (MAC), is known. The US and multiple foreign nations have done similar things in the past. All our major trading partners make at least some effort to actively manage their currencies. But nothing has been done.

Predictably, given the fact that absent currency management, tariffs tend to induce currency appreciation (though not necessarily a full offset), this has doomed any hope of rebalancing America’s overall trade. America’s trade deficit in 2025 was thus a higher percentage of GDP (2.9%) than in 2016 (2.7%), when Trump was initially elected in large part on the strength of voters’ belief that America’s trade had been mismanaged. The resulting trillion dollars of missing demand for domestically produced goods remains a bar to America’s reindustrialization.

As another example of coordination failure, consider current industrial policy in AI. This industry has received an ambitious package including a federal framework to coordinate AI R&D, federally funded university-led AI research institutes, subsidized domestic semiconductor fabrication, export controls on advanced AI chips, federal coordination on AI safety, fast-tracked federal permitting for AI data centers, federal procurement of AI models, and proactive AI adoption in national security.

However, the Trump administration has also, in the face of soaring demand for electricity, pulled the plug on multiple green sources thereof, most capriciously but not only offshore wind, where it spent nearly $3 billion to compensate firms for terminating already-permitted projects. China, while no friend to the environment in its electricity generation policies, has pursued a source-agnostic “all of the above” approach.

At a minimum, the US should not have industrial policies operating at cross-purposes to each other. Ideally, policies should strengthen and support one another.

Danger #7 of Ad Hoc Industrial Policy: Irrational Trade-Offs

Ad hoc policymaking will never be able to handle trade-offs honestly or efficiently. This is critical, because industrial policy at meaningful scale inevitably imposes significant costs on the rest of the economy. Whether any given policy is worthwhile depends upon understanding these costs and their corresponding benefits, including non-economic benefits such as national security, public health, or environmental protection.

One of the classic ways a nation can fail at industrial policy is to see only the benefits of any given policy, such as standing up an industry domestically, without counting the costs. This remains true even when these costs are widely diffused, poorly understood, or otherwise pass unnoticed.

For example, tariffs, at the modest levels the US has thus far imposed them, have not proven to induce overall inflation upon detailed examination of the data. But they can certainly produce price increases in individual products (though sometimes this is only a temporary spike). A similarly complex pattern holds for producer, as opposed to consumer, goods.

Whether these potential price effects are worthwhile depends on the value of the industry thereby saved domestically or reshored. The analytical wrinkle is that, contrary to oversimplified theory and the expectations of most economists, a straight 1:1 relationship between tariffs and price increases is not empirically what happens, due to competition, sunk costs, demand and supply elasticities, corporate pricing strategies, and other factors. This complex reality will need to be part of any weighing of any future protectionist measures.

Another example of difficult trade-offs: Restricting exports of advanced technology will indeed handicap rivals who need that technology to build their own industries. But it will also reduce revenues for the restricted American exporters, making it harder for them to afford to remain at the technological frontier. And, of course, it will encourage the rival to develop its own technologies.

Another example: Trade barriers don’t only raise buyers’ costs. They can diminish product variety and reduce the competitive pressure that, paradoxically, benefits the domestic industry by driving it to excel and innovate.

And subsidizing one technology path may crowd out another that would ultimately have proved superior.

Danger #8 of Ad Hoc Industrial Policy: Rational Decisions Require Institutions

For government to make the above choices with adequate information, correct logic, and public accountability requires institutions. It cannot be handled like a succession of repeated one-off projects like presidential campaigns.

Industrial policy needs to become a standing national mission like defense, energy, transportation, or environmental protection. It requires permanent analytic capacity, technical expertise, and institutional memory. Among other things, this means a cadre of career officials.

The authors’ proposal for an Industrial Policy Council (IPC) modeled partly on the National Security Council and National Economic Council is one possible way to institutionalize these capacities. The IPC would coordinate agencies and require the government to examine industries, supply chains, competitor policies, trade balances, capital flows, and domestic capabilities as an integrated whole. It would produce an indicative multi-year industrial strategy, “indicative” meaning that the strategy would not embody prescriptive authority, would not be intended for adoption as a single bill, and would not claim to be an exhaustive description of US industrial policy.

If industrial policy is left entirely to the executive branch, it will be unstable, insufficiently vetted, and vulnerable to capture, so Congress will also need formal, expert decision-making capacity. Its committees will need staff who understand industrial organization, trade, technology, manufacturing processes, finance, and supply-chain structure. The old Office of Technology Assessment should be revived or replaced with something stronger.

Government agencies that will actually implement industrial policies, ranging from the Treasury to the NSF, will also need staff specializing in industrial policy.

None of this should aim at “depoliticizing” industrial policy or making it something like the process by which the Fed sets interest rates or the FDA approves drugs. That would be impossible, because:

1. Industrial policy is an insufficiently deterministic body of knowledge to be reduced to a standardized routine.

2. Industrial policy involves too many choices about ends, as opposed to means, are inherently political.

This would also be undesirable, even if it were possible, because this is a democracy. Instead, the objective should be to make political disagreements occur on top of a shared understanding of facts, tools, and goals.

Danger #9 of Ad Hoc Industrial Policy: Lack of Legitimacy

Industrial policy at sufficient scale will require true political legitimacy, not the mere passive public acceptance that we have today, which apparently suffices for a bundle of scattershot policies.

If voters are to accept the costs of industrial policy, they are going to need to understand how they will benefit. Selling Biden’s green energy initiatives as climate policy, rather than support for an inevitable technological transition designed to protect the competitiveness of American industry, was probably a mistake.

Legitimacy means, as with any major government policy, acceptance of the need for it as a general proposition, so that people are not put off the whole idea when the inevitable failures occur.

In a technocratic democracy like the US, legitimacy must hold with both Washington policymakers and with the electorate at large.

Individual voters will not need to become experts in semiconductor lithography, shipyard finance, or pharmaceutical precursors. But they will need to understand, albeit in the simplified way mass electorates do, certain basic ideas:

1. It matters what industries, and what economic activities within them, exist in the US, because this will determine whether good jobs will be available for them and their fellow citizens.

2. Markets are great, but they’ve never been the whole story, not in the US or any other successful country. Successful nations use a long list of other policies.

3. Trade is a good thing for many reasons, but foreign nations will take advantage of us if we let them. So America needs tariffs and a competitive dollar.

4. Technology doesn’t just come from Silicon Valley entrepreneurs, but from a long road many of whose steps need to be executed by the government.

Without such understandings, voters will have to choose between presidential and congressional candidates who claim to support sound industrial policy, but without any basis upon which to judge. Both demagogy’s appealing but ineffective solutions and the phony expertise of “experts” will have free rein.

Without entrenched public appreciation of industrial policy, the lazy generalizations of “picking winners is wrong” on the right, and “helping corporations is corporate welfare” on the left could potentially combine to snuff out industrial policy entirely.

The fact that the average voter realized America’s “free trade” policy was a mistake long before the experts in government or academia did should serve as a reminder not of the superior wisdom of the masses, but of the fact that a successful democratic society needs policy wisdom (wisdom, not expertise) at both the popular and elite levels.

But voters don’t bother to understand isolated fragments of things. They bother to understand things they view as important, and they form a gestalt impression of the issue. For this to happen, industrial policy must form a whole in their minds.

Danger #10 of Ad Hoc Industrial Policy: Lack of Discipline

Industrial policy must include discipline. The state must have the capacity not only to support industry, but to demand performance from it. Firms receiving protection or support should be expected to invest, upgrade, train, export where appropriate, and meet measurable milestones, such as market share and job creation.

Industrial policy must not become an open-ended subsidy machine or get stuck in the rut of programs continued out of mere institutional inertia. Programs will need clear objectives, performance benchmarks, sunset provisions or periodic reauthorization, independent evaluation, and the ability to terminate failures.

The US is quite capable of doing this. In decades past, the US successfully terminated huge programs for synthetic fuels and a supersonic transport when their usefulness or feasibility became dubious. This sort of behavior is most likely to become standard operating procedure when industrial policy is systematic, i.e., when voters and politicians have expectations.

An entrenched public sense of the legitimate objectives of industrial policy will also help discipline against the hijacking of economic objectives by extraneous political ones, such as Trump’s retaliatory tariffs on Brazil for its prosecution of its former president or on India for buying Russian oil. The same goes for worthy but peripheral social-policy objectives, such as the affordable childcare required by the CHIPS Act.

As shown by Trump’s termination of many of Biden’s green energy initiatives, policies depending upon the political will of individual administrations are vulnerable to abandonment and failure. A better grounding in public opinion would also militate against such policy flip-flops, which in this case have thus far cost the auto sector alone $52 billion in write-downs (source, source, source).

A solid grounding in public opinion can only come from the public being explicitly told, over and over again, by America’s highest leaders in both parties why the US is embracing industrial policy. Said explanation will need to be comprehensive because, as argued above, only a comprehensive industrial policy truly makes sense or will hold the public’s attention.

Conclusion: This is Required for Industrial Policy to Survive

If America doesn’t get truly systematic about its industrial policy, there is a grave danger the current trend will stall out. As the authors documented in their book Industrial Policy for the United States: Winning the Competition for Good Jobs and High-Value Industries, movements in favor of proactive industrial policy have stalled out before. Nixon was secretly researching one in the early 1970s. In the early 1980s, the decline of American competitiveness produced a serious debate about industrial strategy. But it then fizzled. Bill Clinton campaigned in the early 1990s like he was going to impose one, but in office he did not. We cannot afford to make this mistake again.

The Fed has now joined the IMF, the World Bank, and other pillars of the neoliberal establishment in acknowledging, after decades of denial, the potential of industrial policy. Industrial policy is increasing in industry after industry in this country, with more almost certainly to come.

The need now is to transcend this scattering of ad hoc measures and make US industrial policy comprehensive, coherent, institutionalized, and a fixture of our national policy consensus. Otherwise, the industrial policy revival may ebb like previous episodes of interest in the early 1980s and early 1990s.

The consensus we need is not a rigid doctrine or an ideology – indeed it would be a disaster if it became one. Instead, that consensus should contain room for partisan differences, accommodate debate about specifics, and incorporate independent reviews to weed out failed programs. It should be flexible enough to adapt over time to changing technologies, markets, and foreign-country policies.

This consensus must be grounded in a generally agreed understanding of why industrial policy is desirable, what its major elements are, and why they need to be coordinated. Industrial policy for economic purposes should be based on a rigorously defined vision of what economic objectives are worth achieving, why markets cannot realize them on their own, and why industrial policy can sometimes help. Often these policies will also serve other important objectives such as public health, national security, or environmental protection. When there tradeoffs, they should be made with a full understanding of costs and benefits.

We have explained our version of this economic foundation here. In a nutshell, industrial policy should be about the pursuit of what we term “advantageous” industries. These industries have the scale, pricing power, and technological dynamism to be the basis of a high-wage, high-profit economy. They are often rife with the non-market dynamics that open the door for policy interventions to add value. American industrial policy should seek to establish or retain leading positions in almost all such industries: positions of size, technological sophistication, and sustained profitability matching or exceeding the best foreign competitors.

The policies America will need to achieve this fall into three broad categories:

1. Managing international flows of goods, services, and capital with tariffs, quotas, capital controls, and related policies.

2. Managing international flows of capital with moderate, variable capital controls to bring the dollar to a trade-balancing value.

3. Nurturing the entire innovation pipeline, from pure science to production of commercial products at scale, to fill in gaps where markets cannot suffice.

The US needs not just new laws, but new standing authorities with the ability to maneuver and shift gears when needed. Now is a uniquely fluid moment to start establishing the practices and building the institutions we will need for decades to come.

Reflecting Political Realities

Political power in this country is dispersed: between the president, the courts, and congress; within congress itself; and between the states and the federal government. Social solidarity is weak due to regional, cultural, and political diversity. There are few limits on interest groups’ ability to bend policy to private ends.

American industrial policy will therefore never be as smooth as in, say, Japan, with its strong social solidarity, preference for consensus, and deference to technocrats. So, there will never be a single official, agency, or committee able to impose a centrally directed and prescriptive US industrial policy. Instead, policies from trade to technology to education will have to be developed through our existing multi-player political process.

But for the result to be coherent – or sufficiently coherent, as industrial policy doesn’t need to be perfect any more than any other kind of policy – these players will have to behave differently than they have done in the neoliberal era. Successful industrial policy will require a shift of mentality at the federal, state, and local levels, one that is already starting to occur.

To keep all these disparate players pulling on the same rope, strong, consistent presidential is critical. Donald Trump and Joe Biden both believed, imperfectly but with breakthrough accomplishments, in proactive industrial policy, so we know this can come from either party.

Over time, the US should consolidate many of the existing offices whose functions will require coordination. In the longer term, for industrial policy to have staying power and efficiency, it should be institutionalized. (We discuss proposals for both below.)

Eventually, a dedicated corps of civil servants should exist for day-to-day administration, to maintain institutional memory, and to track ongoing changes in technologies, the strategies of other governments, and the competitive positions of US and foreign companies. As institutional design scholar Ganesh Sitaraman has written,

“This is not work that can be done on an occasional basis by politically appointed advisors who serve for a couple of years before their next gig.”

Industrial policy is no different in this respect from other major governmental missions such as defense, environmental protection, transportation, and energy. The US graduated from “no standing armies” to a professional military a long time ago.

What the President Should Do

Industrial policy will require coordination of many executive branch agencies and the congressional committees overseeing them. The only government official with the electoral mandate, legal authority, and broad purview to lead this effort is the president.

Therefore, all relevant White House staff, cabinet secretaries, agency heads, and members of the Council of Economic Advisers should be chosen, in part, for their commitment to industrial policy as a core component of the administration’s economic program, and have the skills and experience to contribute.

The President should establish a new Industrial Policy Council (IPC) modelled on the existing National Security Council and National Economic Council. It would coordinate the work of the executive branch agencies while consulting with congressional leaders of both parties.

The IPCs diverse membership would ensure that all relevant administration voices are heard and that its policy choices are understood and accepted. It should include the Secretaries of the Treasury, Commerce, Defense, Energy, Labor, and Education; the US Trade Representative; the Directors of National Intelligence, the National Science Foundation, the National Institutes of Health, and the National Institute of Standards and Technology; and the Assistant to the President for Science and Technology. It should also have high-level participation by the Council of Economic Advisers and the Council on Environmental Quality.

The IPC’s staff should have expertise in the many policy areas that will require coordination, including technological research, manufacturing, commercial lending, and venture capital. But to reduce the risk of capture by private sector interests, the IPC’s director should not come from the business community.

Industrial policy offices should be established in each department and agency whose head serves on the IPC. The chiefs of these offices should report directly to the departmental secretary or agency head and liaise with the senior staff of the IPC. To support coordination with the NSC and NEC, select staff from these councils could serve jointly on the IPC.

Of course, other staff structures are possible, so long as they include institutionalized cross-disciplinary expertise and leadership with direct access to the president.

What the Industrial Policy Council Should Do

The IPC would produce an indicative multi-year industrial policy for the US. “Indicative” means that it does not embody prescriptive authority, is not intended as an exhaustive description of US industrial policy and is not designed for adoption as a single comprehensive bill. Instead, it would be analogous to the existing National Defense Strategy drawn up by the Defense Department, laying out fundamental realities, objectives, and the means of achieving them.

This document would provide a shared vision and a division of responsibilities for the agencies involved. It would offer a coherent baseline for congressional oversight. And it would communicate the US government’s intentions to key audiences at home and abroad. The president could direct the production of this document by executive order, but it would have more impact if authorized by and delivered to congress.

It should include:

1. A list of advantageous (in the sense the we explained here) and otherwise economically critical industries, such as those (rare earths) that serve as chokepoints for many others..

2. A list of those industries critical for the major non-economic industrial policy objectives: national security, public health, and environmental protection.

3. An assessment of where the US stands in each of the above industries, including the percentage of its domestic market, including supply chains, held by imports.

4. An assessment of the same for major competitor nations.

5. Trade balances in these industries by sector, product and counterparty nation, with assessments of their effects on US employment and manufacturing capacity.

6. Statistics by counterparty nation on public and private inbound and outbound investment in these industries.

7. An assessment of the main industrial policy tools being used by major competitor nations in these industries, with evaluations of their effectiveness and their effects on the US.

8. An assessment of the policy tools being used by the US in these industries, also with evaluations of their effectiveness.

9. A report on currency manipulation by economically significant trade counterparties.

10. A report on the volume and composition of gross and net debt, and capital inflows (equity and debt) into the US, and on their impact on the undervaluation of foreign currencies.

11. Estimates of effect of different levels of taxation of incoming foreign capital on the US trade balance over a 5-year period

12. Proposals for new or changed policies and programs to meet the needs identified by the foregoing analyses.

The process of developing this document will be at least as important as its recommendations. The interagency consultations required will sharpen awareness of the interconnectedness of, and tradeoffs between, different elements of industrial policy and help establish channels of communication and protocols for coordination.

What Congress Should Do

Congress should develop its own nonpartisan capacity for industrial policy analysis. For example, using the US-China Economic and Security Review Commission as a model, it should establish a commission with members from the private sector appointed by the majority and minority leaders of the House and Senate. This should be supported by a small permanent staff with the capability to model the economic effects of existing and proposed technology, trade, and exchange rate policies.

Expert staff focused on industrial policy, who should liaise with the IPC, should be added to all relevant Congressional committees. The Congressional Research Service should add staff experts in industrial policy. The 1972-1995 Congressional Office of Technology Assessment should be reactivated. Project Socrates, a technology analysis platform developed in the 1980s by the CIA and DIA, should be restored.

The National Institute of Standards and Technologies should establish an office to monitor and analyze American and foreign innovation capabilities in key industries. Its output should include a detailed digital map of the technological capabilities of companies, universities, and government labs around the world, including current and potential supply chains.

The House Ways & Means Committee and the Senate Finance Committee control tax and trade policy. They are thus deeply involved in developing and enacting industrial policy legislation, and therefore should have staff with an understanding of and commitment to industrial policy.

Despite agency turf-guarding and Congress’s sensitivity about committee jurisdiction, large-scale governmental restructurings have taken place after major crises, most recently post-9-11. Therefore, when the political stars align, many of the agencies and functions that should be coordinated for a maximally effective industrial policy should be consolidated.

With its many existing industrial policy-related agencies, the Commerce Department is the most suitable center for US industrial policy. It should therefore take on the Small Business Administration, the Trade Adjustment Assistance program, the State Department’s several trade bureaus, the Trade and Development Agency, the Export-Import Bank, and the International Development Finance Corporation.

A new Economic Security Agency in the Commerce Department should assume the existing fragmented controls over sensitive exports, technology transfers, and outbound investments. It should take on the Committee on Foreign Investment in the United States, State’s Directorate of Defense Trade Controls, the Bureau of Industry and Security, and the Office of Foreign Assets Control.

