Industrial Policy Returns — CHIPS, IRA, and the China Comparison

The Scale of the Push

The 2022 Inflation Reduction Act and CHIPS Act together represent the largest US industrial-policy commitment since the Apollo program. The IRA's energy and manufacturing provisions are roughly $369 billion in tax credits and direct spending over ten years. The CHIPS Act adds $52 billion in semiconductor manufacturing subsidies. The combined commitment dwarfs any prior US industrial policy of the post-1970 era.

What the IRA Subsidizes

The IRA structures most of its incentives as tax credits with specific eligibility rules. Major categories:

• Clean energy generation: production tax credits for wind, solar, geothermal, and storage at $25-30 per MWh.
• Clean energy manufacturing: investment tax credits for facilities making batteries, solar components, wind components, and processed critical minerals.
• Electric vehicle credits: $7,500 per vehicle subject to domestic-content and assembly requirements.
• Carbon capture: $85 per ton stored permanently, $60 per ton for enhanced oil recovery.
• Clean hydrogen: $3 per kg for hydrogen produced at near-zero emissions intensity.
• Heat pumps and home efficiency: rebates and tax credits for residential electrification.

The tax-credit structure is more durable politically than direct grants — once enacted, removing tax credits requires affirmative legislation, which is harder to pass than discontinuing appropriations. The structure also distributes benefits across many congressional districts, which builds political coalitions for continued funding.

What CHIPS Subsidizes

The CHIPS Act focuses on a single sector with direct grants and loans rather than tax credits. Major awards:

• TSMC Arizona: $6.6 billion for two fabs in Phoenix.
• Intel: $8.5 billion across Ohio, Arizona, New Mexico facilities.
• Samsung: $6.4 billion for Texas fab.
• Micron: $6.1 billion for New York fab.
• GlobalFoundries: $1.5 billion for Vermont and New York expansions.

Direct grants give the federal government more leverage over specific outcomes (where the fab is built, what node it produces, when production begins) but create more risk of poor selection or political interference. The CHIPS implementation has so far been more disciplined than 1970s-era industrial policy, partly because the specific knowledge required to evaluate semiconductor manufacturing proposals is more concentrated than for other industries.

The Comparison with China's Industrial Policy

China's "Made in China 2025" program, announced in 2015, targeted ten high-priority sectors for domestic capability development. The program has been substantially successful in some sectors (electric vehicles, batteries, solar manufacturing, high-speed rail) and substantially unsuccessful in others (commercial aircraft, leading- edge semiconductors). The successful sectors have been ones where China combined market access (large domestic demand), strategic funding (subsidies and state-bank lending), and access to global technology (joint ventures and acquisitions).

The US industrial policy approach differs in several ways. The US has less government control over private firms; the subsidies must work through tax incentives and procurement rather than direct state ownership. The US has a more litigious and political environment that constrains the speed and flexibility of decision- making. And the US has not historically had the long-term policy continuity that Chinese industrial policy has demonstrated.

The Korean Developmental-State Model

Korea's industrial-policy success from the 1960s through the 1990s was built on a tighter government-business relationship than the US permits. The Ministry of Trade and Industry directed credit allocation, forced export discipline (subsidies were conditional on export performance), and concentrated capital in chaebol conglomerates that could compete at world scale. The model produced what is now the world's eleventh-largest economy.

The model also has costs: the chaebol concentration produced political-economy problems that Korea has struggled with for decades. The US would not want and could not implement the Korean model wholesale. But the elements — performance conditionality, export discipline, focused capital concentration — are partially present in the IRA design.

The Implementation Risks

Several specific failure modes have been seen in prior US industrial policy:

• Capture: the targeted firms become political constituents of the program rather than performance-disciplined participants. The 1970s steel and shipping subsidies are the canonical cases.
• Mis-selection: the program backs technologies or firms that turn out to be wrong bets. The Solyndra failure under the 2009 ARRA was the most visible example, though the broader ARRA energy program mostly succeeded.
• Cost overrun: the appropriated amounts turn out to be insufficient to achieve the targeted outcomes, and political support erodes before the targets are reached.
• Foreign subsidization: the tax credits flow disproportionately to foreign firms operating in the US, which is the source of WTO complaints from Europe and Korea against IRA.

Each of these risks is addressed in the IRA and CHIPS design at some level. None has been fully prevented; the next several years will reveal which materialize.

The Five-Year Test

2027 will be roughly the five-year mark. By that point, several empirical tests will have results:

• Whether US battery and solar manufacturing has reached competitive scale.
• Whether the US semiconductor fabs are producing at competitive quality and cost.
• Whether the IRA tax credits have produced the announced investment, or whether substantial commitments are walked back.
• Whether the political coalition for the spending holds across administration changes.
• Whether the trading-partner conflicts (WTO complaints, retaliatory subsidies) have produced durable problems.

The 2027 evaluation will substantially shape whether the next generation of US industrial policy continues the IRA model, retreats from it, or expands it. The intellectual case for industrial policy has won the academic debate; the empirical case is still being made.

Reading List

• Mariana Mazzucato, The Entrepreneurial State (2013)
• Ha-Joon Chang, Kicking Away the Ladder (2002)
• Robert Wade, Governing the Market (1990)
• Dani Rodrik on industrial policy (various)
• Brookings IRA tracker and CHIPS implementation reports