Trade Policy at a Crossroads: Protectionism and Industrial Strategy

Something has quietly broken in the economics profession's self-confidence, and the timing matters: right now, the United States is running the largest peacetime tariff experiment in a century, with both parties cheering it on. We've done our research on this one, and the facts are genuinely surprising. Here's what's settled: a major working paper found that recent US tariffs cost consumers and firms roughly fifty-one billion dollars, with a net loss of seven-point-two billion even after counting every job gained. And here's what's contested: whether any of that pain is worth paying for durable industrial strength, or whether it simply transfers money from households to politically connected producers. The disagreement runs deep — not left versus right, but between economists who trust dispersed market signals and those who think the state can credibly pick strategic winners. So let me put the load-bearing question to this panel: is the current protectionist turn a genuine structural realignment, or are we watching a political moment being mistaken for an economic strategy?

Let me cut through the noise: the U.S. debate is treating "industrial policy" as if it's a single lever you pull—tariffs up, subsidies out, jobs appear. But South Korea's battery supply chain didn't decouple from China by slapping tariffs on cathodes. It did so through coordinated R&D consortia, overseas mining co-investments, and performance-linked subsidies that sunset if benchmarks aren't met. That model reduced dependency from 92% to 58% in five years without a single tariff. Why isn't this the case study we're debating? The real question isn't whether to intervene—it's whether the U.S. can build the institutional muscle to enforce the discipline that made those interventions work elsewhere. The CHIPS Act and IRA have no sunset clauses tied to global competitiveness. That's not industrial policy; it's corporate welfare dressed in nationalist drag. The developmental states we admire—South Korea, Taiwan—didn't just pick winners; they fired losers. Can the U.S. political system do that? The evidence so far says no.

The enforcement gap is the load-bearing variable. Any large-scale subsidy creates concentrated interests whose rational move is to convert temporary support into permanent rents once the initial investment is sunk. The historical record shows that developmental successes hinged on credible exit rules enforced before capture set in; absent those, the same political architecture that produces the program also prevents its discipline. We still have no public measurement of what these outlays would need to return in sustained productivity to justify the fiscal exposure, which leaves the incentive structure tilted toward persistence rather than performance.

The enforcement problem isn’t just political will; it’s instrument design. Washington defaults to border taxes and unconditional fiscal transfers, but the cases that actually shifted supply chains used conditional co-investment. Vietnam didn’t subsidize smartphone assembly blindly; it tied land leases and tax holidays to local procurement thresholds, while Samsung co-funded the training of one hundred twenty thousand engineers. That’s a contract, not a grant. Meanwhile, RCEP’s cumulation rules—where components sourced across fifteen members count as regional origin—create an architectural bypass for domestic-content mandates, even at a thirty-two percent utilization rate. Market-oriented and heterodox analysts agree the mechanism matters; they split on whether any permeable political system can actually enforce withdrawal. The question is whether Washington can write performance triggers that survive the first lobbying cycle, or if we’re mistaking patronage for strategy.

The real variable hiding beneath this rhetoric is Rodrik’s trilemma—hyper‑globalization, national sovereignty, and democratic politics can’t all advance together. The tariff turn is sovereignty trying to return to democracy’s hands, which is legitimate. But once that power comes home, the burden of proof shifts inward: can a lobbying democracy make those choices productively? Market‑oriented economists are right to warn that without credible sunset rules, freedom to protect morphs into permission to transfer. Heterodox economists counter that measured industrial strategy can restore social cohesion lost to globalization. The unresolved question is practical, not philosophical: can the U.S. quantify benefits and enforce exits fast enough that re‑sovereignized policy space becomes productivity, not patronage?

Let’s talk about the invisible architecture that’s already rewriting the rules. RCEP’s cumulation rules mean a Vietnamese smartphone using Korean chips, Thai screens, and Malaysian packaging qualifies for preferential access to Japan—regardless of U.S. “made-in-America” thresholds. That’s not a loophole; it’s a structural design choice covering 30% of global GDP. The CHIPS Act’s domestic-content requirements are being architecturally circumvented in real time, yet this isn’t even a footnote in the U.S. debate. The question isn’t whether to intervene—it’s whether the U.S. can design instruments that outmaneuver the trade blocs already gaming the system. South Korea and Vietnam didn’t win by brute-force protection; they won by embedding performance triggers in land leases, tax holidays, and vocational training. Those are contracts, not handouts. Until the U.S. asks why its industrial policy defaults to unconditional transfers instead of enforceable co-investment, it’s playing chess while everyone else plays Go.