The US Trade Representative should also become part of Commerce, but with the USTR retaining his or her cabinet position. Commerce should have an Office of Policy Planning that, like the State Department’s equivalent, reports to and advises the Secretary.

The economic intelligence functions of the Central Intelligence Agency, Defense Intelligence Agency, National Security Agency, and Treasury Department should be centralized in a new Economic Intelligence Agency.

General Guidelines for Industrial Policy

Industries should be supported not only by economy-wide policies, but also by programs tailored to their specific needs. Industries’ internal economic and technological structures differ, how much help they merit differs, and what helps them differs.

Industrial policy programs should be assigned to agencies with the technical skill, underlying inclination, and statutory missions to execute them. The US has a diversity of governmental institutions and program types for good reason.

Programs should be lodged at the appropriate levels (sometimes more than one) of government. For example, however welcome state-level support may be, nurturing internationally competitive regional clusters requires the money and policy authority of the federal government, because only it can, for example, impose tariffs to protect the cluster’s market.

To ensure buy-in and commitment, funding responsibility should ideally be split among the parties benefitting from the program. Often these parties will be the federal government, private industry, a state, and a region or municipality. State and local commitment should be important criteria for determining where place-specific federally funded programs are sited.

Time scales should be appropriate. Regional clusters usually take decades to become internationally competitive and self-sustaining. Development projects applying existing technologies to specific commercial uses sometimes take only weeks. A sense of urgency has helped agencies like DARPA, but a long-term view has been essential for agencies like the NSF.

Projects involving technologies close to or already at the stage of commercialization, should have the involvement of, and co-direction by, the private sector. It generally has the expertise and the right incentives for this stage, and the government generally does not.

Programs should reflect structural differences between the industries they nurture. For example, hard-science-intensive manufacturing, such as new materials, advanced semiconductors, renewable energy systems, and large aircraft, requires more capital than software development.

Which initially promising technologies will pan out is inherently uncertain for radical innovations, so multiple technology paths should be supported. This process should be disciplined by tools, such as required performance levels, to enable timely culling of failing approaches.

Programs should have structured protocols, with consistent, albeit evolving, criteria for selecting, modifying, and continuing projects. Programs should draw on business and technical experts from both inside and outside government. For pure science and science-intensive technology development (not deployment), peer review should be used.

The “eternal life” of government programs should be preempted by building in termination dates or periodic reauthorization reviews. But these reviews should not be so frequent as to deprive programs of the stability needed for long-term investments and credible commitments from the private sector. There should be funding for effectiveness assessments.

For programs for which appropriate criteria of success were established at the outset, government should stick to them. Conversely, programs should not be terminated on inappropriate grounds such as a failure of precompetitive technologies to turn a profit, a failure to hit budgetary estimates in the face of scientific and technological uncertainty, or a failure to consider indirect benefits such as environmental protection.

Audit and oversight to reduce both hard and soft corruption should be funded, because the close collaboration between government and business required for many industrial policies to work increases the risk of political capture by the industries involved.

Programs should be held accountable for their performance at the aggregate level, but insulated from political interference at the detail level. Some projects of even the best programs will fail, so programs should be judged by the success of their overall portfolios.

Marc Fasteau and Ian Fletcher are the coauthors of Industrial Policy for the United States: Winning the Competition for Good Jobs and High-Value Industries (Cambridge University Press, 2025). Both are connected with the Coalition for a Prosperous America. There is more about Marc on his personal website here, and about Ian on his personal website here. Their book’s website is here.

The revival of proactive industrial policy in this country since 2017 has centered on tariffs (universal, country-specific, industry specific), subsidies (e.g. the CHIPS Act), and federal equity investments (e.g. rare earth producers). But because these are only a few of many industrial policy tools and they are most effective in combination, it’s worth reviewing the others. Some listed below are of low direct relevance to the US, but are still worth understanding, as other countries use them against us, and we need to respond effectively. To wit:

Infant-Industry Protection means protecting emerging industries from foreign competition so domestic sales will support growth of the nation’s own industry.

This policy is only worthwhile if it pushes the nation into more-advantageous industries. There is a risk of launching industries that remain eternally dependent on tariff protection, and undermining downstream industries by cutting the them off from state-of-the-art imported goods and services.

If protected firms just pocket the profits, instead of investing them in upgrading their capabilities, protecting industries will merely enrich their owners. Therefore, measures to pressure firms to upgrade are generally required. This requires a state that can sometimes say “no” to big business. This comes more easily to authoritarian governments, like present-day China or pre-democracy Taiwan and Korea, to highly technocratic regimes such as Japan, and to social-democratic governments like those in Europe.

Infant-industry protection needs a large enough domestic market, so this policy is less useful in small nations. This is why small countries, such as Singapore, that have had very aggressive industrial policies overall, have therefore tended not to be protectionist, while large countries, like China and the pre-WWII US, have been.

Right now, the US is applying infant-industry protection most visibly in electric vehicles, by imposing a prohibitive tariff on ones from China. But protection rarely works without supporting policies, and we have recently abandoned many that we had enacted, triggering massive write-offs in the industry.

Local Content Rules mandate that 50 percent, say, of a car must be made in-country. The advantage over a tariff on the entire product is that they can be adjusted over time to move more steps of the production process onshore as the nation’s producers upgrade their capabilities.

Historically, these rules were often attached to incentives for foreign direct investment (FDI) to create jobs. More recently, their purpose often has often included forcing transmission of technological knowhow.

The danger is that by driving up the cost of the finished product or compromising its quality, such rules can undermine its competitiveness both as an export and against imports. This is likely to happen if the nation hasn’t yet developed a sufficiently sophisticated supplier base – illustrating how industrial policies need to be pushed hard enough, but not too hard.

The US imposes local-content rules, albeit mildly. For example, the automobile section of the 2018 United States-Mexico-Canada Agreement (USMCA) requires a certain percentage of a car to come from the US, Mexico, or Canada to qualify for a lower tariff.

Stage-Differential Tariffs place, for example, a higher tariff on cotton made into thread than on raw cotton, and a higher tariff still on cloth woven from thread. Such duties were proposed by Alexander Hamilton in 1791 and had been used by European nations for at least 200 years before that. Their purpose is to push the economy out of raw materials production and into the more advantageous manufacturing-oriented, steps of supply chains. For example, China a few years ago had duties of 5.5 percent on raw lithium oxide, a component of lithium-ion batteries, 12 percent on battery parts, and 14 percent on completed batteries.

Import substitution aims to replace items currently imported. The advantage of targeting such goods is that these are often products on the next rung up of industrial sophistication, and the market already exists. The disadvantages are a) the policy focuses industry on the domestic market, which may not make demands as rigorous as overseas customers, and b) some goods will not be manufacturable domestically at acceptable cost within an acceptable timeframe. One example: in 2006, Beijing introduced its National Medium-and Long-Term Plan for the Development of Science and Technology (2006-2020), which sought to reduce China’s reliance on products containing foreign technology from 60 percent in 2006 to 30 percent 2020. (Although the metrics were changed, it basically succeeded.)

Selective Importation is the deliberate importation of product components and foreign capital goods, typically production machinery, by nations pursuing export-led growth. This policy often allows tariff-free importation of machinery and technology while forbidding or heavily taxing imports of consumer and luxury goods. Such restrictions conserve foreign exchange, often scarce in developing nations, for investments that upgrade the nation’s productive capacities. They also increase the national savings rate by tilting expenditures away from consumption, resulting in more investment capital. For example, after WWII, the Japanese government compiled a list – periodically updated to reflect the country’s rising living standards – of consumer goods deemed luxurious. And as of 2019, in Korea all imports of equipment for manufacturing computer chips but only 45 percent of computer chips were duty-free .

Export Subsidies and Targets The subsidies are about helping new firms succeed in difficult industries. The targets make the subsidy conditional on achieving a specified quantity of export sales, which avoids creating a “lazy” industry content with the easy profits from its protected domestic market.

Export sales are a simple test, and easy to enforce because foreign cash is hard to fake. In combination with tariffs or other protection, the policy generates the invigorating pressure of foreign competition without risking losing the industry to imports. Domestic firms are forced to become competitive in the global market, where the best in the world compete, setting the standard for technology, quality, and cost.

Export subsidies can not only nurture infant industries, but also allow mature industries to underprice and decimate foreign competitors. Most such subsidies violate WTO rules, but since any policy that reduces the cost of producing a product can serve, this restriction only tells against the most explicit ones.

Incentives for Foreign Firms to locate in a country and export include low corporate taxes, investment subsidies, preferential import tariffs for inputs, special export zones, and streamlining of bureaucracy.

Foreign firms are valued for their capital, technology, and distribution networks. Inducements also include investments in infrastructure and education in the targeted sectors, such as electronics in Southeast Asia and automobiles in Mexico and Brazil. America’s federal government has not offered such incentives, but individual states have, for example to German, Japanese, and Korean automakers, especially in the South.

Nations with internal markets large enough to create leverage, like Japan in past decades and China today, have required foreign firms wishing to sell there to form joint ventures with local companies to force the transfer of product and production knowhow. To skirt WTO rules, this tends to be implemented informally, rather than by law or written agreement.

To avoid getting trapped in a competition to offer the lowest taxes, lowest wages, highest subsidies, and laxest regulation, host nations must transition over time to competing on attributes other than price. For example, they can foster the accumulation of specialized skills by local workers and advanced capabilities by local supply chains, creating advantages that are difficult to duplicate elsewhere.

Export Processing Zones (EPZs) are geographic regions with special laws and policies designed to make them attractive for foreign industrial investment. Provisions include tariff exemptions on inputs used in producing goods for export and liberalized rules on foreign ownership. Some EPZs are individual factories, while others, such as the Chinese SEZs, are entire major cities.

By 2006, 130 countries had over 3,500 EPZs, with 66 million workers. They have been key growth drivers in, for example, China: In 2015, they accounted for 60 percent of its exports, 22 percent of GDP, and 45 percent of FDI. A less successful example is Mexico’s maquiladora plants along its US border. By 2000, these plants employed 1.3 million workers, and by 2013 produced 65 percent of Mexico’s manufactured exports. But due to weak supporting policies, the country has mostly remained stuck in mid-range processing and assembly, and its per-capita GDP has not significantly converged with that of the US. (It remains to be seen whether Plan Mexico will change this.)

Regulatory Competition means lowering regulatory burdens or expanding the range of permitted activities. In manufacturing, this generally means weakening environmental protections or worker health and safety rules. In services, strategies include low transparency and reporting requirements, compliance with which can be expensive. A more sophisticated practice is establishing effective legal governance frameworks to enable emerging industries to achieve sustained growth, rather than the boom, corruption, and bust that tends to accompany the absence of regulation. Online gambling is an industry currently in search of the friendliest regulatory environment, and AI regulation is still up in the air, with the EU having enacted rules in 2024.

Credit Allocation means directing capital to uses more conducive to the long-term growth of the economy than the market would choose.

In China, the state-controlled banking system is used to finance industries the Communist Party wishes to prioritize. In Germany, KfW (Reconstruction Credit Institution) has played a crucial role in post-WWII industrial development, providing long-term financing for expansion, innovation, and exports. Today, its Mittelstandbank focuses on small and medium enterprises (SMEs).

In many countries, industrial conglomerates have grown up around financial institutions owned by the conglomerates themselves. Such combinations, often buttressed by cross-shareholdings and shared directors, have been and remain a major factor in the growth of Japan, South Korea, Brazil, and Indonesia, among others. These conglomerates have all been actively or passively supported by their respective governments.

The US itself engages in credit allocation. The federal government subsidizes loans for housing (Fannie Mae, Freddie Mac, the mortgage interest deduction), small businesses (the Small Business Administration), universities (federal student loan programs), and exports (the Export-Import Bank). It de facto subsidizes financial spec-ulation by bailing out the financial sector and creating the expectation of future bailouts.

Forced Savings Policies aim to increase investment capital available to industry. Governments encourage or force people and firms to save money, especially in designated financial institutions and instruments. Policies include restricting opportunities to spend money (import restrictions on luxury goods, curtailed shopping days and hours, limits on non-business foreign travel, restrictions on currency convertibility), limiting opportunities to invest abroad (prohibitions on foreign banking or securities accounts), and incentivizing people to want large savings (a sparse social safety net and prohibiting home mortgages).

Singapore, for example, forces people to save a quota of their income for retirement and doesn’t pay significant unemployment benefits. In China, the state-owned banking system keeps returns on savings low to keep capital cheap and forces citizens to accept these low returns by limiting overseas investment.

Developing nations often prefer forced domestic savings to foreign capital because they fear losing control. Forced savings policies go hand-in-hand with trade surpluses and currency manipulation because someone other than the squeezed domestic consumer has to buy the output resulting from the increased capital investment. They are thus a hidden but significant cause of international economic conflict.

Currency Management is about the fact that a nation’s currency has a huge impact on its export competitiveness, its ability to avoid trade deficits, and in the long run, its international indebtedness. Therefore, most successful nations try to maintain an optimum value for their currency, i.e., one which balances the above concerns, which imply a weak currency, with others, such as servicing existing foreign debts and buying imports cheaply, which imply a strong one. Many, including China, mani-pulate the value of their currencies by controlling what their exporters can do with the foreign currency they earn. The US today fails to manage its own overvalued currency, and this is a major bar to rebalancing our trade and rebuilding our industrial base.

Sovereign Wealth Funds (SWFs) are large pools of capital owned and controlled by governments for funding investments and covering long-term liabilities such as pensions. The SWF Institute in 2023 estimated the total size of the top 100 funds at $11.5 trillion. (Total global wealth is about $600 trillion.)

In some nations, SWF money derives from overseas assets accumulated as a byproduct of currency manipulation. In others, it comes from petroleum exports. Countries invest these funds abroad because of limited domestic investment opportunities and because investing abroad reduces the currency appreciation that would otherwise undermine their export competitiveness.

Sovereign-wealth money is often used to support national-champion companies. For example, Singapore’s fund, Temasek, has successfully promoted state-owned champions across a wide range of industries, including shipping, airlines, telecommu-nications, shipbuilding, and financial services, many of which have become dominant regional players. In recent years, China has become the most aggressive country in using its sovereign wealth to promote exports by such means as China State Construction and Engineering Corporation (CSCEC). A state-owned and -financed company, it is by far the largest construction firm in the world. It is at the forefront of China’s Belt and Road infrastructure initiative to open up Eurasian markets for Chinese exports.

Government Procurement is the purchase of goods and services by government to create demand. Typical areas include health, education, public works, energy supply, public transport, military hardware, and telecommunications.

In developing nations, government is often the only entity large and sophisticated enough to buy products at sufficient scale to launch domestic producers. It can also promote new technologies ahead of private-sector demand: In China, government fleet purchases jumpstarted the electric vehicle market.

Since WWII, the US government has played a significant role as a lead buyer of technologies and services ahead of private sector demand. The early stages of most of America’s high-technology industries of today were created by federal investments in basic and applied research followed by large-scale government procurement. For example, the military was a major early buyer of transistors, integrated circuits, and jet aircraft.

State Entrepreneurship is not socialism, with its renunciation of the profit motive, but the state as a risk-bearing, profit-seeking capitalist. Consider these examples:

China’s State-owned Assets Supervision and Administration Commission (SASAC) is now the single largest economic entity in the world, with the 96 companies under its control having combined assets of $10 trillion in 2020.

In 1968, for example, the Korean government founded a state-owned steel manufacturer, Pohang Iron and Steel, with the help of US and Japanese investments and technology. It has since become a world leader, and its price and quality have been an advantage for Korean carmakers. POSCO was eventually taken private, with foreigners buying 50 percent, in 1998.

Dutch State Mines was established in 1902 to mine Dutch coal because the nation’s existing coal industry was foreign owned. After 1960, it phased out coal production and transitioned to distributing Holland’s recently discovered natural gas and developing its petrochemical industry. In the 1990s, it transitioned again, out of commodity chemicals and into fine chemicals, materials, biotechnology, and food ingredients. Privatized in 1989-1996, it is today an innovative, diversified global company employing 23,000.

India’s state-owned pharmaceutical firms have had government support since the 1950s. Their science base has been supported by government research institutes. An intellectual property regime that did not recognize foreign patents allowed them to manufacture drugs designed and tested elsewhere. Protected from imports and subsidized, they were able to invest in R&D and accumulate production skills. Today India is the world’s largest (though sometimes unsafe) supplier of generic medicines, accounting for roughly 20 percent of global output.

The Brazilian government founded Embraer in 1969 as part of a strategy to develop the country’s aircraft industry, and today it is the world’s largest builder of regional jets. Public procurement, government-established engineering schools, research and training centers, as well as financial support from the Brazilian development bank BNDES, allowed it to survive in a highly competitive and capital-intensive industry. Today it is a privatized company, but with the government retaining a “golden share” conferring veto power over key decisions.

National Champions are firms supported by their governments to be industry leaders. In many advantageous industries, only firms as large as the global leaders can be competitive.

At worst, such policies have produced bloated failures soldered together from incompatible companies. Examples include the automaker British Leyland, whose final rump successor went bankrupt in 2005 after decades of decline, and the still-extant French computer firm Machines Bull, which, like Italy’s Olivetti, struggled to compete with IBM in the 1960s and 70s.

But the policy has worked elsewhere. Saudi Basic Industries Corporation is one of the world’s largest and most profitable chemical firms. Dubai’s DP World is among the world’s biggest operators of ports and cargo terminals. Firms as diverse as Hyundai, Samsung, Siemens, Embraer, Volkswagen, and LG have been supported as national champions at various times in their histories. In large passenger aircraft, a multi-nation, half-century effort has created Airbus, the European multi-national champion that has taken roughly half the global market.

Many Chinese national champions are among China’s largest firms, and several are state-owned enterprises (SOEs). In 1998, Premier Zhu Rongji began reforming the SOE sector, including a major push to turn the largest into champions that could dominate the domestic market, compete on the global stage, and be instruments of China’s national interests abroad. American companies competing against these firms are, to say the least, not on a level playing field. China’s strategy emphasizes maximizing assets, rather than profits, and the ability of these firms to ignore the profit imperatives under which US companies operate is a significant advantage.

Imposing Competitive Industry Structure means driving excellence by enhancing competition.

For example, Korea, during the 1961-1987 heyday of its proactive active industrial policies, usually made sure there were at least three or four chaebol conglomerates competing in each sector. Taiwan fostered hundreds of small firms in electronics so that a few large ones could emerge from the pack. Singapore, Malaysia, and China have all ensured that multiple foreign multinationals competed against each other on their soil.

Economies as large as the US, the EU, and China can usually have firms of global scale and competitive internal markets, as they can support multiple competitors in almost every advantageous industry. And many of their single-firm industries will face strong international competition, as with Boeing and Airbus.

