Industrial policy is often designed as temporary support: subsidies, directed credit, or protection that is eventually withdrawn. The empirical literature now contains sharply divergent evidence on what remains after withdrawal. In South Korea's Heavy and Chemical Industry (HCI) Drive, subsidized firms grew faster than never-subsidized firms for 30 years after subsidies ended, and benefits persisted in targeted industries and downstream users; in pre-1978 China, temporary state construction interventions left lasting regional industrial development, but only where targeted industries matched local factor endowments; in the United States, temporary steel tariffs left persistent damage to downstream employment after removal; and simulation work on bioeconomy support finds sectors reverting to roughly no-support levels once support ends. No general account explains which mechanisms and conditions make the effects of temporary support permanent rather than transitory, and estimates of the magnitude and incidence of long-run spillovers onto untreated firms, industries, and regions remain fragmented and mutually inconsistent. This problem asks for the identification and reconciliation of those determinants and spillovers.

Background

Industrial policy denotes government intervention targeting particular sectors or firms — subsidies, directed credit, tax breaks, tariffs, or public procurement — rather than economy-wide instruments. A classic rationale is the infant-industry argument: temporary support lets an industry descend a learning curve or attract financing until it can compete unsupported, after which support is withdrawn.

Several specialist concepts recur in this literature. Learning-by-doing refers to productivity gains that accumulate as cumulative output rises, so early production — even if unprofitable — can permanently lower future costs. Financial frictions are constraints that prevent firms from borrowing against future returns, which can make temporary public credit or subsidies permanently change which firms grow. Total factor productivity (TFP) measures output not explained by measured inputs. Input–output linkages describe how one industry's output is another's input, so policy for targeted industries can spill over to untreated downstream users. Sunk or fixed entry costs are non-recoverable costs of establishing a plant, which can convert a temporary shock into a permanent change in the set of operating firms. Factor endowment structure is the mix of skilled labor, capital, and natural resources available locally; an intervention is congruent with the endowment structure when the targeted industry's technology uses the locally abundant factors intensively, i.e. follows local comparative advantage.

The empirical evidence on what happens after temporary support ends is conflicting. Juhász (2018) exploits the Napoleonic blockade of Britain (1806–1813), a natural experiment in temporary trade protection, and finds that French regions exogenously better protected built mechanized cotton-spinning capacity that persisted, with higher industrial value added per capita up to the mid-nineteenth century but not later. Lane (2025) studies Korea's Heavy and Chemical Industry (HCI) Drive (1973–1979) using its introduction and termination as a natural experiment: output, export, and plant-level TFP gains in targeted industries persisted long after 1979, downstream industries benefiting through input–output linkages gained too, while horizontal spillovers to other industries were limited. Choi and Levchenko (2025), using data on the universe of Korean firm-level subsidies, find that subsidized firms grew faster than never-subsidized firms for 30 years after subsidies ended, and build a quantitative heterogeneous-firm model — a model in which firms differ in productivity — in which learning-by-doing plus financial frictions rationalize this persistence, with welfare gains of 3–4 percent. Lin, Wang, and Zhang (2026) compare industries that received state investment during China's centrally planned period (1953–1978) with comparable untreated industry–prefecture pairs: temporary interventions had positive and lasting effects on the size and efficiency of local industries, operating through spillovers within and between industries, but deviations of the targeted industry's technology from local factor endowments significantly weakened the effects. On the other side, Criscuolo, Martin, Overman, and Van Reenen (2019), using a regression discontinuity in UK investment-subsidy eligibility rules, find employment effects concentrated in small firms with no TFP gains and no evidence of spillovers to other firms in the same local labor market. Lake and Liu (2025) show that the 2002 US steel tariffs — a temporary intervention — did not raise steel employment but depressed employment in steel-consuming local labor markets for many years after removal, with persistent exit of steel-intensive establishments attributable to sunk entry costs. In a different mode, Maes and Van Passel (2019) simulate bioeconomy support policies for the Belgian manure-transformation sector and find that operational subsidies generate fast sector growth during support but reversion toward the no-support level afterward, a result that depends on the modeling assumption that bankrupt firms lose their technological knowledge.