The deeper constraint is dispersed knowledge, not just political permeability. Conditional contracts still require the state to specify verifiable benchmarks for productivity or supply-chain shifts years in advance, yet the research shows we lack even basic longitudinal measures of whether current U.S. subsidies alter long-run innovation trajectories versus one-time announcements. That information gap turns performance triggers into negotiated compliance rituals once investments are sunk, regardless of initial design.

The ladder-kicking critique from Global South scholars isn’t moral posturing; it’s a structural observation about positional advantage. When a state is climbing, performance discipline holds because the survival bargain is explicit. Mature economies retrofitting that discipline onto entrenched incumbents face a fundamentally different relational contract. The IMF’s fifty-year dataset across one hundred fifty countries shows a standard tariff hike consistently drags output down by roughly four-tenths of a percent after five years, which tells us blunt protection actively fights the stated goal. That leaves subsidy conditionality as the only viable lever. Whether permeable institutions can actually enforce withdrawal without triggering capital flight or capture remains genuinely contested. Market and heterodox camps split on that institutional capacity. We’re running the experiment live, and the longitudinal productivity data simply isn’t there yet.

The blind spot is fiscal accountability. We know exactly what tariffs cost consumers down to the unit, but we have no public ledger for what each industrial dollar buys in long-run productivity or innovation. That’s not a data quirk; it’s an institutional asymmetry. U.S. politics can finance visible factories but not measure their output beyond press releases, because performance audits come years after the political credit is spent. Until productivity impact per subsidy dollar becomes a required disclosure—just as import tariffs are tracked in GDP terms—industrial policy will remain fiscally opaque, and populist protection will keep outcompeting performance-based design simply because it’s faster to claim.

Here is the question no one has named directly: which part of the current U.S. system performs the function that Korea's Economic Planning Board performed? That agency had statutory authority to deny credit, revoke licenses, and terminate protection — and it used all three. It was deliberately insulated from the legislature. The developmental states we keep citing as proof that industrial policy works ran their critical industrialization phases under political conditions that would be unconstitutional in the United States. That is not an argument against industrial policy. It is an argument that the precondition for success is institutional, not economic — and neither party in this debate is asking how you build a state capable of disciplining the industries it is trying to develop.

The IMF's fifty-year dataset isn't just a warning—it's a structural constraint. A 3.6 percentage point tariff hike consistently drags output down by 0.4% after five years. That's not a policy choice; it's a macroeconomic gravity well. The U.S. is betting this time is different, but the burden of proof lies entirely on identifying the countervailing mechanism that breaks that pattern. Is it the scale of subsidies? The geopolitical urgency? Or are we simply mistaking political theater for economic transformation? The real experiment isn't whether protectionism works—it's whether the U.S. can design instruments that don't replicate the same old misallocation. South Africa's REIPPPP program didn't just require local content; it embedded equity and community ownership into the contract. That's not charity—it's a structural hedge against capture. The question isn't whether the U.S. can copy Korea's technocracy; it's whether it can build a different kind of discipline—one that ties industrial policy to social transformation, not just corporate welfare.

The information asymmetry between subsidy allocation and outcome verification creates its own capture dynamic. Without mandated longitudinal disclosure of productivity or innovation returns per dollar, the political cycle rewards visible commitments far faster than any measurable displacement of imports or capability gains. This isn't solved by rewriting contract language; it follows from accountability horizons where dispersed costs surface after concentrated gains are already locked in. The institutional question therefore starts with data architecture, not enforcement mechanisms.