The US has traditionally used antitrust to ensure competition in industries where competition is feasible, and utility regulation to restrain natural monopolies where it is not. Since 1980, changed doctrine (read “Robert Bork”) and less enforcement has led to greater concentration in industry after industry, and some public utility regulation has been eliminated outright.

Concentration caused by abusive, anticompetitive conduct is very different from that caused by scale economies and technological dynamism. Antitrust is complicated by the fact that many industries are global. A concentrated domestic market may thus be irrelevant, and membership in a global oligarchy that extracts rent from nations outside it may be in the national interest.

Geographic industry clusters are often extraordinarily successful, so governments foster them.

Transaction costs are lower with local suppliers. Clusters facilitate complex communication between firms at different stages in supply chains, easing innovations requiring coordination across these stages. Clusters tend to develop support firms, such as specialized law firms, consultants, and venture capitalists. Companies can share infrastructure, such as specialized training or testing facilities, which no one firm could support. The top people in an industry are attracted to a region where they will not be dependent on one potential employer. Firms are attracted to the deep talent pool.

Clusters accumulate social capital as a result of repeated interactions, personal relationships, and shared educational paths. Accumulated trust increases willingness to take risks. Clusters expose employees to other firms, helping prevent companies from getting stuck in ruts, and foster cross-fertilization of ideas. If someone with an innovative idea is stymied at their current employer, it is easy to move to another. Clusters encourage rivalry because they make each firm’s successes more visible to the others. They tend to spawn new firms. Clusters thus help avoid the economic brittleness that can result when a single firm, which can decline or go bankrupt, is the dominant employer in a region. When one firm declines, others can pick up the slack. And if an industry’s products or technology change, the ascendant firms are more likely to be local.

It is much easier for governments to support existing clusters than to create new ones. A cluster requires a critical mass of many firms, skilled employees, and supporting institutions, so creating one from nothing requires huge investments, on multiple fronts, often for decades.

Supporting the Industrial Commons refers to the networks of suppliers and know-how that industries rely on.

Industries are often critically dependent upon resources shared with and supported by other industries. For example, in the words of Gary Pisano and Willy Shih of Harvard Business School,

Software knowledge and skills, for instance, are vital to an extremely wide range of industries (machine tools, medical devices, earth-moving equipment, automobiles, aircraft, computers, consumer electronics, defense). Similarly, capabilities related to thin-film deposition processes are crucial to sophisticated optics; to such electronic products as semiconductors and disk drives; and to industrial tools, packaging, solar panels, and advanced displays. The knowledge, skills, and equipment related to the development and production of advanced materials are a commons for such diverse industries as aerospace, automobiles, medical devices, and consumer products. Biotechnology is a commons not just for drugs but also for agriculture and the emerging alternative-fuels industry.

These relationships mean that the thriving of one industry can depend on the health of quite different industries, and a lack of local capability in key supporting industries can make it impossible to produce or innovate certain products in a country or region. When one industry moves offshore, it can damage others, and once a tipping point is passed, a cascading collapse of interdependent industries can follow. It can be difficult and costly to rebuild the lost capabilities – often requiring strategic, expensive, and sustained government support.

Supporting Research: There are often positive spillovers to the wider economy from the research of private firms, so governments subsidize it.

The federal R&D tax credit was created in 1981. It offers a credit of 20 percent of qualified expenditures over a base calculated from prior years to avoid incentivizing research that would have happened anyway. Other nations have similar policies.

Some governments have agencies that perform industrial research on a cost-sharing basis with private firms. Germany’s Fraunhofer Society is one, as are parts of Japan’s National Institute of Advanced Industrial Science and Technology. The Industrial Technology Research Institute of Taiwan has played a major role in transforming Taiwan’s economy over the past 50 years. America has the excellent but far too small, Manufacturing USA.

To have economic impact, knowledge must not only be created, but also transmitted to firms and deployed at scale. The federal Manufacturing Extension Partnership focuses on this for smaller firms.

Basic science is usually impossible for any firm to keep to itself, so it is rarely funded by for-profit entities. In the past, it was sometimes funded by state-sanctioned monopolies, such as AT&T, and by oligopolistic private companies such as GE and Westinghouse, as they had sufficient commercial scope to profitably harvest its benefits.

Infratechnologies are the often hidden but crucial technologies used, among other things, for communication, measurement, standardization, testing, and cataloguing. They include the protocols, such as Wi-Fi, that enable devices to interact. In the US, the underappreciated National Institute of Standards and Technologies (NIST) is the agency focused on infratechnologies.

General purpose technologies, such as jet engines and the Internet, have since WWII rarely come from the private sector on its own. The last GPTs to develop without significant government support were the electric grid and the automobile, above all because their intermediate stages were monetizable. The dominance of the private sector in developing artificial intelligence constitutes a reversion to this older pattern, showing how basic rules of industrial policy can endure for decades and then change.

Intellectual Property (IP) Policy – This covers patents, trade secrets, trademarks, and copyrights.

Nations, above all China, have allowed their firms to steal IP and have frustrated American attempts to respond. Movies, designs for consumer goods, computer chip designs, cellular telephony software, and other major technology systems have all been stolen. The Commission on the Theft of American Intellectual Property has estimated the value of this theft at one to three percent of US GDP. It is no accident that IP theft was the major stated justification for the Trump administration’s first round of tariffs against that nation in 2017. Ironically, China is now strengthening its domestic patent system as it transitions from being a net consumer of IP to being a significant producer in its own right.

Nations also design IP-acquisition policies not involving theft. For example, IP rules were used in several countries, the outstanding example being India, to support development of a generic drug industry – though doing so has become more difficult since the TRIPS Agreement of 1995 (and later). Other lawful tactics include governments’ organizing cartels of domestic buyers of foreign IP to drive the price down, as Japan has done.

Standards Setting covers the rules, such as car safety standards and cell-phone ­­protocols, required for technologies to function, interface with other technologies, and be accepted in the marketplace.

Standards setting is often done by governments, as opposed to the private sector, for good reasons. The private sector may not be able to agree, resulting in a fragmented industry with incompatible products. Allowing one firm to impose its proprietary standard can sometimes work, as with IBM from the 1960s to the early 1980s, but raises antitrust concerns.

Standards setting can also serve as a form of protectionism. As an obscure, WTO-compliant policy, it can strengthen domestic producers due to their familiarity with the idiosyncratic national standard.

Standards setting has also been used to force up quality of demand, compel replacement sales by making older models obsolete, and push the local industry to the technological frontier. Japan’s rigorous annual auto inspection, the shaken, incentivizes drivers to drive a relatively new car.

Going forward, China aspires, as shown by its China Standards 2035 plan, to set not just domestic but global standards for emerging technologies such as the Internet of Things (IOT) – which would be a major step toward global Chinese technological dominance.

Technology Mapping means understanding dependencies between technological capabilities, so as to know in what order they must mastered to achieve a targeted end product, and to block rivals attempting the same thing. During the Reagan administration, pathbreaking work on technology mapping was done in Project Socrates, a CIA-Defense Intelligence Agency project that should be revived. In 2023, NSF’s new Directorate for Technology, Innovation and Partnerships announced funding of $30 million to develop a new program, Assessing and Predicting Technology Outcomes (APTO, later folded into other efforts). Its purpose was to understand America’s competitive position in the technology focus areas designated in 2022’s CHIPS Act.

Technology Denial means efforts to deny rival nations access to key technologies. This goes back centuries: in the 18^th century, China banned export of its porcelain technology, Britain of its textile technology. During the Cold War, the US and its allies established the Coordinating Committee for Multilateral Export Control to block Soviet access to military and dual-use civilian-military technologies. In 2019, Japan even blocked chip exports to South Korea in a political spat unrelated to trade. The CHIPS Act prohibits funded companies from making advanced integrated circuits in China and other “countries of concern” for 10 years after a grant. Following the Act, the Commerce Department, in cooperation with America’s allies, has tightened export controls to deny China and other hostile nations advanced semi-conductor and semiconductor-manufacturing technology.

With few exceptions, industrial policy tools work only when used in concert. A common way to mistakenly dismiss the entire concept of industrial policy is to examine the track record of one specific tool, such as cash subsidies, in isolation, and find that it doesn’t correlate with successful outcomes. (Examples are here, here, here, here, here, here, here, and here.) The American debate about industrial policy must be broadened from its narrow focus on the best use tariffs to the coherent and coordinated deployment of all its applicable tools.

There is no universally applicable protocol for how to combine industrial policies because the right strategy will be specific to each industry and its national and international context. Industrial policy is not an ideology and, unlike neoliberalism, does not traffic in simple universal solutions. Modern economies are complex and have many “pressure points” where industrial policies can be applied. This article has focused on the more obvious tools, but there are many others. Intricacies of the tax code and accounting rules, tacit understandings between government officials and industry associations, deliberate gaps between what laws say and what actually gets enforced: All can serve as tools of industrial policy.

President Trump has made American reindustrialization and reshoring central to his economic policy agenda. Reviving American manufacturing enjoys broad public support and has increasingly become a bipartisan priority. We have previously written about the importance of reindustrialization—a cornerstone of the Biden administration’s agenda—and one that has won over a significant part of America’s policymaking elite.

One of the most effective and least controversial tools for advancing this objective already exists: the Manufacturing Extension Partnership (MEP). Yet despite its stated commitment to rebuilding American manufacturing, the Trump administration has not consistently supported MEP, at times proposing reductions or eliminations that would weaken one of the most effective mechanisms available to strengthen small and mid-sized U.S. manufacturers.

Strengthening MEP would directly reinforce the administration’s stated goals of supply chain resilience, domestic capacity expansion, and small-business manufacturing growth.

The Manufacturing Extension Partnership

The Manufacturing Extension Partnership (MEP), housed within the Commerce Department, helps smaller manufacturers adopt advanced technologies, adopt state-of-the-art management techniques, qualify for OEM supply chains and compete more effectively in domestic and global markets[MF1] . In the past, budget proposals have sought to eliminate or reduce the program, though Congress — including many Republican members — has consistently preserved it due to its demonstrated results.

In a modern economy, innovation must not only be invented—it must be diffused. Knowledge has economic impact only when it is transmitted to firms and deployed at scale. Larger corporations often have the internal resources to track technological developments and implement them. Smaller manufacturers typically do not. MEP fills that gap.

These firms—often tier-three suppliers—play a vital role in American manufacturing. While large Original Equipment Manufacturers (OEMs) dominate headlines, they depend on deep networks of smaller suppliers. Strengthening these suppliers strengthens the entire industrial system.

MEP is analogous to the agricultural extension services that have long supported American farmers by bringing state-of-the-art techniques into widespread use. The program operates on a cost-sharing basis among federal, state, and private-sector partners. Companies contribute financially to the services they receive, ensuring both accountability and relevance. It is not a giveaway program; it is a performance-based partnership.

A Nationwide Network with Measurable Results

MEP has a nationwide network of 51 regional centers and 460 service locations that provide a wide variety of technical and business assistance, either using their own staff or by connecting firms to state and federal agencies, trade associations, universities, research labs, and other institutions. It has over 1,400 in-house manufacturing experts and 2,100 outside consultants. In 2022, MEP conducted 33,500 interactions with 9,111 different clients.

Its 2025 federal funding was $175 million—modest by federal budgetary standards.

Yet its economic returns are substantial:

• As of 2022, every federal dollar invested generates approximately $40.50 in new private investment.

• Every $1,353 in federal funding creates or saves one manufacturing job.

• Since its founding more than three decades ago, MEP has assisted roughly 130,000 companies, contributing to an estimated $139 billion in sales, $26 billion in cost savings, and the creation or retention of 1.45 million jobs.

• A 2023 study estimated its benefit-to-cost ratio at 18 to 1.

• A 2022 analysis concluded that MEP produced $40.50 in new client investment and $35.80 in new sales for each federal dollar spent.

Few federal programs can plausibly claim such measurable leverage and return.

Case Studies

The MEP centers differ in the scale and complexity of their services. Some are small not-for-profit entities, while others are far larger affiliates of universities or state and regional economic-development organizations. They vary according to the focus of local manufacturing, the presence of major research universities, regional startup rates, and the type of entity managing them.

For example, the Ohio Development Services Agency’s Office of Small Business and Entrepreneurship oversees the Ohio MEP through six sub-recipient centers. One, MAGNET in Cleveland, serving the state’s northeastern industrial corridor, has over 50 experts and support staff. By contrast, New Mexico’s MEP Center is a standalone non-profit. It has an office in Albuquerque, branches in Farmington, Las Cruces, and Roswell, and a total staff of 10.

Projects can be carried out by center staff, a partner organization, or third-party consultants. Here are three case studies from MEP’s website:

Doyle Equipment Manufacturing Company is a leading manufacturer of equipment for blending, conveying, tending, and spreading dry fertilizers… In 2016, Doyle consolidated most of its manufacturing and administrative functions in Palmyra, Missouri, where it built a new 220,000 square foot facility near its sister company, Riverview Manufacturing… Doyle recognized the importance of garnering the knowledge and expertise of long-serving employees who were nearing retirement. Losing the experience of critical employees can dramatically impede operational efficiency and productivity, especially as companies expand and prepare for additional growth. To combat the potential loss of tribal knowledge, Doyle decided to introduce an Enterprise Resource Planning system (ERP) to capture and consolidate “islands of information” into one integrated system. Facing the immense task of properly implementing an ERP, company leaders turned to Missouri Enterprise, part of the MEP National Network, for guidance.

Ingevity Corporation provides specialty chemicals, high-performance carbon materials, and engineered polymers that purify, protect, and enhance the world around us. Specifically, the company’s plant in Waynesboro, Georgia, makes a carbon honeycomb filter used to collect automotive gas emissions within a vehicle and reuse them, rather than release them into the atmosphere… In 2016, after a series of automotive recalls (unrelated to Ingevity), the entire automotive industry worked to improve quality standards. Many automotive OEMs began to require new certifications for companies in their supply chain. Even though Ingevity already had a quality management system in place and was certified to ISO 9001 and its automotive supplement, TS16949, the company needed to certify to the new, more stringent automotive- quality standard, IATF16949:2016, by September 2018, to maintain its customer base… Chuck Sabo, the company’s quality systems manager, quickly realized that his team needed some help and reached out to Elliot Price, Augusta Region Manager, who connected them with a GaMEP project manager and quality management system expert.

A 17-year-old spinout of the Bigelow Laboratory of Ocean Science, Fluid Imaging Technologies is based in Scarborough, Maine with 15 employees. The company manufactures industry-leading particle analysis instrumentation based on digital imaging technology. Its flagship product, the FlowCam, is the first automated particle analysis instrument to use digital imaging for measuring size and shape of microscopic particles in a fluid medium. With applications in oceanographic research, municipal water, biopharmaceutical formulations, chemicals, oil and gas, biofuels and many other markets, Fluid Imaging Technologies leads the way in imaging particle analysis… The challenge was to create tools and protocols to enable a manageable product development process… Maine MEP, part of the MEP National Network, worked with the company on a weekly basis to not only design tools and protocols, but also run a test product through the system to provide both training and validation of the system with the staff.

Supporting Supply Chain Resilience and National Security

In addition to solutions to specific manufacturing problems, the MEP centers increasingly offer more generalized services such as supply chain optimization, sustainability measures, workforce development, procurement assistance for government contracts, and technology acceleration. MEP emphasizes helping firms adopt all-important lean manufacturing practices, which it has been instrumental in spreading among small firms.

The centers also help the 35,000-40,000 larger, more sophisticated, managerially proactive firms with tasks such as process improvement, capacity expansion, technology and supplier scouting, technology-driven market intelligence, capital access, cybersecurity, and international quality standards such as ISO 9001.

Most MEP client companies are “tier three” firms, and the centers help them connect with the larger tier two firms that supply the tier one Original Equipment Manufacturers (OEMs) at the top of the manufacturing food chain. With MEP assistance, tier three SMMs often enter into cooperative arrangements with one another to develop products they can then sell to tier two companies.

Although a genuinely collaborative national manufacturing network had been a goal from the start, MEP centers for many years collaborated only regionally. Thus, in 2017, the MEP National Network was launched.

The CHIPS Act of 2022 expanded MEP’s funding, especially for reshor­ing, domestic supply chains, critical technologies, and foundational manufacturing capabilities. It authorized $2.3 billion for 2023 to 2027, triple the previous five-year period, though the actual appropriation for 2023 was $175 million, only $17 million higher than the previous year.

During 2017-23, MEP engaged in several other initiatives:

• The MEP-Assisted Technology and Technical Resource Program connected MEP’s clients with technical resources at NIST laboratories. These have capabilities in advanced manufacturing technology, collaborative robotics, additive manufacturing, materials design, nanotechnology, information and communications technology, quantum information, biosciences, and industrial standards.
  

• The Competitive Awards Program supported projects at the centers aimed at solving new or emerging manufacturing problems. These awards required no matching funds from the individual centers. They were made on a peer-reviewed and competitive basis, and could extend up to three years. Only centers with sufficient performance ratings were eligible for the awards, which ranged from $50,000 to $1 million.
  

• The Defense Manufacturing Supply Chain Support program enabled MEP centers to provide technical assistance to small and innovative manufacturers in the defense industrial base. This included assistance in capitalizing on market opportunities associated with DOD’s MFG-USA institutes. It also helped DOD’s agencies integrate new technologies, best practices, and optimal manufacturing approaches into their suppliers’ operations.
  

• Helping Manufacturers Source Domestically: In 2021, President Biden signed E.O. 14005: Ensuring the Future is Made in All of America by All of America’s Workforce. This specifically requested that MEP Supplier Scouting and MEP-Assisted Technology and Technical Resource Services help government procurement offices connect with domestic manufacturers.
  

• Improving Domestic Supply Chains: NIST awarded approximately $400,000 apiece to each of the MEP centers to develop programs to make domestic supply chains more efficient and resilient to geopolitical and pandemic disruptions, including funds for connecting small and medium-sized manufacturers with OEMs.

A Strategic Opportunity

MEP’s budget is small compared to similar foreign programs like Germany’s Fraunhofer Gesellschaft and Japan’s Kohsetsushi, both major contributors to their respective nations’ manufacturing prowess. It is also modest relative to the scale of America’s manufacturing base.

At a time when policymakers are scrutinizing federal programs for efficiency and measurable impact, MEP stands out as a high-performing investment in domestic capacity.

If the United States is serious about reindustrialization, strengthening firms inside the tariff wall is just as important as erecting it. Tariffs protect the market; productivity and technological diffusion determine who thrives within it.

Strengthening the MEP would reinforce the administration’s commitment to domestic production, supply chain resilience, and small-business manufacturing growth. It would be a practical and effective step toward cementing America’s industrial revival.

Reindustrialization requires both market protection and firm-level capability. MEP provides the latter. It deserves sustained and strengthened support as part of a comprehensive strategy to rebuild American manufacturing.