The survey of Juhász, Lane, and Rodrik (2024) synthesizes this landscape and concludes that long-lasting local effects of industrial policy are common in some (not all) settings, while the evidence on spillovers — positive downstream linkages, absent local spillovers, and leakages through multi-plant firms — is fragmented and inconsistent (the persistent negative effects on unprotected input users come from later work, Lake–Liu 2025). What no study or body of work provides is a general account of the conditions and mechanisms that determine whether temporary support leaves permanent or only transitory effects, together with credible estimates of the magnitude and incidence of long-run spillovers on untreated firms, industries, and regions.

Problem Statement

For temporary industrial-policy interventions — subsidies, directed credit, or protection that is eventually withdrawn — determine what makes their effects persist versus fade after withdrawal, and establish the size and incidence of their long-run spillovers onto untreated firms, industries, and regions. Concretely: (i) identify the mechanisms and conditions (such as learning-by-doing, financial frictions, sunk entry costs, input–output linkages, congruence between targeted industries' factor intensity and local endowment structure, and institutional or political-economy factors) that govern whether treated outcomes, untreated outcomes, or both are permanently shifted; (ii) provide credible identified estimates of the magnitude and incidence (who gains, who loses) of long-run spillovers to untreated units; and (iii) reconcile the conflicting empirical patterns of persistence, fade-out, and positive, absent, or negative spillovers documented across settings. The contrasting episodes described in the Background — persistent post-subsidy firm growth in Korea (Choi–Levchenko), endowment-congruent persistence in pre-1978 China (Lin–Wang–Zhang), persistent downstream losses after temporary tariffs (Lake–Liu), fading productivity and absent local spillovers for UK investment subsidies (Criscuolo et al.), and simulated reversion to no-support levels when exiting firms lose knowledge (Maes–Van Passel) — are motivating evidence for the conflict. The question concerns generalizable determinants and spillover incidence rather than determinants or spillovers confined to a single country, instrument, or parameter regime.

References

  1. Jaedo Choi and Andrei A. Levchenko (2025), 'The long-term effects of industrial policy', Journal of Monetary Economics 152, 103779. DOI: 10.1016/j.jmoneco.2025.103779. URL: https://doi.org/10.1016/j.jmoneco.2025.103779
  2. Nathan Lane (2025), 'Manufacturing Revolutions: Industrial Policy and Industrialization in South Korea', Quarterly Journal of Economics 140(3), 1683–1741. DOI: 10.1093/qje/qjaf025. URL: https://doi.org/10.1093/qje/qjaf025
  3. Justin Yifu Lin, Xin Wang, and Haochen Zhang (2026), 'Government interventions in industrial development and regional development: the role of endowment structure', The Scandinavian Journal of Economics, article sjoe.70012 (working-paper version: SSRN 4410618, 2023). DOI: 10.1111/sjoe.70012. URL: https://doi.org/10.1111/sjoe.70012
  4. Dries Maes and Steven Van Passel (2019), 'Effective bioeconomy policies for the uptake of innovative technologies under resource constraints', Biomass and Bioenergy 120, 91–106. DOI: 10.1016/j.biombioe.2018.11.008. URL: https://doi.org/10.1016/j.biombioe.2018.11.008
  5. James Lake and Ding Liu (2025), 'Local Labor Market Effects of the 2002 Bush Steel Tariffs', American Economic Journal: Economic Policy 17(4), 292–330. DOI: 10.1257/pol.20220472. URL: https://doi.org/10.1257/pol.20220472
  6. Réka Juhász (2018), 'Temporary Protection and Technology Adoption: Evidence from the Napoleonic Blockade', American Economic Review 108(11), 3339–3376. DOI: 10.1257/aer.20151730. URL: https://doi.org/10.1257/aer.20151730
  7. Chiara Criscuolo, Ralf Martin, Henry G. Overman, and John Van Reenen (2019), 'Some Causal Effects of an Industrial Policy', American Economic Review 109(1), 48–85. DOI: 10.1257/aer.20160034. URL: https://doi.org/10.1257/aer.20160034
  8. Réka Juhász, Nathan Lane, and Dani Rodrik (2024), 'The New Economics of Industrial Policy', Annual Review of Economics 16, 213–242. DOI: 10.1146/annurev-economics-081023-024638. URL: https://doi.org/10.1146/annurev-economics-081023-024638 (working-paper version: NBER WP 31538, DOI: 10.3386/w31538)