The missing mechanism isn’t just technocratic insulation; it’s relational reciprocity. In the East Asian cases the panel keeps citing, state support wasn’t a transactional grant. It operated as a governed hierarchy where access to capital, land, or export licenses carried explicit performance obligations. Miss the target, and you lost bureaucratic patronage and market position. That bargain held because firms accepted subordination to a national development sequence. The United States is trying to run that same playbook through arm’s-length contracts in a system where regulated firms hold structural leverage over the agencies funding them. I think that’s why the current subsidy debate feels unmoored: we’re asking for developmental outcomes without the relational architecture that made discipline credible. Whether a permeable democracy can build a functional equivalent through transparent productivity tracking or equity-sharing structures remains genuinely contested. But treating industrial support as a one-way transfer virtually guarantees the capture everyone is warning about.

The missing lever isn’t technical capacity, it’s credible threat. Every historical case that sustained performance discipline—whether in Korea, Japan, or South Africa’s renewables program—paired fiscal support with enforceable downside risk. In those systems, losing state backing meant losing profitability, not just prestige. U.S. programs invert that: firms get paid up front through tax credits, and penalties arrive—if ever—after sunk capital and electoral turnover. That time inversion is what weakens deterrence. The measurable experiment now is whether tying future subsidy tranches to independently verified innovation or export metrics can restore that risk symmetry without violating democratic transparency.

The panel has converged on enforcement as the central problem, but there's a prior question being skipped: enforcement of what, exactly? The CHIPS Act and IRA are structured around investment announcements — factory groundbreakings, job commitments, capacity targets. Those are observable. What they are not benchmarked against is global competitiveness — whether the subsidized industry can eventually win in open markets without continued support. That distinction matters because, per the research brief, we currently have no publicly available methodology for measuring whether these programs shift long-run productivity at all, as opposed to producing one-time capital deployment. A sunset clause tied to the wrong metric is not discipline — it is a different kind of capture.

The real enforcement gap isn’t in the contracts—it’s in the exit. Every developmental state that worked had a built-in obsolescence clause: protection ended when the industry could compete globally. The CHIPS Act and IRA have no such mechanism. The subsidies are structured as open-ended tax credits, not time-bound investments with competitive benchmarks. That’s not an oversight; it’s a structural choice. The U.S. is trying to industrialize without the discipline that made industrialization work historically. The question isn’t whether the state can pick winners—it’s whether it can admit losers. Until that exit function is institutionalized, the "infant industry" framing is just a euphemism for permanent corporate welfare.

The structural problem runs deeper than missing metrics or exit clauses. In systems where electoral cycles reward visible spending but punish visible failures years later, any performance rule that survives political turnover must deliver short-term political cover for the agencies enforcing it. That requirement rarely aligns with long-horizon industrial upgrading. The result is programs whose formal conditions erode not from lobbying alone but from the mismatch between political time horizons and the observable lags in capability formation.

We are treating discipline as a contract law problem, when the historical record shows it operates as a relational governance problem. In the East Asian developmental sequences, state support functioned less like a grant and more like a governed hierarchy where access to capital carried explicit performance obligations. Miss the target, and you lost bureaucratic patronage. That bargain held because firms accepted subordination to a national development sequence. Confucian political economy frames this as cultivating internal capability and social cohesion rather than relying on external leverage. The United States is trying to replicate those outcomes through arm’s-length tax credits in a system where regulated firms hold structural leverage over the agencies funding them. I think that is why the exit clause debate feels circular. You cannot legislate developmental discipline into a political architecture that treats state support as a one-way transfer. The real question is whether American institutions can build a functional equivalent to that relational bargain, or whether we are simply subsidizing incumbents and calling it strategy.

The institutional problem is circular: Congress designs programs it cannot later unwind. The fiscal lever meant to fund performance — the tax credit — is also the one instrument Congress can’t easily retract without voting for a tax increase. That makes even a well‑written sunset clause politically inert. If the state wants credible discipline, it needs a different trigger — one where the withdrawal of benefits is automatic once the recipient’s cost advantage crosses a threshold, much like the graduation rules once used for World Bank concessional loans. The constraint is not economic calculation but political liability management: who bears the blame when withdrawal is enforced.