People have been warning lately about how hard and how fast artificial intelligence is about to hit every profession, how it alters Man’s very status in the universe, and how it threatens utter catastrophe. I can’t claim greater insight on these topics, so I won’t elaborate on them. But I do think the long-term economics of AI hasn’t been well analyzed yet (unless someone I’m unaware of has done it), so I’d like to address this topic.

Right now, people seem split between four views:

1. Panic epitomized by Bernie Sanders’s call for a moratorium on constructing data centers.

2. People like Rob Atkinson who believe AI’s impact will be like any other technology: unpleasant for some, but overall raising productivity and therefore real wages.

3. People like Andrew Yang who think that AI is shortly going to do all the work, so humans should be given free money to just sit back and enjoy it.

4. Consulting firms offering estimates of which jobs and which industries will feel the greatest impact.

These are either talking only about specifics, obsessing over one aspect of what AI is likely to do, or simply assuming that the future will resemble the past. (It may or it may not, but this should be established by argument, not just be assumed.) Nobody is trying to give an economic theory of the whole.

I believe the correct way to analyze the long-term impact of AI is as follows:

1. It will produce a radical shift in the relative scarcities of goods.

2. This will, in turn, produce a radical re-politicization of the social order.

The first dynamic derives from the fact that AI will make the cognitive aspects, broadly defined, of tasks now done by humans very cheap. By this, I mean its galloping ability to write code, answer phones, et cetera, closely followed by its ability to drive cars today, trucks soon, and so forth. So the part of any task, blue-collar or white-collar, that is done by a human brain can soon be done by a machine, and cheaper.

It follows that AI will generate an unevenly distributed negative cost shock throughout the economy. Most of its consequences in the intermediate term, i.e., until AI is fully deployed to the limits of its capability, will turn on how this tsunami rolls unevenly through the economy, affecting certain areas harder and sooner than others.

In the long run, the real key is that while brain-based operations, as a factor of production, will become cheap, certain other things won’t. For example, even if automobiles can be wholly built by robots that are themselves built by robots, cars will still be made out of raw materials like steel, plastic, and glass. Even if houses are built by robots, there is no good reason for lumber and cement to became cheaper (and if the zoning laws that determine the scarcity of land change, that’s a whole other issue).

Indeed, these other inputs are quite likely to become more expensive, definitely in relative but possibly even in absolute terms, precisely because the brain-substituting inputs have gotten so cheap. Right now, the housing industry cannot demand more lumber and land than the number of houses people buy, which is constrained by the price of houses, which is partly a function of the labor that goes into building them. If the labor portion of producing a house gets cheaper, then, other things being equal, the price of houses will go down, more houses will be demanded, and therefore more lumber and land will be demanded and their price will rise.

Remember when TV sets were expensive and suburban real estate in major metro areas was not? Over the next 20 years or less, AI is going to produce a comparable but greater reshuffling of the relative prices of goods. Bottom line: land and raw materials are the two things AI cannot make cheaper. These will therefore become the key goods of contention, the things whose possession (or lack thereof) is the object of both global geopolitical conflicts and the key determinant of an individual’s living standard. (And there’s a third thing I’ll get to later, but I need to explain more of the argument before it will make sense.)

Granted, the above argument presumes:

1. That AI will continue to get cheaper and cheaper. I know of no reason why it shouldn’t, but this is still an empirical premise, not a conceptual truth.

2. That AI will become so ubiquitous (and possibly self-deploying) that the deadly serious contest going on right now to achieve the most advanced AI will end in an equilibrium of equal capabilities all round.

3. That some other factor, like energy, doesn’t turn out to be constrained in ways we do not now expect.

Note that we’re talking here about the ultimate end-point of AI’s development and deployment. Until we get there, competence at AI itself will be the biggest theatre of rivalry, and it would be very unwise to let the US fall behind. But if we accept the above givens, then land and raw materials (which economists often define as a single thing, as in “land, labor, and capital” because raw materials come from land) will be the key constraints on both national economic output and individual consumption.

So, paradoxically, the triumph of AI will lead to the revenge of the physical world.

The implication here is that AI is unlikely to usher in a world where everything is so cheap to produce that we can just give goods and services away for free. That, minus the mediating institution of a government check every month, is what basic income amounts to, in its AI-related version. Giving stuff away for free only makes sense, and is only economically sustainable, if things are free, and they won’t be. (Basic income is also problematic for other reasons that need not detain us here.)

The next interesting dynamic to note is that if everything whose major production cost is the work of human brains drops in price, then the relative price of everything else must go up. That’s just arithmetic. For the last 80 years or so, the opposite has been happening, due to so-called Baumol’s Cost Disease. Goods that could be produced with the rising non-AI automation of this period, mainly manufactures, became relatively cheaper. Simultaneously, non-offshorable labor-intensive services – everything from schoolteachers to doctors – got more expensive. This is all well documented and not that controversial: Will Baumol framed the problem in terms of explaining a crush of rising costs, e.g., in government budgets, as a difference in the rate of inflation in goods vs. services.

But Baumol could just have accurately described this dynamic as “Baumol’s prosperity sharing.” Consider: The productivity of steelworkers in the US rose by a factor of eight from 1950 to 2020, while the productivity of barbers remained unchanged. But we didn’t see the wages of steelworkers increase by a factor of eight while those of barbers remained the same. Instead, with some variation, the wages of American workers as a whole rose during these years on the back of the economy’s overall increase in productivity. (Some argue that wages and productivity diverged, but read this, and the logic here holds even if wages didn’t rise quite as fast as productivity.)

The point here is that sector-specific productivity gains don’t go only to workers in the gaining sector. This has been a good thing for society, but, crucially, it has depended on the fact that huge numbers of humans were needed to perform cognitive tasks. This is why past technological change did not impoverish the workers it put out of a job. But when this fact disappears, humans no longer have the monopoly on cognitive capability that forced the world to keep paying them no matter how expensive they got. Baumol’s prosperity sharing, that is, only works when human brains are a bottleneck through which the economy’s output must pass.

If you take this bottleneck out of the picture, there’s no longer a reason for there to be a difference in the inflation rate between goods and services. Looked at one way, this will be a good thing, because services that people want will get cheaper. Medical services, for example, will probably stop gobbling up a larger and larger share of GDP. And good luck persuading a public sweating from the cost of health insurance to stop this: Anyone wanting to restrain deployment of AI is going to discover that people on the “customer” side of the counter will always prefer a cheap robot to an expensive human. Granted, this won’t be true if customers are willing to pay for the human touch, but then we’re talking about something robots can’t do, which is by definition an exception to this logic.

The reassuring fact here, among the many alarming ones, is that AI can’t actually destroy incomes faster than it lowers costs. For example, if we replace a $50 per x-ray technician with a $5 per x-ray robot, workers as a class lose $45 in income from this change. But they also face $45 less in medical bills. This is, of course, the essence of view #2 noted in the opening paragraph, and for much of the transition period to full AI deployment, it will hold true. At the terminal equilibrium, this is simply the truism that the hell where robots take all the jobs is also the heaven where robots do all the work. (More on this later.) During the transition period, who ends up happy and who unhappy will depend on whether the cost shock hits their income or their expenses sooner.

Another important implication here is that AI will be very deflationary. Given that developed nations’ post-1930s Keynesian economic systems are based (unlike, say, the deflationary late 19^th-century US) on low but non-zero inflation, this probably won’t result in actual deflation. What is more likely is the loose monetary policy that governments will be able to get away with, without triggering the inflation that is its normal consequence, because of this huge deflationary pressure. Stable prices with much of the economy deflating necessarily means prices rising elsewhere, so again, we have our radical shift in relative prices, with inflation piling up in whatever AI can’t touch.

Here’s where politics enters the picture, because we’ll also see a decoupling of two extremely important dynamics that have operated in parallel for a very long time, but have been taken so for granted that they are rarely even discussed:

1. The economic leverage (derived from society’s demand for their work) that people have to extract income from the economy.

2. The political leverage (derived from politicians’ demand for their vote) that people have to extract income from the economy.

A major reason the economies of democratic nations function at all is that the ability of people to extract income from the economy is, despite a political system that could hypothetically redistribute income by fiat, close to what the free market would give them. Granted, a) this dynamic is not perfect, b) disputing how some groups allegedly take more than their fair share is a major issue in our politics, and c) some classes of people (children, retirees) are not expected to earn their own way. But broadly speaking, it is true, and if it weren’t, the incentive systems of a market economy couldn’t function.

The problems start when dynamic #1 above sharply diminishes while #2 does not. If vast numbers of people start to lose income while retaining political leverage, they will inevitably try to get government to do something about it. Many of them will have convincing explanations for why they deserve interventions in their favor, and this will be complicated by the fact that some of them will even be right.

A giant stabilizing factor in our politics has always been the fact that whatever extreme or foolish political ideas people may entertain, 99% of the population still has to get up the next day, go to work, and do their jobs. It’s no accident that in times of mass unemployment, political upheaval is common. My point here is not so much the likelihood of mass unemployment as the fact that the economic leverage of a lawyer is going to converge on that of her secretary. Some may welcome this, because it implies a reduction in inequality, but there’s a problem. If we extrapolate this trend, what happens when the economic leverage of human labor approaches zero? The work of distributing income shifts entirely to the political system, as opposed to the economy. More and more people will be in the current situation of Social Security and welfare recipients, whose incomes are governmentally determined.

This is where radical politicization emerges. Right now, we’re used to having a social structure that is mostly determined by people’s occupations, starting with the fact that if I say, “social hierarchy,” you assume automatically that I mean the hierarchy of how important people’s jobs are, usually but not always measured by how much they earn doing them. Outside of situations like a full-blown Marxist revolution, this imposes a natural limit on how much any political intervention can change society. But if you take this economic backbone of our social structure away, then:

1. The scope of what people have good reason to fight over politically expands.

2. The economic stabilizer independent of political interventions weakens.

In the unlikely thought-experiment of everyone one day subsisting on UBI, who gets how much UBI will obviously be the major issue in politics, period.

It would seem to follow that, compared to the politicization we’ve seen since 2016, “You ain’t seen nothin’ yet.” This doesn’t necessarily mean we will experience extreme antagonism between left and right. Indeed, the above dynamics may well be potent enough to scramble existing left-right configurations of who votes for whom and what policies they want. But the days of being able to rely upon the fact that society has a natural, self-enforcing structure if government just enforces property rights and a few other things, will be gone.

We Americans have generally been able to assume that government should only proactively intervene to shape social structure only when something goes wrong, i.e., when this default outcome is unsatisfactory in some way. Granted, what counts as “unsatisfactory,” and what government should do about it, are legitimately controversial, but there are not today, and have never been, significant players in American politics believing that the structure of society should be determined by government per se. But we are increasingly going to be cornered into that result, simply because there will be no other source of social structure.

This can’t happen immediately, and there will be a thousand complications and countervailing factors, starting with the fact that human labor is unlikely to ever be completely eliminated. But this is still where we are eventually headed.

Between now and the terminal equilibrium, there will be all sorts of profound secondary effects. For example, what will happen to unions if the labor of their members becomes irrelevant? What will public opinion about immigration become when “come here and work” becomes meaningless? What will people go to college for, if not to learn well-compensated skills?

The above changes pose particular risks to American democracy. Ignoring many details of implementation, the basic social contract in the US has never been paternalistic, i.e., based on the idea that it is the function of government to take care of people. The assumption has always been that people should take care of themselves and that, as noted, government should step in when they can’t. And the backbone of this logic has simply been people earning income according to the market. Take this factor away, and you inevitably end up with a world in which the social contract has to be “government taking care of people,” or people will get nothing.

We Americans have no experience running that kind of government.

Another implication, which I alluded to earlier, is that in addition to land and raw materials, politics itself will become a binding constraint on the economy, in the sense of quality of governance. At any given level of AI capability, a society is going to be able to exploit it, and distribute its fruits to its population, effectively if it enjoys the governance quality of Japan or Switzerland. At the level of Russia or South Africa, I doubt it. I mean here “governance” here in the sense of “implementing a regime that aims at the common good,” not just technical state capacity, though obviously the latter will be required. And if the importance of politics sharply increases, then the importance of civic virtue will necessarily increase. People will quite rationally care a lot more about who governs them.

What has all this to do with industrial policy, the subject of this Substack? Nothing and everything. The above facts will, over the next few decades, so massively condition whatever economy and political system U.S. industrial policy is implemented in, that not having a concept of what’s going to happen would be like a nation trying to formulate a foreign policy in 1950 without having a settled opinion on communism vs. capitalism. Most of America’s current industrial-policy needs have to do with transition period to full AI deployment, but without an idea of the end game, we will be distracted from being able to formulate policy correctly. It will be important not to get blown off course.

In 2016, Trump’s policies crystallized a turn of public and elite opinion in favor of America’s reindustrialization, a turn which survived Biden’s election and continues today. We have previously explained why this is a good idea. But despite rebounds in important individual sectors such as semiconductors, plus spotty gains elsewhere (steel, batteries, appliances) the US is not now broadly reindustrializing. And it will not until a problem that has received astonishingly little attention is addressed.

We refer to the overvaluation of the dollar. Until this is reversed, America will continue the huge trade deficits in manufactured goods that foreclose reindustrialization. Tariffs on their own cannot solve this problem. Nor can quotas, subsidies, better governmental support for technological innovation, or any other package of policies that does not include devaluation.

Raising the percentage of manufacturing in US GDP will require higher industrial output in existing sectors where America has lost market share to foreign nations. Growth in new industries is welcome, but will not be enough to carry the burden of reindustrialization alone.

Our trade deficit is mostly in durable goods, which is just over half of US manufacturing. Nondurables are mainly sectors that the US should not want (apparel), are tied to natural resources (petroleum), or are where the US runs trade surpluses (chemicals). Food processing, along with agriculture generally, has gone into deficit but presents issues outside the scope of this article. The major nondurable problem industry is pharmaceuticals.

We will have to grow production in pharma and the durable goods industries where we now run huge deficits:

1. Electrical Machinery ($262bn deficit in 2024)

2. Motor Vehicles & Parts ($239bn)

3. Computers & Parts ($130bn)

4. Pharmaceuticals ($117bn)

5. Buses, Tractors, Others ($42bn)

6. Iron, Steel & Products Thereof ($38bn)

7. Gas turbines ($17bn)

8. Optical, Medical & Precision Instruments ($16bn)

9. Air conditioning machinery ($12bn)

10. Refrigerators ($11bn)

11. Printing machinery ($10bn)

If US production increases in these sectors, either the US must reduce its imports or increase its exports in them. Either way, our trade deficit in manufactured goods will go down. This will bring down our overall deficit because the US runs roughly balanced trade in raw materials and a surplus in services, and we cannot expect a sharp change in either.

Here lies the connection to the value of the dollar: We are not talking about a causal chain where increased output → fewer imports or more exports → lower trade deficit. The causation could just as accurately be described as the reverse: reduced trade deficit → fewer imports or more exports → increased production. The reason is that we’re not actually looking at a sequential causal chain at all, but at things that take place simultaneously, not in sequence. Which means that if any of these events cannot occur, none of them can. As a result, anything that prevents a decline in the trade deficit will prevent growth in US industrial output and thus prevent America’s reindustrialization.

Right now, the US has an average effective tariff (disputable to some extent depending on different ways of calculating it) of about 14%. But it also has a dollar overvalued by roughly 17% compared to the price that would balance our trade. (Dollar overvaluation was estimated at that level in October 2024, and since then, the dollar has risen and then fallen back.) So, effectively the US has a net negative tariff.

This small negative tariff, i.e., an export subsidy for our trading partners, is one reason they have not retaliated against our tariffs to the extent predicted. They know perfectly well that the US hasn’t even leveled the playing field yet, let alone tilted it to America’s advantage.

The overvalued dollar also creates a 17% headwind against our exports; this dual effect is one reason overvaluation is so harmful. This is compounded by the fact that almost all our major export markets levy value-added tax (VAT) on our imports, averaging 19% in the OECD. Each has added or increased its VAT – which is refunded on exports – to replace taxes, like the corporate income tax, that are not refunded on exports, a combination that produces so-called “fiscal” devaluation.

But even very high tariffs would not produce balanced trade. Any tariff that succeeds in cutting imports will ipso facto reduce the quantity of dollars Americans must sell in international currency markets to buy imports. Reducing the supply of dollars in the currency markets drives up its price. In the real world, where a number of factors affect currency values, the resulting rise in the dollar would not completely offset the tariffs, but would still generate enough upward pressure on the dollar to keep balance out-of-reach .

Pushing tariffs ever higher in pursuit of trade balance would create other problems. So far tariffs have not produced inflation, but at some higher level and/or broader coverage they will. Exorbitant tariffs would also tend to push the US into industries it doesn’t belong in, such as those based on cheap labor, on raw materials we don’t have, or on agricultural products that don’t grow here. There is good reason for us to import coffee and t-shirts – just not electric cars. The correct use of tariffs is for sector policy, i.e., to favor the development or survival in the US of particular industries that are especially desirable for one reason or another.

For these reasons, trade balance will require proactive management of the dollar.

At various times and for different periods, our key trading partners have tacitly or explicitly manipulated the values of their currencies. China requires exporters to redeem dollars they earn at state-controlled institutions that “sterilize” them by using them to buy not American goods, but American financial assets. Germany benefits from using the euro whose value is pulled down by other eurozone economies that lack its export prowess.

Manipulation is currently at an ebb because many currencies are now as weak as their governments desire due to other factors. China, for example, has been suffering from massive capital flight due to slowing growth, a troubled property market, and increasing political interference in the private sector. Still, such manipulation must be countered when it returns.

When a government buys another nation’s currency to drive it up, the victim can buy the aggressor’s currency to drive it back down – so-called countervailing currency intervention or CCI. For currencies that are not fully convertible, such as the renminbi, foreign purchases of US financial assets can be regulated.

But manipulation is not currently the major cause of dollar overvaluation. In addition to serving as our national currency, the U.S. dollar is the world’s primary invoicing unit, safe haven, and reserve asset. It accounts for 56% of global reserves, with the #2 euro at 21%. (China’s renminbi, which isn’t even freely convertible, accounts for only 2%.) So private sector investors, like governments, want to hold dollars as the safest and most liquid store of value.

America has the world’s largest, most secure, most liquid capital markets, open to the world and protected by the rule of law. We have also generally had the most profitable investment opportunities and the greatest openness of those opportunities to outsiders. As a result, the US attracts continuing large net inflows of foreign capital that is spent mostly on existing stocks, bonds, and other financial assets, not on purchasing US goods and services, and therefore does not tend to produce balance in our trade.

So, any solution must reduce foreign purchases of dollars and dollar-denominated assets. This implies that we should tax a) the sale of American assets to foreigners and b) the assumption by Americans of debt to foreigners, which amount to the same thing for these purposes.

One such mechanism was proposed in 2019 by Senators Tammy Baldwin (D-WI) and Josh Hawley (R-MO) in the The Competitive Dollar for Jobs and Prosperity Act. The Act directs the Fed to impose a variable tax on foreign purchases of dollar-denominated assets, a “Market Access Charge (MAC)”. By lowering the net yield on American assets, the MAC would gradually reduce capital inflows until the dollar reached a trade-balancing level.

A 2021 study by the Coalition for a Prosperous America estimated that a 5% MAC would lower the dollar enough to balance trade over five years. It would also produce an additional $1.6 trillion dollars of GDP, 3.9 million jobs, and $300 billion per year in additional tax revenue. The costs would be small relative to the benefits: it would raise interest rates by a mere 0.12 percent and inflation by 0.22 percent.

A MAC would reduce the dollar’s reserve currency status, but not extinguish it. Reserve status creates $35 billion dollars a year of “seigniorage,” the U.S. government’s ability to print money that doesn’t just inflate away because foreigners desire to hold it. It allows the US to borrow internationally 10 - 30 basis points cheaper. American leverage over so much of the international payment system enables us to sanction enemies by preventing them from using it. But reserve status is emphatically not a free lunch: The value of these benefits is swamped by the trillions in trade deficits it has enabled.

Unlike tariffs, tit-for-tat retaliation is ineffective against capital controls like the MAC. In fact, retaliation by foreign nations imposing capital controls of their own would make the MAC even more effective. Furthermore, because balanced trade is a) something every nation can have at the same time, and b) intrinsically sustainable, the MAC – unlike a tariff war – would have a stabilizing influence on the global economic order.

Economic history suggests that that balanced trade doesn’t happen without proactive measures. Our last major era of balanced trade was the 1944-1971 Bretton Woods period, when our currency’s value was pegged to a global system of fixed exchange rates, enforced in part by capital controls all over the developed world. From 1963 to 1974, the US had a MAC-like Interest Equalization Tax, which could rise as high as 15 percent, to control America’s capital outflows – pushing up the dollar, then under downward pressure, and defending its fixed value. And during our previous tariff era before WWII, we were on the gold standard (de facto 1873-1931), another fixed-exchange-rate system. Before that, in our previous major era of unmanaged currencies from 1800 to 1872, the US mostly ran deficits.

Despite acknowledgements by Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, Chairman of the Council of Economic Advisors Steven Miran, and Trump himself of the need for a lower dollar, why has nothing been done? In part because dollar devaluation remains politically taboo. A strong dollar is equated with national strength. But, given the effects of its excessive valuation, “strong” is a mistaken adjective. One could more accurately call it a “bloated,” “overpriced,” or “uncompetitive” dollar.

In addition, the beneficiaries of dollar overvaluation are concentrated and powerful: financial institutions who manage the incoming investments; owners of financial assets whose prices are pushed up by the foreign buyers; and large corporate importers. The costs – loss of productive capacity, declining technological knowhow, and indebtedness – are diffuse and delayed. Yet avoiding the overvaluation issue guarantees continued failure. No effective strategy to achieve balanced trade can ignore the exchange rate. And without balanced trade, our deindustrialization will continue.

The pressure of events has inexorably pushed the U.S. into proactive industrial policies in one industry and technology after another. These events have ranged from the chronic (the trade deficit), to the threatening (China’s rare earths embargo), to the promising (the exploding potential of AI).

The result has been a set of ad hoc policies, of varying degrees of rationality, with critical gaps and some at cross-purposes to others, but a lot has happened. Let’s take a look.

First, the pivot away from China. Thanks to tariffs, now averaging 41 percent on a trade-weighted basis, Chinese imports as a percentage of America’s total fell between 2017 and 2024 from 22 percent to 13 percent. Uncontrolled trade with China had been damaging in ways from regional deindustrialization to loss of cutting-edge technological capabilities, so this was appropriate defensive industrial policy. These tariffs were explicitly justified as retaliation against industrial subsidies, theft of intellectual property, and coercive technology transfers. Although this should have been done 25 years ago, America is now enforcing what should be basic prerequisites for opening our economy to a foreign nation.

Next, export controls. These are now being used not only for security, as during the Cold War, but to proactively shape the global distribution of some of the most important present and future industries. They constrain where innovation can occur, shape which firms can scale, and implicitly subsidize domestic and allied producers. There are now controls on advanced microprocessors, equipment for making leading-edge semiconductors, GPU chips, AI accelerator chips, software for training AI models, and some aspect of those models themselves, such as their weights.

Last, a wide range of industry-specific policies.

On December 20, 2017, Executive Order 13817 marked the U.S. government’s first explicit recognition that mineral supply chains are a strategic industrial asset rather than just a commercial commodity. It flagged three interlocking vulnerabilities with respect to rare earths and other critical minerals such as lithium:

1. The U.S. was in many cases 100 percent import-dependent, frequently on China, which dominated not just mining, but the true choke points of processing and refining.

2. Beijing had already demonstrated its willingness to use their supply as a geopolitical weapon.

3. The U.S. had excessively slow permitting for new mines and processing facilities, fragmented governmental authority over this issue, and regulatory uncertainty.

In response to Chinese actions, the administration escalated further with EO 13953 in 2020, which activated emergency powers under the International Economic Emergency Powers Act and the Defense Production Act. Today, critical minerals policy now includes Federal equity stakes:

1. The Defense Department took a 15 percent stake (preferred / convertible stock and warrants) in the rare earths producer MP Materials.

2. The U.S. took a 5 percent stake in Lithium Americas and a separate 5 percent stake in its Thacker Pass mining joint venture with GM.

3. The U.S. invested $35.6 million for a 10 percent stake (plus warrants) in Trilogy Metals, which is developing a mine in Alaska to produce copper, zinc, lead, gold, silver, and potentially germanium.

Continuing to less exotic metals: We have today a 50 percent tariff on steel from most nations, plus tariffs on many downstream products. Here, too, the U.S. government has embraced corporate equity, taking a “golden share” in U.S. Steel, with special veto rights, after Nippon Steel’s takeover of the firm. The steel tariff has helped rebuild U.S. production capacity: Since 2018, some 21 million metric tons (MT) of new or expanded steelmaking capacity has been announced or completed, the largest expansion in decades.

The US also has a 50 percent tariff on aluminum, but has not rebuilt the industry. Capacity utilization in existing smelters has improved, but new smelters have not been built, nor have existing smelters significantly expanded. (A $4 billion proposal by Emirates Global Aluminium for a in Oklahoma remains contingent on a cheap electricity deal.) The cost of aluminum is 20-40 percent electricity (depending on the electricity’s price) so the U.S. is disadvantaged relative to nations that have either electricity subsidies (China), proximity to hydropower (Iceland, Canada, Russia), or cheap fossil fuels (Russia, UAE). Everywhere primary aluminum is produced, it is because of an explicit governmental decision to host this industry, so the U.S. will need additional policy tools to regain its position as a major producer.

The Trump administration is clearly contemplating a wide range of other industries for protectionist measures. The U.S. today has ongoing Section 232 (threats to national security) investigations into the following:

1. Semiconductors, including tools for photolithography (including EUV and DUV), deposition, etching, ion implantation, wafer inspection, metrology, and testing. This includes logic and memory semiconductors across advanced, mature, and legacy nodes, plus discrete devices and integrated circuits.

2. Pharmaceuticals. This encompasses pharmaceutical products, active pharmaceutical ingredients, key chemical precursors, and intermediate compounds. It includes small-molecule drugs, biologics, vaccines, sterile injectables, and complex formulations, as well as contract manufacturing and fill-finish stages.

3. Raw, refined, processed, and semi-finished critical minerals, including midstream and downstream processing. This includes oxides, salts, powders, concentrates, alloys, and component-ready materials.

4. Refined copper and related products. This includes refined copper cathodes, billets, rods, wire, strip, plate, and other semi-finished copper forms, as well as copper-intensive derivative products. It encompasses upstream refining capacity, rolling and extrusion stages, and downstream fabrication.

5. Wood and lumber products. This covers logs, sawn lumber, engineered wood products, plywood, veneer, oriented strand board, and products. It encompasses harvesting, milling, and secondary processing.

6. Medium- and heavy-duty trucks. This includes their chassis systems, internal combustion and alternative-fuel engines, suspension assemblies, electronic control units, braking systems, powertrain software. transmissions, axles, and drivetrains. It covers both complete vehicles and tiers one and two of component supply chains.

7. Commercial aircraft. This includes airliners, regional aircraft, and business jets. Components include turbofan and turboprop engines, engine cores, nacelles, avionics, flight-control systems, landing gear, and composite structures. The scope includes design, assembly, and multiple supplier tiers.

8. Polysilicon for semiconductors and solar. The investigation also covers related inputs, including upstream production of high-purity silicon feedstock, ingot formation, wafer slicing, and intermediate processing steps.

9. Drones. This includes key components, such as airframes, propulsion systems, sensors, navigation equipment, communications modules, control software, and ground-station equipment.

10. Wind Turbines. This encompasses utility-scale wind turbines and their major components, including blades, towers, nacelles, gearboxes, generators, and power electronics. It includes both onshore and offshore turbine equipment and both the manufacturing and on-site assembly stages.

11. Medical supplies. This includes disposable PPE (masks, gloves, gowns), consumables (syringes, tubing, test kits), medical equipment, medical devices, and hospital equipment. Products include imaging equipment, monitoring systems, ventilators, infusion pumps, surgical tools, diagnostic devices, and their components.

12. Robotics. This investigation covers advanced robotic systems used in manufacturing, logistics, defense, and infrastructure, including robotic arms, autonomous systems, control software, sensors, and integrated automation platforms.

13. Industrial Machinery. Technically a part of the previous investigation, this includes computer numerical control machining centers; turning and milling machines; grinding and deburring equipment; stamping and pressing machines; automatic tool changers; machine tools for cutting, welding, and work handling pieces; metal finishing and treatment equipment, and laser and water-cutting tools.

Such investigations are exactly the first steps for constructing an industrial policy systematically protecting strategically important industries.

Shipbuilding is the major industry in which the U.S. lags the most. With 25 percent of the world’s economy, the U.S. builds 0.1 percent of the world’s ocean-going ships. On April 9, 2025, Executive Order 14269 formally made shipbuilding revitalization a national priority, enjoining preparation of a multi-agency Maritime Action Plan within 210 days. The plan has not yet materialized but the Trump administration has attacked the problem through investment deals, negotiating a strategic partnership with South Korea with Seoul pledging investments in U.S. shipbuilding of up to $150 billion (spread out over several years) as part of broader trade negotiations.

The Korean shipbuilding firm Hanwha Oceans acquired a Philadelphia shipyard in 2024 and committed $5 billion to modernize and expand it. (HD Hyundai and the U.S. firm Cerberus Capital Management are also involved.) Private investment, state-of-the-art technology, and managerial knowhow are necessary but will not on their own rebuild American shipbuilding, given its extreme atrophy and huge cost disadvantage. These undertakings, while welcome, will at best lead to modestly expanded capacity focused on protected corners of the American market, such as US Navy sealift vessels and ships carrying cargo between US ports, which under the Jones Act of 1920 must be American-built, American-flagged, 75% American-owned, and 75% American-crewed.

Enter the SHIPS for America Act. Introduced in the House and Senate in 2025, with apparent support from the administration, it would be a serious program to rebuild American shipbuilding across-the-board, using multiple policy tools. It would require a growing share of U.S. government cargo and strategic trade to be carried on American-flagged, American-built ships. It would expand and reform federal loan guarantees, with support for new commercial vessels, shipyard modernization, equipment and automation.

At the other end of the transportation spectrum, the U.S. today has the world’s largest and most successful private space sector. Since the Commercial Orbital Transportation Services program of 2005, NASA has proactively pivoted from being an owner-operator of space hardware to being a consumer of space services that incentivizes private industry to develop space vehicles. Laissez faire ideologues have characterized the success of the sector as being merely a matter of “privatization,” but in fact it came about because of a carefully phased process of incentivizing private innovation with public money until the industry was de-risked enough for the private sector to take it over.

Even more important than space policy has been the CHIPS Act passed under Biden in 2022. The semiconductor sector, along with AI, is today the only sector in which a comprehensive array of industrial policy instruments are being applied: tariffs, export controls, subsidies, tax credits, federal R&D, defense procurement, workforce development, and coordination with foreign allies.

Although Trump initially condemned CHIPS, by the time he retook office in 2025, major plants were already under construction. Repeal would not have saved money, but have stranded projects, triggered lawsuits, and created chaos in regions that had voted for him. So he settled for changing the program – which should be an object lesson in how to make industrial policy durable.

The Act targeted the entire research and manufacturing ecosystem, not just fabs. The diversity of its programs recognized that successful industrial policies almost always use multiple tools. Its first core component was $39 billion in semiconductor manufacturing incentives, structured as cost-sharing grants rather than blank checks, and tied to domestic production requirements, private co-investment, and guardrails restricting recipients from expanding advanced manufacturing capacity in China or other “countries of concern.” Second was a 25 percent tax credit for capital investments in manufacturing facilities and equipment. Third were long-term investments in workforce development, advanced packaging, and semiconductor research and development.

These financial supports had become the “ante” for American participation in the industry, due to a structural cost gap between the U.S. and East Asia, where governments routinely subsidize 30–50 percent of fab costs. This is unlikely to be the last time foreign governments corner America into industrial policies.

The CHIPS Program Office’s role in administering the Act’s incentives was eventually absorbed into a new United States Investment Accelerator created in March 2025. This office aimed to facilitate domestic and foreign investment projects in many industries, not just semiconductors, especially for deals over $1 billion. It is charged with negotiating investment deals, navigating regulatory processes, and coordinating cross-agency support for large capital projects. According to its establishing EO, it was created because:

1. U.S. regulatory processes are slow and burdensome, making it harder for companies to invest and build facilities in the U.S.

2. Large projects face overlapping federal, state, and local legal requirements that delay permitting, construction, and launch.

3. The federal government should more actively assist major investors, both domestic and foreign, to make the U.S. a more attractive destination for long-term investment.

Although the U.S. has long had programs like Select USA designed to market the U.S. abroad as an investment location, this was a huge and welcome shift of focus at scale to supporting foreign investment in the U.S.

In energy, Trump’s industrial policies have been a decidedly mixed bag. Denying the reality of climate change, he saw no reason to subsidize carbon-free energy. Despite rising electricity prices and demand, his administration halted not only less technologically mature, higher cost offshore wind projects (California), but also projects that were already partly built (the Northeast) whose economics were demonstrably competitive with other green energy sources. Whether these halts are legal is the object of ongoing litigation. In place of green energy, he has favored fossil fuels, attempting to keep coal-burning power plants open (even against the will of utilities) and seeking to expand infrastructure for exporting oil and natural gas.

In the long run, Trump’s most consequential energy policy may be to double down on nuclear power, which he has done in four ways:

1. Guaranteed demand and long-term contracting.

2. Fuel-cycle security and domestic supply.

3. Accelerated licensing and regulatory intervention.

4. Support for advanced and next-generation reactors.

The stated goal is a quadrupling of nuclear generating capacity by 2050.

For decades, this industry has struggled in the U.S. due to high financing costs, regulatory complexity, and long construction timelines. In response, the administration’s interventions aimed to make building new plants less risky, making projects more investable by private capital.

Most explicitly, the federal government entered into a partnership with Westinghouse to deploy at least $80 billion worth of its AP1000 nuclear reactors. Brookfield Asset Management will supply the financing and Cameco the uranium mining, conversion, and enrichment. The deal includes a form of contingent quasi-equity participation by the federal government, which will reap a profit if the projects reach specified success targets.

The administration has also committed to new forms of nuclear power. It has allocated $800 million in cost-share funding for small modular reactor (SMR) projects in Tennessee (a GE Vernova Hitachi BWRX-300) and Michigan (two SMR-300s). DOE’s cost-share cooperative agreements have helped SMR developers advance design certification and licensing, with key recipients including NuScale Power, which has the first American SMR design to enter the NRC’s certification process.

The Trump administration has also proclaimed its support for fusion. But this support has not translated into increased funding for the technology, which is now receiving less federal money than it did under Biden. Instead, while broad, exploratory, long-horizon research funding shrank along with the rest of US science funding, the administration refocused more tightly on aiding commercialization.

DOE issued a Fusion Science and Technology Roadmap outlining a strategic pathway to commercialization. It announced $128M for Fusion Innovative Research Engine Collaboratives, with seven teams selected to build a fusion innovation ecosystem bridging basic science and industry. There have been contracts and cooperative agreements between DOE and firms such as Commonwealth Fusion, TAE Technologies, General Fusion, and others. ARPA-E programs such as ALPHA have provided seed and scale-up funding to private fusion companies and university labs, thus supporting high-risk, high-reward innovation that DOE typically doesn’t fund. The administration also began developing regulatory pathways for fusion, for which existing nuclear regulatory processes designed for fission are ill suited.

In the long term, potentially the Trump administration’s most ambitious industrial policy is the Genesis Mission. Biden had treated AI primarily as a problem technology in need of regulation, emphasizing safety, risk management, ethics, guardrails, and funding across agencies. In its place is now a diffuse but potentially potent federal effort to proactively deploy AI into multiple industries as a source of national competitive advantage. Genesis aims, for example, to use AI to accelerate discovery in materials, energy, biomedicine, and advanced manufacturing. It covers federal data, compute power, AI models, and mission-oriented R&D. It will attempt to accelerate design-test cycles, generating hypotheses, and optimizing processes. It is not a grant program in its own right, but an agenda to coordinate the existing funding activities of many federal agencies. The Department of Energy, for example, recently announced a $320 million investment in cloud computing, AI models, robotics, and the underlying capabilities of AI.

It is high time to debunk the myths used to disparage the value of growing American manufacturing output, creating American manufacturing jobs, and holding onto what we still have. But before we can address them, three things must be understood:

First, only profitable, wealth-creating manufacturers can afford to pay high salaries and provide attractive benefits. So, any US administration that wants to grow the number of good jobs must focus first and foremost on creating and sustaining such industries. A systematic industrial policy, like the one we sketched out in our book Industrial Policy for the United States: Winning the Competition for Good Jobs and High-Value Industries (Cambridge University Press, 2025), is required to accomplish this.

Second, the US needs proactive industrial policies not only in manufacturing. For example, it urgently needs sound industrial policies for non-manufacturing industries such as artificial intelligence, quantum computing, and fusion power. But manufacturing is still a uniquely important sector, for reasons we explain below.

Third, the question of why the US lost so many manufacturing jobs in recent decades, despite being politically inflamed, is distinct from the question of whether we’d be better off with more. So even if the main cause of America’s job losses in recent decades was something other than bad trade policy (it was bad trade policy, but that’s a topic for another day) gaining manufacturing jobs is still a good idea.