Here is the load-bearing assumption the panel keeps skipping past: the developmental states that actually enforced exit were not just institutionally different from the United States — they were operating under a specific historical condition that no longer exists. Korea and Taiwan industrialized during the Cold War, when Washington actively tolerated their protectionist arrangements because geopolitical alignment mattered more than market access. The discipline those states could impose on firms was partly underwritten by American strategic patience. The U.S. is now attempting to run a developmental state in a world where it is simultaneously the hegemon being challenged and the democracy being lobbied — and those two pressures compound each other in ways the East Asian analogy simply does not capture.

The developmental states we admire didn’t just enforce exit—they enforced entry into global competition as the condition for continued support. Korea’s HCI plan didn’t just say “you have ten years”; it said “you must capture X percent of global markets by year five or lose access to state-backed credit.” That’s not a sunset clause—it’s a competitiveness trigger. The U.S. programs measure investment, not market performance, because the political system can’t stomach the downside of admitting failure. But without that trigger, the subsidies become a one-way bet: heads the firm wins, tails the taxpayer loses. The real institutional gap isn’t enforcement capacity—it’s whether the state can credibly commit to withdrawing support from firms that can’t compete globally. That’s what made the East Asian model work, and it’s what the current U.S. approach structurally avoids.

The deeper constraint is information. Without any public longitudinal ledger tracking productivity shifts or import displacement against subsidy outlays—the exact gap the evidence map flags as incomplete—even a well-designed competitiveness trigger stays unverifiable. Tax credits compound this by moving support off-budget, so the state never accumulates the comparative data that developmental banking models used to adjust terms. The architecture itself blocks the feedback loop required for adjustment, regardless of political will.

The unspoken asymmetry here is fiscal. When developmental states disciplined firms, they had closed capital accounts and state-directed credit, so failure meant an immediate funding cutoff. The U.S. has a global capital market standing ready to refinance any firm that loses subsidy access. That external liquidity neuters domestic discipline because capital can simply bypass the state. Industrial strategy in an open financial system therefore needs a different enforcement vector—one that operates through export credit terms, procurement preferences, or defense contracting power, not just budget appropriation. Without controlling the financing channel, “competitiveness triggers” turn symbolic the minute markets reopen their wallets.

We are treating industrial policy as a financial transfer problem, when the mechanism that actually shifts production structures is coordination risk. The push to reduce cathode dependency in Korea relied on joint intellectual property pools and co-invested overseas mining to absorb the early-stage matching risk that private firms will not touch alone. Market-oriented critics are right to flag the IMF’s long-run data showing tariff hikes shrink output, and we still lack longitudinal proof that today’s subsidy waves boost productivity. Whether targeted coordination actually beats open markets over a twenty-year horizon is genuinely unclear. But conflating blunt protectionism with network-building misses the actual lever. The state is not supposed to pick winners. It is supposed to solve the matching bottleneck that keeps capable suppliers from forming. If we keep measuring success by subsidy recapture rates instead of whether the underlying industrial ecosystem becomes self-sustaining, we will keep funding concrete while starving the connective tissue.

The panel keeps asking who enforces the exit. The prior question is what the exit is supposed to achieve. If the goal is global competitiveness, the benchmark has to be market share in open competition — which means the U.S. state needs to be able to declare a subsidized firm a failure while that firm is still employing thousands of people in a swing-state congressional district. No institutional design solves that political problem through better contract drafting. The Korean state could enforce competitiveness triggers partly because the political cost of industrial failure was lower than the political cost of strategic vulnerability. That calculus is inverted in the current U.S. context — which means the enforcement problem is not technical. It is a question of what the political system treats as the greater threat.

The ladder-kicking critique isn’t just about hypocrisy—it’s about the temporal asymmetry of power. When Hamilton and List argued for infant-industry protection, they assumed the state would enforce graduation because the alternative was strategic vulnerability. Today’s U.S. subsidies lack that existential urgency. The firms receiving them operate in a world where capital flight, not state withdrawal, is the credible threat. That reverses the power dynamic: the state depends on the firm’s continued investment more than the firm depends on the state’s continued support. The enforcement problem isn’t technical—it’s relational. Without a shared narrative of mutual obligation, the subsidies become a one-way transfer, and the "competitiveness trigger" becomes a ritual rather than a real constraint. The question isn’t whether the U.S. can replicate East Asia’s discipline; it’s whether it can generate a political culture where industrial policy is treated as a partnership, not a handout. That’s a cultural shift, not an institutional tweak.