Now, the main argument! We’ll start by pointing out that manufacturing is a larger part of the economy than is often claimed – usually with the intent of arguing that even if it is in decline, it’s just not that important and the energies of economic reform are better spent elsewhere. (You can find this line of argument here, here, here, here, here, and here.)

For example, one often hears that manufacturing is “only” 10 percent of GDP and 8 percent of jobs. It is even argued that manufacturing is destined to benignly dwindle to a tiny percentage of the workforce, just as farming did before it. (That argument can be found here, here, here, here, here, here, here, and here.)

But the “just like farming” story is an historical analogy that completely ignores the huge differences in the economic dynamics of farming and manufacturing.

In any case, employment percentages are not a reliable guide to the importance of a sector. Restaurant workers are about seven percent of employment, while electric power employment is one third of one percent (!) But which sector’s disappearance would be an inconvenience, and which would collapse modern civilization? Which sector is the enabler of many other sectors, and which sector enables almost nothing else?

Moreover, the oft-quoted 10 percent figure is an artifact of statistical definitions that assign the inputs into manufacturing, which range from iron ore to accounting services, to other sectors. Measured without subtracting these inputs, manufacturing – as of 2015, the date of the most recent detailed study – accounted for not 12 percent, but 35 percent, of GDP, and not 9 percent, but 21 percent, of all jobs.

Another way to say this is that every manufacturing job supports 1.4 jobs outside manufacturing, the highest multiplier of any major sector. Due to classification issues, this figure may even be an understatement: Almost a quarter of the jobs in the computer systems design and related services industry, for example, were in 2015 attributable to manufacturing.

The trade deficit, despite the seductive “money for nothing” myth that it doesn’t matter, reduces our nation’s wealth every year. A deficit means we are paying foreigners for the things they export to us not with things we produce today, but with things we produced yesterday or things we promise to produce in the future. The former takes place when we sell off existing financial assets to them, the latter when we borrow money from abroad. Either way, we own less and owe more, so we are poorer.

The vast majority of the deficit – $1 trillion – consists of manufactured goods. We consume a trillion dollars more manufactured goods than we produce. Because regaining US markets is much easier than penetrating generally protected foreign markets, manufacturing more here is the largest-scale, most feasible strategy for reducing the overall deficit.

The trade deficit could feasibly be reduced to a long-term average of zero by a combination of tariffs and the deliberate downward revaluation of the dollar by means of moderate controls (explained here) on inflows of foreign capital.

This would definitely result in more US manufacturing jobs than if existing policies were continued. Note those last words: “than if existing policies were continued.” An oft-neglected issue when discussing manufacturing job numbers is that of baselines, i.e., what alternative numbers are the projected results being compared to.

It is sometimes argued that zeroing out our trade deficit would not restore manufacturing to the share of GDP or employment it used to have. For example, Paul Krugman said this here.

This is certainly true. But it is also irrelevant, as no rational U.S. industrial policy would ever aim at restoring manufacturing’s previous share of GDP or employment per se – let alone assume that zeroing out the trade deficit would accomplish this.

Similarly, one sometimes hears that there’s no point bringing back production of goods like t-shirts and toys to the U.S. This is also perfectly true, but reshoring primitive industries should not be part of any future US industrial policy.

This raises the question of what kind of industries should the U.S. try to protect or reshore? A good place to start is with the major goods categories in which the U.S. ran a trade deficit in Q1 2025 (source and source of the original data):

To winnow this list, let us first eliminate raw materials, then products like clothing that are produced with low-productivity, low-wage labor. For products like electronics that have both low-productivity final assembly and high-productivity component manufacturing in their supply chains, eliminate the former.

The implication is that the U.S. should reshore significant production in industries such as machinery, electronics, motor vehicles, pharmaceuticals, organic chemicals, steel, medical instruments, and aluminum. To this should be added products with special relevance to national security and health, such as aluminum, copper, rare earths, active pharmaceutical ingredients (including their precursors) and generic drugs.

We should then add to this list important products of the future, many of which are at risk of never being produced in the U.S. unless policy proactively ensures that they are. A list of such industries has been issued by the White House Office of Science and Technology Policy, and includes:

Recovering, or gaining for the first time, such high-productivity industries would raise our economy’s average labor productivity, and thus necessarily create more high-productivity jobs than would otherwise have existed. (Moving workers from flipping burgers to assembling cars is the classic example.) Other things being equal, high-productivity jobs are high-wage jobs, so this would raise the incomes of American workers.

To benefit from increased manufacturing output, the U.S. wouldn’t even need to increase the number of its manufacturing jobs. This is true because if manufacturing output grew without job gains, it would, other things being equal, mean an increase in average labor productivity. The optimal U.S. economy would thus definitely have a higher manufacturing GDP share, but it might or might not also have a larger manufacturing employment share.

Insofar as a larger manufacturing sector did deliver more manufacturing jobs, these jobs would mostly be, contrary to another bundle of myths, attractive. For example, one often hears that manufacturing work is so dirty, dangerous, and unpleasant that Americans don’t want it anymore. (Examples of this argument appeared here, here, here, here, here, here, and here.)

But with limited exceptions, this is only true of the low-end industries that no rational American industrial policy would target. For high-end manufacturing such as cars, aircraft, and semiconductors and many mid-tech industries, manufacturing jobs today exist on a spectrum of physical rigor. Many are more “technician” than “factory hand” jobs in the old-fashioned sense. Many are white-collar engineering jobs, sometimes with advanced credentials.

It is no accident that recent new auto plants in the US, for example, have attracted many more applicants than the available positions for these “jobs Americans don’t want” (source, source, source, source). Although 75 percent of Americans, according to a 2024 poll, say they’re not personally interested in having a manufacturing job, this still leaves 25 percent who are – more than double the percentage now employed in the sector.

Even the more physically demanding jobs in contemporary manufacturing are comparable in their working conditions with construction and other industries that nobody claims Americans have lost interest in. And let’s not forget that “unpleasant” depends on the pay: Every profession has its hardship posts that attract people who consider the extra pay worthwhile. We rarely hear that American don’t want to work on offshore drilling platforms, where average pay is $100,000 a year.

Granted, manufacturing jobs are often still noisy (especially for products made of metal) and regimented (because they function at the cadence of machinery). But they are rarely hellish like in the old days. The reader is strongly advised to find a local factory that gives tours and have a look at what their working conditions are really like today.

The flip side of disparaging the feasibility of reshoring manufacturing jobs is to claim that America already has too many – so many, in fact, that such jobs are going begging and reshoring more is therefore not feasible. Thus some people have portentously reported facts like “As of July 2025, 400,000 manufacturing jobs were going unfilled for lack of qualified candidates.” (Examples appear here, here, here, and here.) But this widely quoted figure is a nothingburger: There are still 13 million manufacturing jobs in the U.S., so 400,000 jobs is only about 3 percent, which is similar to the percentage of unfilled positions in other fields.

More generally, “It’s hard to find good help these days” is what people on the hiring end always say. Businesses that are willing to pay for good people can find them, but, importantly, sometimes with a lag for the time it takes for word to get out and applicants to get qualified. The right way to think about this problem is that America does not have a labor shortage, it has a training shortage. Training, often provided in schools, community colleges, et cetera, is generally not a market commodity, and is therefore susceptible to actual shortages, as opposed to market prices exceeding purchasers’ wishes.

One sometimes hears that manufacturing jobs no longer pay more than service jobs, so there’s no point in the U.S. trying to obtain more of them. (This was argued here, here, here, here, here, here, and here.) Supposedly, manufacturing was a more unionized sector (because it is easier to organize hundreds of people in a single plant than waitresses), which created a barrier to entry and thus a wage premium. Then unionization rates fell, unions lost leverage due to imports and nonunion competitors, and the premium went with it. [Paragraph revised 04/26.]

But this is actually not true. It only appears true if one uses an incomplete set of data from the Bureau of Labor Statistics, the Current Population Survey, which does not include bonuses and other one-time payments to workers. A more comprehensive dataset, the Quarterly Census of Employment and Wages, does include these, reveals a manufacturing wage premium in 2024 of 12.7%. Our erstwhile Coalition for a Prosperous America colleague Jeff Ferry has documented this in detail here. [Paragraph revised 04/26.]

But absent barriers to entry, labor markets tend to equalize wages at any given skill level. Therefore, a policy that successfully generated good jobs in one sector would not necessarily show this fact by producing jobs that paid more than comparably skilled ones in other sectors. Instead, it would push up wages in all sectors at the skill level of the jobs it created. A wage premium for a specific sector is nice for the workers that receive it, but is not necessarily an indicator of anything good or bad for the labor market as a whole.

Insofar as reshoring of manufacturing increases total demand for labor in the US, we may expect it to produce generalized upwards pressure on the labor market. But as long as the total employed population remains the same, we will be talking about shifting the demand curve for labor upward. We will not necessarily be increasing the total number of jobs, or even of manufacturing jobs, in the economy.

This logic admittedly falls apart, of course, if policies designed to create manufacturing jobs actually destroy them on net. Which is why we now address the argument that tariffs cost multiple jobs in downstream industries for every job they create in upstream ones.

There are two parts to the rebuttal. The first concerns the lack of evidence that this problem has already occurred, while the second concerns the fact that even if it does occur in future, it could be a price worth paying.

Regarding the first issue: because this claim is a generalization covering a vast range of industries with very different characteristics, it cannot be disproved in its entirety. But it can still be answered by debunking one widely publicized case in detail to show how weak the logic of this criticism has been.

Specifically, as one consulting firm put it,

Following the 2018 Trump’s tariffs,.. tariffs on steel may have led to an increase of 1,000 jobs in steel industry. However, according to a Federal Reserve Board of Governors’ study, increased input costs due to those tariffs are associated with 75,000 fewer jobs in the domestic manufacturing sector.

But the Fed study was seriously flawed. The losses it asserted were not empirical observations, but projections from questionable mathematical models. To wit: The authors assumed that an X% tariff on steel increases the price of steel by X%, then compared the share of a steel-consuming industry’s inputs covered by tariffs with changes in the industry’s employment levels, output, and prices. On this basis, they estimated a 4.1% increase in factory-gate prices.

All the study’s conclusions about job losses are derived from this 4.1% number. The problem is that its underlying assumptions about price behavior were wrong. To wit:

1. It is simply not empirically true that tariffs affect prices 1:1. One US International Trade Commission study found that the price of steel sold in the U.S. after the 2018-2019 tariff rose by just 2.4% (source). Why? Largely because consuming industries avoided the tariff by switching to domestic steel, whose producers had large slack capacity. Switching American consumption to domestic sources was, of course, the whole reason a tariff was imposed in the first place.

2. The PPI for steel-using sectors was essentially flat from the 2018 tariffs until COVID hit in 2020. The sharp jumps came during pandemic-era supply shortages and stimulus-driven demand surges, not during the 2018–2019 period studied by the Fed paper (source).

3. PPIs spiked across many industries, not just in those subject to tariffs.

Other industries that have significant domestic slack capacity will behave similarly to the steel industry when tariffed. As will, with a longer adjustment period, industries where new capacity can quickly be brought online.

The are two contrary cases:

1. Industries where the US lacks existing capacity and where it takes a long time to add new capacity. Here, there will be at least a transitional price shock, though this could be mitigated by phasing in tariffs as new US production ramps up.

2. Industries where foreign production is both at the world efficiency frontier and heavily subsidized. Here, US-made products are likely to be more expensive than imports indefinitely, and when this happens, one cannot rule out downstream job losses. The price shock will make downstream products more expensive, downstream producers will sell less, and they will need fewer workers.

But even here, the case for tariffs is not necessarily invalidated. It will depend on the value of the industry being reshored. Why, for example, was the foreign nation subsidizing that industry in the first place? Quite likely because it is a strategically and/or economically important one, like electric cars, that both they and we are rational to want. So it will be worth paying a price to get it.

This is the classic tradeoff in industrial policy: does the long-term value of having an industry outweigh the short-term cost of reshoring or developing it? Can the US maintain economic leadership while being uncompetitive in an industry of this significance? And if developing said capacity is slow and difficult, this is precisely the sort of industry where tariff protection will be key to building it back up.

Finally, any defense of manufacturing jobs needs to address the argument that AI and robotics will soon change the world so much that it is not worth trying to regain such jobs. After all, if robots are going to do all the work, why should we even care whether people have jobs or not, let alone manufacturing jobs? (This or related, slightly less extreme, arguments have appeared here, here, here, here, here, here, here, and here.)

But this argument is the intellectual equivalent of unconditional surrender: Because nobody yet knows what AI and robotics will do in the long run, all current economic policies and assumptions can hypothetically become moot, not just the pursuit of manufacturing jobs. Giving in to this sort of thinking is a recipe for policy paralysis on every forward-looking economic issue.

In any case, whatever happens, the U.S. will have to traverse a lengthy transition period before this this epochal shift arrives. Over, say, the next 15 years, we will be dealing with deployment of AI and robotics into an environment whose conditions are largely, if decreasingly over time, similar to those of today. So current arguments about the value of manufacturing jobs will still apply for years to come, and still define what we should do now.

Furthermore, the U.S. will be poorly positioned for whatever the future has in store if we fall behind during this transition period. For example, the U.S. is weak in industrial robotics: We have 6 of the world’s top 10 academic programs in robotics, but not one of the world’s 10 largest industrial robotics companies is owned and based here. Our robot density in manufacturing lags rival nations, even nations like China that still have vast low-wage manufacturing workforces. So even if 100% robotization is the future of manufacturing, America is not on track to succeed in that future.

Importantly, if the most extreme predictions do come true, the “robots” argument only tells against the desirability of manufacturing jobs. It does nothing to diminish the desirability of greater US manufacturing output. In the long term, even if manufacturing ends up as an activity based almost entirely on capital and technology, with little or no direct labor content, it will still be in America’s interest that this manufacturing take place in the U.S. If it does, then the distribution of its output will be up to us, settled by our political system according to our values. If it doesn’t, then we will have to accept whatever other nations are prepared to give us.

(An audio version of this article is here.)

Industrial policy is back on the national agenda. From Marco Rubio on the right to Elizabeth Warren on the left, a bipartisan consensus is emerging that America’s deindustrialization has passed the point where governmental passivity is appropriate. The intensifying national security threat from China, and supply chain vulnerabilities newly revealed by the COVID pandemic, have extended the urgency beyond purely economic concerns.

And yet, the current ferment is in danger of ending up like previous versions in the 1970s, 1980s, and 1990s: nowhere. Conceiving and marshaling the political will to adopt and effectively administer a sound industrial policy will require a solid, defensible theory of its economic foundations, and such a theory has not been put forth. The absence of such theory was a major factor dooming previous efforts, and without it, the U.S. will continue to be restricted to one-off tactical moves, against only the most obvious problems and using only the most obvious solutions. It will not be able to undertake the coordinated, strategic-scale, all-of-government solutions it needs.

Mainstream contemporary American economics takes a dim view of industrial policy. As Paul Krugman correctly wrote in the early 1990s, when American worries about the Japanese challenge were still fresh, and there was a surge of interest in industrial policy,

Economists have been extremely negative about the idea of industrial policy even in principle. The general presumption of most economic theory is that the best industrial policy is to let the market work – that decentralized incentives of the marketplace will push resources to the places with the highest expected return, and that no second-guessing of market decisions is necessary or desirable.^[1]

But there is an alternative view of the economics underlying industrial policy, a view whose intellectual components are consistent with mainstream economics but which leads to pro-industrial policy conclusions. While mainstream economics holds that the market is always, or with only a few exceptions, best, this alternative view holds that economic success requires not only a) letting the market work, but also b) systematically exploiting the market’s shortcomings. And because such exploitation, by definition, involves things the market can’t do, state intervention is often the only way to accomplish it. This implies governments should pursue the systematic and strategic exploitation of the market’s shortcomings to support the growth of certain economic activities, and industries, which are especially beneficial to the competitiveness and growth of nations.

This alternative theory is closer than mainstream economics to how real-world businesses function. Businesses necessarily make money by exploiting market shortcomings, because perfectly competitive markets squeeze profits towards zero. Everybody wants perfect markets when they are the buyer, but imperfect markets when they are the seller. The consumer-side view of economics and the producer-side view thus differ. One way to understand the alternative approach is thus to grasp that while mainstream economics is biased to take seriously almost entirely the former, the alternative view gives both more equal consideration.

Why Markets Can’t Do Everything

The shortcomings of markets are a large topic, with many of them having little relevance to industrial policy. Those that concern us here are:

1. Externalities

2. Time horizons

3. Systemic effects

4. Fundamental innovations

5. Static vs. dynamic efficiency

6. Increasing returns

Because each of these creates at least the possibility of a market producing a suboptimal economic outcome, they create, in principle, the possibility that competent governmental interventions could produce something better. So, let us examine them in detail.

#1: Externalities

An externality occurs whenever an economic activity “leaks” costs or benefits elsewhere, without the entity performing the activity incurring the cost or benefit itself.

Some externalities are negative, such as environmental pollution. This reduces the value of the environment without this cost being borne by the firm doing the harm. The classic governmental responses are regulations and fines for polluters.

Other externalities are positive, such as a company developing a new technology that will create value all over the economy. Due to the limits of the patent system and other constraints, that firm may not be able to fully capture that value. The classic governmental responses include tax credits for, and direct investments in, research and development.

Research and development also exemplify the problem of appropriability, which refers to when a firm does something that generates economic value but cannot fully capture this value as profit. One example is training personnel who may then go to work for another company. The classic governmental response? Everything from free public schools to subsidies for worker training. Another appropriability problem occurs when a firm invests in upgrading its supplier companies, which will then be able to sell to anyone, not just the company paying for the upgrades. Governments have responded to this by funding technology extension services that help small and medium enterprises upgrade themselves.

#2: Time Horizons

Short-term investing can accomplish some economic tasks, but many of the most important investments must be long term, and there is nothing that guarantees capitalists will have long time horizons. But without long-term investments, whose payoff may not come for years or even decades, businesses won’t develop the next generation of technology, but will stick with variations on what already exists. Companies with short time horizons will cede market after market to rivals with longer time horizons. Entire industries can be out-competed by foreign rivals with time horizons artificially lengthened by foreign industrial policies.

Short-term thinking may be perfectly justifiable for the firms engaging in it, and even more so for managers compensated for short-term results. But society, as opposed to any one firm, goes on forever, so it has good reasons to prefer long time horizons. Thus, one part of industrial policy is about lengthening effective time horizons, with tools such as more readily available, cheaper, and more patient capital. Said capital is often generated by the many, often hidden, policies governments use to coerce, or incentivize, people and firms to save more. Other policies concern corporate-governance structures that favor patient investing. Governments also do things like offer tax credits, or incentivize banks to lend preferentially to projects expected to have long-term payoffs for the economy as a whole, and fund high-risk technological research whose expected payoff is too far in the future to interest private firms.

#3: Systemic Effects

Systemic effects refer to aspects of the economy that are not optimizable at the level of individual economic actors such as people and firms. Many things are optimizable at this level, which is why market economies work. But a market economy, by definition, is not centrally planned, so it will have a hard time achieving outcomes that can’t emerge naturally from the actions of its individual profit-seeking players. As a result, even if every individual player does what’s best for itself, an optimal result for the economy as a whole will not result.

For example, “quality of demand” refers to the fact that firms selling anything from jewelry to jet engines to exceptionally sophisticated customers will be driven by their demands to excel. Competitive pressure to respond to these demands will drive producers to upgrade their capabilities, not just maximize profits while standing pat. But no individual customer will see a financial reward commensurate with the benefits that the higher quality of their demand confers on the whole economy. As a result, markets alone will not always induce optimal quality of demand, and governments have therefore deliberately pushed their nation’s industries to upgrade more aggressively than immediate profitability would have caused. Quality of supply has an analogous logic.

#4: Fundamental Innovations

Per-capita economic growth means that the economy produced more goods and services on Tuesday than it did on Monday because something changed. But what? Adding capital will grow an economy, but how should the money be invested? Simply buying every mechanic in a factory a second wrench won’t make that factory more productive. They need something they don’t already have, something that will enable them to produce more output with the same amount of labor. A better wrench, or perhaps some automation. In other words, better technology.

Technological advance, in developing nations, can mean adoption of technologies already existing elsewhere. But in developed nations, which by definition already use existing technologies close to their full potential, this means creating new technologies. As a result, in developed economies, innovation is necessary for growth in per capita output.[2]

Unfortunately, free-market economics lacks a good explanation for why and how technology advances. It is on reasonable ground explaining advances developed by the private sector under free-market conditions. But the most important technologies have tended to originate elsewhere: in government laboratories, in publicly supported non-profits like universities, and in government-sanctioned monopolies like AT&T. The Internet itself, for example, as opposed to the websites on it, was created to enable government scientists to share data. Jet engines were first used in military aircraft and only later adapted for use in commercial airliners. Most fundamental technologies, the ones with vast effects throughout the economy, are not directly ownable or saleable. And before they can be deployed, they must pass through long stages of risky and expensive development with no supporting revenue stream. Because the market can’t optimize their creation, the most economically crucial technological advances, especially since WWII, have not been produced by it.

#5: Static vs. Dynamic Efficiency

With respect to economic growth, nations face not one question, but two:

1. What is the most productive thing to do today, given the current state of their productive resources?

2. How can they transition to a higher state of the productive resources tomorrow?

Markets, through the efforts of private businesses, are generally good at achieving the former, so-called “static” efficiency. But they are much less good at the latter, “dynamic” efficiency. This is clearest in developing nations: growth is clearly about turning Burkina Faso into South Korea, not about being the most efficient possible Burkina Faso forever. But it is also true for developed economies.

Static and dynamic efficiency are not only different, but also can conflict. In other words, it can sometimes be rational to do things that are inefficient in the short run because there is a long-run payoff. For example, nations sometimes impose protectionism for infant industries, forcing their consumers to buy more-expensive domestically produced goods, to get a lucrative new industry off the ground. This difficult but fruitful tradeoff between static and dynamic efficiency is, in fact, the single most important theoretical basis of industrial policy.

#6: Increasing Returns

Increasing returns means that for a given increase in inputs, the increased value of output exceeds the increased cost of the inputs. For example, if producing a product requires a machine plus raw materials, then cost-per-unit will consist of a fixed cost per unit for the raw materials, plus a cost for the machine which falls as its cost is spread over more output. So, 10% more money spent on inputs might deliver, say, 11% more value of output, 11% more money 12% more, and so on. The opposite of increasing returns is diminishing returns: after a certain point, 10% more money for inputs delivers only 9% more value of output, then 8%, and so on.

In industries where increasing returns are present, the market will not be free. This is clearest in the extreme case, increasing returns that go on forever, costs dropping ad infinitum with each additional unit. (A search engine like Google is probably the thing closest to this in the real world.) In this case, the most efficient outcome is one producer for the entire planet, because two producers would each produce half as much, and therefore have higher costs.

A free market will initially deliver this one-producer outcome, because whenever one producer has slightly higher volume, it will have lower costs, be more competitive, gain volume, lower its costs more, and inexorably drive its rivals out of business. But then the free market will come to an end, because there will be a monopoly. Any new firm entering the industry, unless it can do so with the same production volume as the incumbent, will not be competitive.

Most increasing returns are not this extreme, but the more increasing returns are present in an industry, the less free the market will be. There will be so-called “imperfect” competition: not one single producer, but only a few, so that each firm will be big enough for its actions to affect the entire market. In other words, an oligopoly. Increasing returns naturally tend to make the industries in which they occur oligopolistic.

Advantageous Economic Activities

Increasing returns are the first of a set of characteristics of what we will call “advantageous” economic activities. These characteristics tend to coincide, reinforce each other, and build up cumulative causation. (That’s when something happens because of multiple causes over time.) To wit:

1. Increasing returns

2. Pricing power

3. Technological dynamism

4. Dynamic rent seeking

5. Synergies

Economic activities with these characteristics produce dynamic efficiency, causing an economy to repeatedly gain in productivity and thus grow. The more a nation’s economy consists of such activities, the more prosperous it will be today, and the better its growth prospects will be for tomorrow. Industrial policy, at root, is about increasing the quantity of such activities in an economy.

Pricing power occurs whenever a firm can choose to charge more, and sell less, or charge less, and sell more, as opposed to just having to accept the market price. (The latter happens when there are many small competitors: each will be able to sell as much as it likes at the market price, but nothing if it goes a penny over.) Pricing power requires there be some factor present that reduces competition, so pricing power tends to appear when there are increasing returns.

Pricing power means, obviously, that sellers get a higher price than they would under perfect competition. It can come from being the only coffee shop on the block, but the pricing power relevant to industrial policy is the result of producing things that are hard to produce, such as jet engines or smartphones. When only a small number of highly skilled firms can produce a given product, an oligopoly will develop, and its members will have pricing power. In fact, whenever know-how is the major production cost in an industry, which is generally intrinsic to cutting-edge industries, there will necessarily be increasing returns, because once that know-how has been paid for, each additional unit of product amortizes its cost over more output.

Next, technological dynamism. “Hard to produce” usually means needing sophisticated technology. And because technology ages over time, what was once cutting-edge inexorably becomes commonplace and easy to produce. So, what one really wants as a producer is ongoing technological change, so that one can stick forever with the difficult, relatively new, technology that results in pricing power. This logic creates an incentive to produce this technological change oneself, through R&D.

Pursuing profits in this way is called “dynamic rent seeking.” (The term “rent,” in the peculiar usage standard in economics, means profits that exceed what a free market would grant.) Dynamic rent-seeking is based on continual innovation and upgrading to hold onto the rent. Its opposite is static rent seeking, which means seeking some privilege, such as a legal monopoly, which enables charging a premium without doing anything to earn it. Static rent-seeking does an economy no good, but dynamic rent-seeking increases productivity and thus produces economic growth.

Finally, synergies. Some economic activities, when they first emerge or later improve their productivity, enable other activities to emerge, or improve their productivity. For example, cheap, mass-produced steel, whose first major market was railroad rails, made skyscrapers possible. Vacuum tubes originally flowered as an industry to build radios, then enabled the construction of televisions. Computer chips were created when computers were room-sized, but eventually became small and powerful enough to enable smartphones. Smartphones, in turn, enabled ride-hailing services. Jet engines developed for military aircraft enabled in turn passenger jets, mass tourism, and ultimately… Disney World.

When synergies are present, advances in one part of the economy tend to push forward other parts, too. This is important for long term economic growth because there is a limit to how much any one industry can grow, simply because there is a limit to how much of any one product consumers will buy.

When are synergies absent? Consider single-export economies, producers of bananas or petroleum: a nation may be poor if it produces only bananas, or rich if it produces only petroleum, but either way, it has no good path to producing anything else.

Advantageous Industries Produce Growth

Advantageous economic activities are activities, not industries per se. But economic activities take place in particular industries. Some industries are “tightly packaged:” one must perform all, or nearly all, of their activities, to perform any of them. Others are “loosely packaged:” some of their activities can be delegated to other firms, or geographically scattered, so a country can host some but not others. Thus, in the case of tightly packaged industries, the pursuit of advantageous activities means pursuing the entire industry.

Advantageous industries have six key characteristics:

1. High income-elasticity of demand

2. Susceptibility to repeated improvement.

3. Competition not on pure price

4. Capacity to absorb investment

5. Path dependence

6. Human capital accumulation

Advantageous industries tend to produce products for which income elasticity of demand is high. For example, people buy more (and more expensive) cars as their incomes go up. They do not buy more milk. As a result, demand for such products can rise along with the overall income growth of the economy. Therefore, productivity gains in making such products don’t just drive down their price, as output can rise along with productivity. Productivity gains largely accrue to producers, in the form of higher profits and wages, rather than just to consumers in the form of lower prices. (The latter is what happens with products of relatively inelastic demand, such as commodity foodstuffs.)

Advantageous industries tend to produce goods susceptible to repeated improvement, such as computers or airplanes, while disadvantageous industries produce goods whose character is basically fixed, like fruit or t-shirts. When a product exhibits meaningful variety, producers can establish mini-monopolies in specific niches, leading to dynamic rent-seeking. Product variety also increases opportunities for innovation, both because inventing new varieties is itself an innovation, and because innovation can now advance on multiple fronts.

Advantageous industries tend not to compete on pure price. (Price is obviously a factor with anything sold for money, but it is relatively less important in such industries.) They tend, instead, to compete on quality, technology, reliability, reputation, marketing, service, variety, style, under-standing of buyer needs, rapid innovation, vendor financing, managerial sophistication, and customer relationships. These forms of competition make competition based on cheap labor much less relevant, especially cheap foreign labor, a crucial determinant of a traded industry’s ability to raise its workers’ incomes.

Advantageous industries tend to have a high capacity to absorb investment. Buying another $1,000,000 worth of tractors for a coffee plantation that already has them won’t increase its productivity much. But putting $1,000,000 into improved production machinery in a television factory will. Advantageous industries tend to activate a virtuous cycle, in which innovation absorbs capital and repays it by raising profitability, generating more capital and repeating the cycle. Economy-wide growth often involves a multi-industry virtuous cycle, in which the upgrading of one industry causes others to upgrade and so on.

Advantageous industries tend to exhibit “path dependence.” That is, having advantageous industries makes it easier for a nation to acquire other advantageous industries. For example, nations that produced radios were better equipped to start producing televisions, because they had already mastered key technologies like the vacuum tube. As a result, economic growth is path dependent: what nations can produce tomorrow depends on what they produce today. This is why much industrial policy, historically, has centered on using various policies, from infant-industry protection to outright cash subsidies, to break into industries expected to lead somewhere. Quintessentially, this has meant getting out of agriculture and raw materials and into manufacturing (especially of products sophisticated for the time) although a small segment of advanced service industries has been entering the advantageous category since the late 1970s.

Advantageous industries tend to experience human capital accumulation, because they use technology the workers must have training to operate. They thus create a premium on more skilled, not just the cheapest possible, workforce. This human capital, manifested in the workers themselves, encourages better treatment of labor, for the same reason factory owners do not let valuable machinery rust away. Advantageous industries thus make at least possible the “countervailing powers,” like bargaining leverage by unions, that spread the profits of industry beyond its owners. Such leverage is not guaranteed, but workers can’t bargain for a share of profits that aren’t there. Rising worker incomes also provide the purchasing power to sustain growth, and as incomes rise, consumers demand better products, driving the industries of the nation (which will generally depend in large part upon domestic sales) to upgrade.

It is not guaranteed that any given advantageous industry will have all six of the above qualities. They are, however, interlinked, mutually dependent, and do tend to appear as a package. For example, technological advance is linked to expanding production because it is much easier to invest in new, better technology when one is adding new machinery to expand production. And expanding production is linked, of course, to elastic demand. Remember, finally, that the advantageous industries that count are big industries, like cars or airplanes, or groups of many small industries that together amount to a lot, not tiny niche industries.

Disadvantageous Industries

Advantageousness vs. disadvantageousness is not binary, but rather a sliding scale from very advantageous to very disadvantageous, with many industries falling in between. In disadvantageous industries, all or some of the six previously discussed dynamics are absent – or, worse, run in reverse. Such industries not only tend to have low wages and profits today, but also do not lead to better industries tomorrow. For centuries, “disadvantageous” has meant most agriculture, natural-resource extraction, and services like retail. And since the mid-1970s, low-skilled manufacturing has been inexorably joining this category.

Agriculture and natural resources generally exhibit diminishing, not increasing, returns, as once the best land and most accessible resource deposits have been exploited, increasing production means turning to inferior land and deposits. Such industries rarely have any way of acquiring pricing power, as they generally produce undifferentiated commodities. Because income elasticity of demand for their products is low, productivity growth tends to flow to consumers, in the form of lower prices, rather than to producers, in the form of higher wages and profits. American farmers, for example, use advanced technologies from genetically engineered seeds to satellite positioning systems for their tractors, but struggle to stay solvent.

Fields such as retail and restaurants have far less scope for productivity improvement than manufacturing: a 1950s diner isn’t that different from one today, while a 1950s auto assembly line would be so uncompetitive today that nobody would build one. Technological innovations in disadvantageous industries tend to come from other industries: farmers don’t themselves invent satellite-navigating tractors or genetically engineered corn. As a result, when innovation does occur, it tends to just increase the productivity of all producers at once, driving down the price. Agricultural prices also tend to be volatile: soybean or banana prices can fall or rise by 50 percent year-to-year, but not car prices.

Disadvantageous industries generally lack synergies with other industries. When the Santa Clara Valley in California (better known today as Silicon Valley) specialized in plum production before WWII, this brought about a prune industry and a fruit-canning industry, but little more. When the same region became a center for aerospace and defense electronics, this led to transistors, then integrated circuits, then computer hardware, computer software, and, contemporaneously, a venture capital industry feeding on and nurturing all these industries. The output of all these industries are worth many, many times what the region’s fruit production once was.

The presence of advantageous industries in an economy generally improves wages in even that economy’s disadvantageous industries. For example, growth of manufacturing in a developing nation raises the incomes of the nation’s farmers, as rising manufacturing wages create more demand, and thus higher prices, for farm goods that are consumed locally because of transport costs and perishability. Advantageous industries often also help set a “wage floor” for all local industries, preventing unemployed workers from crowding into disadvantageous industries and dragging down productivity due to these industries’ diminishing returns.

Underdeveloped nations are poor because their economies are predominantly composed of disadvantageous industries, and they can’t find a way up and out of such industries. Developed nations are not wealthier because they engage in the same economic activities as developing nations, only with greater productivity: they engage in fundamentally different economic activities because they have fundamentally different industries. (Commodity agriculture, for example, does exist in developed nations, but is a small percentage of GDP and employment.) Nations that used to be poor and agricultural, like South Korea in 1960, didn’t become rich by finding ways to produce rice and fish (South Korea’s main industries then) with 20 times the productivity. These nations used proactive industrial policies to break into industries, like steel and cars, where 20 times the per-worker productivity of a rice-growing peasant farmer is the norm.

Note, finally, that having a lot of advantageous industries is not the same thing as merely being rich, as small-population states with large natural resources, such as Argentina in 1900 (beef) or Kuwait today (oil), can have high per-capita incomes. But the resource-based model doesn’t scale: there has never been a large nation that was rich purely on the strength of natural resources, because no nation has ever had enough resources. But advantageous industries, on the other hand, can scale, and when their ability to scale is exhausted and their products commoditized, they lead to new industries.

The Dynamics of Advantageous Industries

The core justification for industrial policy is that, for any given nation, the market alone will not automatically generate the optimum amount of advantageous activities and industries. This is so because the key characteristics of advantageous industries are shot through with non-free-market dynamics.

1. Increasing returns industries are, though lucrative, protected from entry by the superior scale of incumbents. The market, on its own, will thus tend to keep new entrants out. As a result, nations can benefit from subsidizing entry into these industries. (And nations already hosting such industries must beware that potential rivals may be doing the same.)

2. Pricing power exists when markets are not perfectly free, so getting it requires advancing into oligopolistic, increasing-returns industries, which in turn entails overcoming the incumbent scale advantages previously mentioned.

3. Fundamental technological advances generally do not come from the private sector operating under free market conditions. They come from public sector research, from the private sector enjoying public sector help, or from government-sanctioned monopolies.

4. Synergies do not generally result in profits for the parties creating them commensurate with the value they generate across the entire economy. As a result, the market doesn’t correctly “price” the industries that have these synergies, and therefore won’t create the optimal amount of them on its own.

5. The private sector, on its own, often has short time horizons, not the multi-decade time horizons needed to lift a nation from poverty to prosperity or keep it prosperous if it already is.

For these reasons, markets are less effective for breaking into and retaining advantageous industries than the combination of markets plus sound industrial policy. And in a competitive international environment, where rival nations are already using effective, all-of-government industrial policies, relying on markets alone will result in the loss of existing traded advantageous industries and failure to establish leading positions in those of the future. As nations become more developed, markets get closer to being sufficient, which is why industrial policy matters more in developing nations, but even fully developed nations never reach a state where markets alone are sufficient.

Advantageousness Changes over Time

How advantageous particular industries are changes as technology’s leading edge moves forward and yesterday’s advanced technology becomes mundane and diffuses around the world. Any given industry’s advantageousness will thus generally decline with the technological status of its products and production processes, although barriers to entry can slow this process. As a result, most industries must constantly upgrade to hold their advantageousness. (Different activities within an industry can, of course, evolve at different rates, and modern communications and transportation systems allow these activities to be geographically dispersed.) Thus, in traded industries, nations must move forward just to stay in place. This is mostly a healthy process, because it forces developed nations to upgrade and allows developing nations to develop by entering industries advanced nations have left. However, it also means developed nations can lose the advantageous industries they already have.

Industries that have lost advantageousness in recent years include laptops, which has largely ceased innovating and entered a phase of price competition between similar models. Decades ago, at the dawn of consumer society, products such as refrigerators, washing machines, microwave ovens, and bread-making machines were also once highly advantageous, but then they, too, declined.

Conversely, some industries have seen their advantageousness revive, thanks to new waves of innovation. The most obvious example today is automobiles, where advances in battery technology enabled electric cars, and artificial intelligence is on the way to enabling self-driving vehicles.

Other products have held their position through incremental innovations: televisions now have Bluetooth connectivity, ultra-high definition, curved screens, and intelligent TV. Refrigerators have TV screens in their doors, see-through doors, separate cooling chambers, in-door ice and water dispensers, and Internet connections.