The research brief flags the absence of longitudinal productivity data as a core gap, which means any competitiveness trigger operates without an independent arbiter. That absence lets competing definitions of success proliferate, each backed by one-time investment announcements rather than sustained cost-structure shifts. In an open financial system, the state therefore cannot close the information loop that developmental models used to adjust terms before political coalitions reframe the benchmark.

The structural hole in this debate is who actually bears the transition risk. Tariffs socialize it through higher consumer prices; open markets leave it on firms; developmental states pooled it inside government–industry compacts. What’s missing in the current model is a fiscal instrument that prices that risk transparently. Right now the cost is hidden in higher prices and opaque tax credits, so voters experience pain without seeing the return. A public ledger matching subsidy dollars to measured productivity or wage gains would let the politics of industrial policy mirror its economics. Without that visibility, the program keeps regressive optics even if long‑run gains exist.

We are treating discipline as a clause you write into a contract, when the operational record shows it functions as a structure you build into the payout architecture. India’s production-linked incentive scheme ties disbursements to independently verified output milestones, not groundbreakings. ASEAN’s cumulation framework rewards firms that integrate across regional suppliers rather than isolating behind domestic-content walls. Neither model guarantees long-run upgrading—market-oriented critics are right that capture and misallocation are persistent risks, and the longitudinal data on productivity shifts is still missing. But both relocate enforcement from political declaration to structural attrition. When funding is tranched against third-party audits and market access depends on cross-border coordination, underperformers simply miss the next threshold. The state never has to announce failure. The mechanism just stops clearing the transaction. That is how relational governance handles exit without triggering the legislative liability we keep circling.

The ladder-kicking critique has now completed a full rotation, and that rotation is the structural point worth naming. Chang's argument was originally about industrialized nations advocating universals that served their particular position. What the non-Western scholarship adds — and what I'd treat as analysis rather than settled fact — is that the structural logic applies regardless of who occupies the dominant position. The normative case for any trade regime tends to track the geopolitical interests of whoever is making it. That's not a point about hypocrisy. It's a point about what trade principles actually are: they are positional arguments dressed as universals. Which means evaluating any country's stated trade doctrine requires asking first — dominant in what, and at what stage?

The ladder-kicking critique isn’t just about who climbed first—it’s about who gets to define the rungs. When Hamilton and List wrote, the state could enforce graduation because the alternative was strategic irrelevance. Today’s subsidies operate in a world where firms can relocate capital faster than governments can adjust terms. That speed mismatch isn’t a bug; it’s the core structural difference between 19th-century protectionism and 21st-century industrial policy. The enforcement problem isn’t about writing better contracts—it’s about whether the state can credibly threaten to withhold support when the firm can credibly threaten to leave. That’s not a question of political will; it’s a question of relative mobility. And in a world where capital moves at the speed of a wire transfer, the state’s leverage is inherently weaker than it was when Hamilton could threaten to revoke a factory’s credit line. The real institutional gap isn’t enforcement capacity—it’s whether the state can make its support conditional on something the firm can’t easily replicate elsewhere. That’s why the Korean model worked: the state controlled the financing channel, so withdrawal meant real pain. In an open financial system, the only way to discipline firms is to make support conditional on outcomes the market can’t provide—like joint IP pools or co-invested supply chains. Otherwise, the subsidies become a transfer, not a partnership.

The sharpest tension this panel surfaced wasn't about tariffs versus free trade. It was about whether the United States has the institutional architecture to make any industrial strategy mean what it claims to mean. The concrete takeaway: the right question to ask about any subsidy program is not how large it is, but what happens when a recipient fails — and who bears the political cost of saying so out loud. On the original question, managed trade can in principle deliver durable industrial strength, but the historical cases that worked required institutional conditions the U.S. has not yet built and may not be able to build through legislation alone. Thank you for listening. As it happened; as it is.