Other industries have held advantageousness because their internal structures create strong barriers to entry. These include large integrated systems like aircraft, other major aerospace systems, military systems, flight simulators, factory automation systems, Formula I race cars, and offshore drilling platforms. They are design-intensive, small-volume, high priced sectors. Their products are generally made of complex, often custom-tailored, components, which are themselves high-value and produced in relatively small volumes. Component suppliers to these industries tend to be located close by, so they can work closely together on the design and integration issues inevitable with complex and continually innovating products. These suppliers are often themselves operating at the technological cutting edge, such as the capital equipment for semiconductor production.

Other industries that have held onto their advantageousness center on product design and continuous processing, rather than unit manufacturing: non-generic pharmaceuticals, oil refining, chemicals, biotechnology, and advanced materials. In these industries, the physical manufacturing step is often a relatively small fraction of the cost of the product. For many, accumulated intellectual property, in the form of patents and trade secrets, creates barriers to entry.

Proximity-sensitive manufacturing, such as custom kitchen cabinets and architectural components, lacks some attributes of advantageous industries, but makes up for this by being sheltered from foreign, and sometimes even national, competition. Also sheltered from long-distance competition are industries like food processing that are tied to local supplies of agricultural inputs, especially those that are perishable, finicky to transport, or need to be consumed shortly after production. Similar factors apply to industries, like many construction materials, where transport costs are high relative to the price of the product.

Evolution of production technology is not, of course, the only reason industries migrate from developed nations. This migration has also occurred because of advances in transportation, communication, business software, and other things that make it easier for firms to take advantage of cheaper foreign labor. The invention of the container ship was a key factor. Others include the ability of foreign nations to artificially suppress the value of their currencies, features of private-sector markets that keep the dollar overvalued, and the subsidies and other mercantilist tactics foreign governments use in pursuit of advantageous industries. In recent decades, trade agreements have facilitated the migration of production abroad by guaranteeing access to the U.S. market and protecting investments in production facilities. The U.S. has also made deliberate choices, such as allowing imports of textiles from certain foreign nations and deliberately transferring technology to U.S. allies in service of non-economic foreign-policy objectives.

Industrial Policy Is Not Just about High Technology

“Advantageous” does not merely collapse in practice into “high tech.” Industrial policies singularly focused on pursuing more-advanced technology as an economic strategy have a poor track record all over the world. (Consider the technological triumph, but commercial failure, of the supersonic Anglo-French Concorde aircraft.)

From an employment point of view, everybody can’t work at Apple or its equivalents. There aren’t enough jobs in these industries, and some people are too old or uneducable to be retrained for them, or live in places where there are none. There are also less advantageous industries that society needs simply in order to function. Many “mid-range” industries have non-zero but modest amounts of increasing returns, pricing power, technological dynamism, and synergies: utilities, infrastructure, construction, transport, healthcare, energy production, and government. Together, these industries represent a large portion of GDP and employment. Many cannot be performed remotely, protecting them from foreign competition and making them viable sources of large numbers of at least medium-wage jobs. But precisely because these industries are so large in terms of GDP and employment, it is important to ensure that they make full use of available technology to become as advantageous as possible.

Productivity gains come not from inventing technologies per se, but from deploying them. As the consulting firm McKinsey has observed, most jobs created by technology are outside the technology-producing sector itself, estimating that the personal computer enabled the creation of 15.8 million net new jobs in the U.S. since 1980. [3] The number of people employed building PCs in the U.S. in 2019? A mere 156,000.[4] As a result, in addition to technology-creating industries, a nation also needs industries capable of absorbing technology. Some industries are thus advantageous, in part, because they serve as matrices for the deployment of technologies developed elsewhere. Without a thriving automobile industry in this country, for example, the productivity gains from newly emerging robotic technologies in auto manufacturing will accrue to other nations. Similarly, America loses many potentially high-productivity jobs when advanced products designed here are made elsewhere – and this tends to result in eventually losing the design work, too.

Advantageous Industries Benefit Workers Generally

There are a limited number of big sectors where large, ongoing productivity gains are possible: manufacturing and a few others, such as information technology. These are where R&D pays off in profitable productivity gains, and thus why manufacturing accounts for 70% of U.S. R&D. Productivity growth in these sectors is the ultimate source of economy-wide income growth.

As previously noted, a restaurant today, and one from 1950, are not all that different. In comparison, a steel mill, or a telephone switchboard, are unrecognizably different: productivity at the mill, in man-hours per ton, is about eight times higher, and the switchboard has almost disappeared, although the task is still performed out of sight electronically.[5] But it is because of productivity growth in industries like steel and telecommunications that more people can afford to eat in restaurants than in 1950.

Barbers earn over ten times as much in Germany as in the Philippines.[6] Do they use more capital or better technology in Germany? Not much. The main difference is that wages generally are higher in Germany, so barbers have to be paid more, or nobody would be one. But how did wages get to be higher there? Somebody’s productivity must have gone way up, because average wages must reflect an economy’s average productivity, and it wasn’t the productivity of barbers, which hasn’t changed much in millennia. Or the productivity of other local service jobs, like retail, dining, and local government. (In America, these make up two thirds of total jobs.)[7] Instead, rich nations are rich because they experienced wave after wave of productivity surges in the advantageous industries of the day over many decades. This allowed capital, skills, and technology to relentlessly accumulate, pulling up wages across the whole economy. As a result, everybody in the economy, not just workers in the most advantageous industries, has a stake in the health of such industries – though how much workers benefit also depends on other factors affecting labor’s bargaining power.

The Core Concepts of Industrial Policy Economics

The foregoing argument does not, of course, evaluate the effectiveness of the hundreds of specific policies, from currency manipulation to subsidies for scientific research, that governments use in the pursuit of more-advantageous economic activities and industries. Nor does it imply a specific set of such policies for the United States, a large and legitimate topic for debate. But it does explain why industrial policy can work:

1. Economically advantageous activities and industries exist.

2. They are the key to economic growth and prosperity.

3. A laissez faire policy will not deliver the maximum possible amount of them.

4. Industrial policy interventions can potentially deliver more of them.

These four points, together, are the foundation for any sound industrial policy. They imply that every country, including the U.S., can benefit from good industrial policy and provide, at a high level of generalization, the logic for evaluating specific proposals.

NOTES

[1] Paul Krugman, “The Current Case for Industrial Policy,” Chapter 7 in Dominique Salvatore, ed., Protection and World Welfare, Cambridge University Press, 1993, p. 160.

[2] Innovation doesn’t only mean technology. Consider the mail-order catalog, or stock exchanges, or retail franchising: all these innovations in business technique unlocked new opportunities for growth. But technology is the main event. And this is not to say that innovation itself is the only bottleneck of growth.

[3] “Jobs Lost, Jobs Gained: Workforce Transitions in an Time of Automation,” McKinsey Global Institute, December 2017, p. 5.

[4] Computer and peripheral equipment manufacturing, excluding digital camera manufacturing (NAICS 3341), 2018: Bureau of Labor Statistics, bls.gov/emp/tables/employment-by-major-industry-sector.htm, table 2.7, row 86, accessed January 5, 2020.

[5] Steel production man-hours per ton in the U.S.: 1950=15.1 (from Lee, K., and Ki, J. (2017). “Rise of latecomers and catch-up cycles in the world steel industry,” Research Policy, 46(2): 365-375 (p. 368)) 2017=1.9 (from American Iron and Steel Institute (2019). “2019 Profile of the American Iron and Steel Institute,” retrieved from steel.org/industry-data/reports, p. 9)

[6] According to the Economic Research Institute, German barbers make an annual salary of €33,991, or $37,419 USD (salaryexpert.com/salary/job/barber/germany). Barber salaries in the Philippines at ₱169,976 ($3,341 USD) per year. (salaryexpert.com/salary/job/barber/philippines).

[7] World KLEMS database, worldklems.org/data/basic/usa_wk_apr_2013.xlsx, tab DATA, rows 1043-1080, accessed February 1, 2020.

A popular argument has emerged (examples of which you can read here, here, here, here, and here) that America’s economy is so large and doing so well that we don’t need to change course — that our wealth and growth rates prove we’ve already found the right formula. On the surface, America looks like an unrivaled success story. Yet beneath the headline numbers lies an economy increasingly dependent on foreign production, hollowed out in critical industries, and unable to provide secure, high-wage work for many of its citizens.

This argument is a major obstacle to the US adopting the kind of comprehensive industrial policy that it actually needs, the kind we sketched out in our book Industrial Policy for the United States: Winning the Competition for Good Jobs and High Value Industries. So it’s worth analyzing why it isn’t true – if the relevant statistics are interpreted correctly.

The problem isn’t that America lacks prosperity — it’s that our prosperity has become detached from production. A truly strong economy doesn’t just measure how much we consume or how high our GDP ranks; it measures how much we make, how much we innovate, and how much value we create at home. When the foundation of our productive capacity erodes, so does our ability to sustain growth, defend our nation, and deliver opportunity to our people.

That’s why we need a national industrial strategy — one that values the strategic importance of industries located on U.S. soil. The following data and trends, properly understood, show not that America is “too successful to need reform,” but that our current success is masking deep structural weaknesses that only a coherent industrial strategy can fix.

Let’s start with the common claim that America surpasses all other significant countries – indeed, apart from a few accounting anomalies, microstates, oil states, and city states, all other countries simpliciter – in GDP per capita. We’re just richer than anyone else.

On the surface, this does appear to be true, as shown in the table below (source):

We really do seem to be a standout economy: The largest country richer than us, Switzerland, has fewer people than Greater Chicago. And our closest rival, Australia, has roughly one tenth our population and is 22 percent behind us in per-capita GDP.

Here’s how this all looks on a chart (source):

It certainly looks like impressions of utter American superiority are justified. But now let’s start correcting for statistical distortions.

For a start, the above data is all nominal figures. This means taking the economic output of a country, which is mostly denominated in its own currency, and converting it to dollars at market exchange rates so different countries can be compared. The problem is that the same economic output costs more in developed nations – the price of a cup of coffee, for example, reflects the wages of the coffee shop – so all developed nations, including the US, appear to be richer than they really are simply because the same amount of output is more expensive there.

Adjusting for these differences in prices between nations uses a well-established technique known as purchasing-power parity (PPP). On a PPP basis, US per-capita GDP is only 3.9, not 6.3, times the global average. The US also has a smaller advantage over the other leading economies because its economy is more distorted by higher prices. Here’s how a similar chart looks using PPP numbers:

Next, we need to adjust for the fact that the U.S. has longer work hours than most other developed nations. As a result, our PPP per-capita GDP per hour worked is actually lower than than Belgium, Sweden, Austria, and Holland, barely above Germany, and not much above France and Finland. Here’s the chart (source):

Of course, longer hours can hypothetically be either good or bad. One can argue that they reflect a strong work ethic. Or, conversely, workers being squeezed by heartless management. One can argue that that they’re bad for family values. Or that they’re just a personal choice. Or that they should be. But higher output caused merely by higher inputs, such as work time, does not constitute economic superiority. We’re not better bakers: we just bake more bread because we use more dough.

Another way to put it is that America isn’t as rich as it looks because while it’s strong in producing priced goods – things sold in the marketplace for money – it underproduces unpriced goods like leisure time. It’s easy to mock unpriced goods, but the fact that you personally would almost certainly pay hard cash for more free time if the price was right confirms that it is indeed something of real economic value. So, contrary to what some people will tell you, this is not a noneconomic (and therefore supposedly irrelevant) criticism of economic facts.

Finally, consider income inequality. Because whatever your political interpretation, i.e. whether you think inequality is deserved or that government should to try to reduce it, the US has greater income inequality than most other developed nations.

One measure of this is that only 53% of total American income goes to the bottom 90% of the population, which is less than over a dozen other developed nations. As a result, US per capita income for the bottom 90% of the US population is less than in Holland and is not much above several other countries. Here’s the chart:

It follows that America’s margin of economic superiority, for practical purposes for most Americans, is just not that large if it exists at all.

Don’t misunderstand here. The US is not secretly a poor country. We are indeed among the world’s richest economies. But we are not in a league of our own with performance so superior as to justify either a) policy complacency or b) disinterest in the economic policies of other nations some of whose policies are better than ours.

Other apparent indicators of America strength, such as our unemployment rate, should also be peeled back a bit. The rate (4.3% as of September 2025) usually quoted is a Bureau of Labor Statistics construct technically known as “U-3.” It is defined as the percentage of the civilian labor force that is unemployed and actively seeking work.

But this number deliberately ignores the following people:

a) “Discouraged workers,” i.e., those who have stopped looking for work because they don’t expect to find any

b) “Marginally attached workers,” i.e., those who are available to work and have looked in the past year, but are not currently looking

c) People who want to work full-time but can only find part-time work.

The statistic that includes all of the above is known as U-6, and it stands right now at 7.9%. And if one expands this metric to include also people who make only a poverty-line wage or less, the number is 24.1%.

A better statistic than the unemployment rate is the employment rate, also known as the employment-to-population ratio. This measures a society’s success in getting people productively employed. As you can see below, US performance by this metric is nothing special (source):

Another misleading statistic? Quite a few people people have argued in recent years that American manufacturing is doing just fine because total US manufacturing output is supposedly at an all-time high. Examples may be found here, here, here, here, here, here, and here.

For a start, this is actually an absurd statistic by which to measure the health of America’s manufacturing sector. U.S. output levels of everything from peanut butter to Disneyland visits are at all-time highs simply because the country’s population is at an all-time high. So even if this figure were true, it would prove absolutely nothing about good or bad economic performance.

Furthermore, the “manufacturing is at an all time high” statistic is misleading because this only holds true if one includes semiconductors and other IT products, which are a small fraction of the economy and a small fraction of the manufactured goods we consume. Take these industries out of the picture and U.S. manufacturing output has been stagnant since 2000. That’s a full 25 years without manufacturing growth – the first time this has happened since the country was founded.

Even worse, there are good reasons to doubt that the reported output growth in IT is even real. As economist Susan Houseman has noted,

The extraordinary output and productivity growth in computers and semiconductors reflects the way statistical agencies account for improvements in the products produced in this industry.

This last sentence refers to the fact that, as has been analyzed by Fed economists, standard statistics mistakenly count cost savings from offshoring as increases in domestic productivity. So we have been imagining that our own industries have been gaining in productivity when they have merely been benefiting from productivity gains abroad! These foreign gains do benefit the US under some circumstances, but they are not the same thing as higher productivity at home.

A better statistic for assessing the health of America’s manufacturing sector would be the percentage of our consumption of manufactured goods that is satisfied by domestic production. As of 2025, that figure was down to 63%, a full 13 points less than 2004.

This decline might have been acceptable if America was paying for all these imports with expanding exports of services, as some people appear to think we have done. But while service exports have indeed gone up, but they haven’t gone up nearly enough to cover the difference. We are instead running huge trade deficits that we pay for by selling off existing assets and going into debt to foreign nations.

Some Pollyannish statements about America’s economy don’t have even the excuse of misleading statistics as their origin, merely popular intuitions that don’t survive contact with hard data. For example, a surprising number of commentators appear to believe that the U.S. is doing just fine in manufacturing because it is still succeeding in advanced manufacturing and has only sacrificed primitive sectors. (You can find people saying this here, here, here, here, here, here, and here.)

In reality, this is an argument whose last plausible connection to fact was around Y2K. Since then, American high-tech manufacturing has been in clearly documented decline. The most obvious indicator of this is our huge and rising trade deficit in the official statistical category “advanced technology products” – $297 billion in 2024.

If the US were strong in these industries, we would be exporting these products on net, not importing them. That it matters what industries, or more precisely economic activities, a nation has is the root premise of industrial policy. So, this deficit is the kind of problem industrial policy exists to solve, and indeed a problem that is basically impossible to solve any other way. This is why the US needs proactive industrial policies, not just old standbys like a good business environment, moderate tax rates, reasonable regulations, etc., to succeed.

Let’s zoom in on the decline of America’s tech industries – masked by today’s spectacular successes in some fields like artificial intelligence – for a minute. Consider the condition of three major high-tech sectors in the late 1980s versus today:

1. Pharmaceuticals and medical devices

2. Aerospace

3. Computers and electronics

These industries’ combined portion of GDP has declined from 4.8% to 2.6%, and their share of employment has fallen from 2.8% to 1.3%. This is a clear sign of a nation that is retreating from advanced sectors, not growing them. Other nations, above all but not only China, have been expanding here thanks to their proactive industrial policies. At a minimum, we need defensive industrial policies, such as tariffs, to protect the industry positions we already have.

America’s global market share, a good indicator of our competitiveness, has been falling in one high-tech sector after another. For example, since the mid-1990s the U.S. has fallen from 37% of world semiconductor production to 12%. This was, of course, a major reason the CHIPS Act was passed under Biden and reluctantly kept alive under Trump.

This problem extends beyond high tech to many of the mid-tech industries that are the backbone of developed economies. For example, according to a recent McKinsey study, from 1995 to 2020 the US lost the following amounts of global market share:

1. 11 percentage points in “learning curve” manufacturing industries such as precision tools, semiconductors, and communications hardware.

2. Six points in scale-based and standardized sectors such as basic metals, automobiles, fabricated metals, petrochemicals, and specialty chemicals.

3. Four points in flexible and customizable sectors such as special-purpose machinery, marine and railroad equipment, aircraft, and military hardware.

These industry-level indicators of decline are arguably more important than the GDP figures we discussed earlier, because success in industries is the basis for a nation having a high per-capita GDP, not the other way around. The only way to be a high-income nation is to have people working in jobs with a high value-added per man-hour, and only certain industries can achieve this. This is why most Americans instinctively grasp that it’s not a winning move for our people to be flipping hamburgers rather than assembling cars.

Another reason to avoid complacency is that current per-capita GDP is a trailing indicator of the underlying soundness of a nation’s economy. Nations on the way up establish industry-specific competitiveness first, then grow rich on the strength of it, while nations on the way down have considerable economic “momentum” that can mask their underlying problems.

The lesson of history is clear: economic statistics often lag reality. In the 1959 election, the British Conservative Party ‘s unofficial slogan was “You’ve never had it so good.” With Britain’s economy growing 4.1% that year and 6.3% the next, this was true at the time, but their industrial base was already eroding. Within a generation, the country’s manufacturing power, global influence, and middle class had all declined.

America now faces a similar crossroads.

We remain a wealthy nation, but that wealth rests increasingly on fragile foundations. Our share of global industrial production is shrinking, our high-tech sectors are losing ground, and our prosperity is built on borrowing and consumption rather than on making and building. These are not signs of enduring success — they are warning lights.

A national industrial strategy is not nostalgia for the past. It is the only way to ensure that America’s economic strength, innovation, and security endure into the future. By rebuilding domestic production, restoring advanced industries, and investing in the skills and technologies that drive value creation, we can turn today’s temporary advantage into lasting prosperity. Without such a strategy, our apparent success will prove as fleeting as Britain’s — but with one, America can again become the world’s most dynamic industrial power.