## Annex I. The Model and Data

## Source details

**Canonical URL:** [Annex I. The Model and Data](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024249-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024249-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024249-print-pdf.pdf.json)

---

### I. Introduction
- Context and motivation
  - Global fragmentation, heightened by Russia’s war in Ukraine and intensified US and Chinese industrial support, compounds Europe’s pre-existing low growth and limited fiscal space challenges.
  - Industrial policy usage in the EU has risen markedly:
    - State aid in the EU is about 1.5 percent of GDP (as of 2022), compared to around 0.5 percent of GDP a decade ago.
    - State aid reached 2.4 percent of GDP as governments mobilized spending in response to the COVID pandemic.
  - Most recent state aid measures are predominantly related to environmental protection and energy efficiency; this trend is likely to grow with the green transition.
- Key new EU-level initiatives noted
  - European Green Deal, European CHIPS Act, and the Digital Europe Program.
- Economic rationale and risks summarized
  - “Good” industrial policies narrowly targeted at correcting domestic market failures (e.g., scale externalities, knowledge spillovers) can increase production efficiency.
  - “Bad” industrial policies driven by rent-seeking misallocate resources toward low-productivity industries and generate fiscal costs.
  - Structural modeling (multi-country, Krugman-style model of Lashkaripour and Lugovskyy (2023)) indicates unilateral policies can create international spillovers (production relocation) and adverse terms-of-trade effects depending on foreign demand elasticities.
- Governance suggestion highlights
  - Coordinated, well-targeted industrial policies within the EU and with non-EU partners avoid distortion away from comparative advantages and mitigate adverse terms-of-trade effects.
  - Suggested practical governance: a singular EU decision-making body for strategic priorities, streamlined programs, common EU funds for collective actions, and harmonized regulations and approval processes.
- Paper organization
  - Stylized facts (Section II), structural model and simulations (Section III), illustrative case studies (Section IV), conclusions (Section V).

### II. Industrial Policy in the EU Single Market: The Current State of Play
- Single Market and trade patterns
  - The Single European Act (SEA, 1987) guarantees free movement of goods, services, capital, and people; complemented by State Aid regulations prohibiting assistance that distorts competition.
  - Intra-EU trade accounts for about 60 percent of European trade; exports to non-EU partners account for the remaining 40 percent of EU trade or over 15 percent of EU GDP.
  - EU27 total exports (imports) totaled €4.0 (€4.1) trillion in 2023.
  - China’s share of EU trade has roughly doubled over the past two decades.
- Evolution and scale of state aid
  - COVID-related measures increased overall state aid to 2.4 percent of GDP at peak; by 2022 state aid remained about 1.5 percent of GDP.
  - The four largest EU member states—Germany, France, Italy, and Spain—account for 70 percent of total state aid expenditure.
  - The average EU country spends over 1 percent of GDP on industrial policies; some smaller EU countries spend up to 2 percent of GDP (e.g., Hungary).
  - Instrument composition (2022, non-crisis state aid):
    - Direct subsidy grants and interest rate subsidies: 57.8 percent of total expenditure.
    - Tax incentives: 30.7 percent of total aid.
- Regulatory and institutional changes
  - Temporary Crisis and Transition Framework (TCTF) replaced TCF in March 2023 and allows state aid for the transition to a net-zero economy until end-December 2025.
  - TCTF supports accelerating roll-out of renewable energy, storage, renewable heat relevant for REPowerEU, decarbonizing industrial production processes, and accelerating investments for manufacturing strategic equipment such as batteries, solar panels, wind turbines, and heat-pumps.
  - Revision of the General Block Exemption Regulation (GBER) in 2014 reduced the need for prior notification when conditions are met:
    - State aid measures provided under GBER increased to about two-thirds of all active measures in 2022 from about 40 percent in 2014.
    - By expenditure size in 2022, notified aid declined to about 70 percent and GBER aid increased to about a quarter of total state aid expenditure.
  - Green Deal GBER amendment (endorsed March 2023) further streamlined rules for green and digital transitions.
- Governance trade-offs and coordination gaps
  - Modernized State Aid control framework increases flexibility but may reduce incentives for cross-national coordination due to lighter ex ante scrutiny.
  - IPCEI initiatives (enhanced in 2022) support cross-border innovation but require stronger coordination and transparency.
  - Policy reviews recommend strengthening coordination among member states and expediting decision making on EU strategic priorities.
- Policy focus implication
  - Ensure state aid is well-targeted at correcting market failures to avoid fiscal costs and cross-border efficiency losses; strengthen ex-ante coordination, transparency, streamlined decision-making, and consider common EU funds.

### Box 1. EU-wide Initiatives
- Objectives of the EU industrial policy strategy
  - Boost economic resilience.
  - Increase diversification.
  - Enhance energy security.
  - Preserve the competitiveness of the EU as a green investment location.
  - Make the European Union (EU) climate-neutral by 2050.
- Key initiatives
  - European Chips Act: provides funding (similar to the US Chips Act) for semiconductors.
  - Foreign Subsidies Regulation: new regime to combat distortions caused by foreign subsidies.
  - Foreign Direct Investment Regulation: screens and regulates FDI to safeguard critical infrastructure and strategic assets.
  - Export Control Regulation: monitors exports, technical assistance, transit and transfer of “dual-use” items.
  - Green Deal Industrial Plan (GDIP) (including REPowerEU funding): streamlines regulations, promotes R&D in clean technologies, and was proposed in February 2023 as part of the European Green Deal (launched 2020).
- GDIP main components
  - (i) Providing more flexibility on the use of the REPowerEU loans.
  - (ii) The Temporary Crisis and Transition Framework that relaxes state aid rules and allows matching subsidies offered by non-EU countries in clean-tech industries.
  - (iii) The Net Zero Industry Act: defines industries benefiting from streamlined permitting and financing.
  - (iv) The Critical Raw Materials Act: seeks to boost EU production and diversify sourcing of critical resources for the green transition.

### Model setup, key assumptions, and data
- Model used
  - Multi-industry, multi-country Krugman-style model of Lashkaripour and Lugovskyy (2023).
- Structural features and assumptions
  - Long-run version with “free entry”; firms can pay a one-off entry cost to transition across industries.
  - Additional fiscal instruments include import tariffs and export subsidies; tax revenue not required for production subsidies is rebated to consumers via lump-sum transfers.
  - Model omits intermediate goods, is static, and uses a stylized representation of fiscal policy.
  - Calibration draws on empirical evidence and retains key features of Lashkaripour and Lugovskyy (2023).
  - Key parameters with high uncertainty: elasticity of foreign demand and elasticity of scale economies.
- Model structure specifics
  - Model included 15 tradable goods sectors and a services sector.
  - Tradables explicitly include agriculture, mining, and a range of manufacturing industries “from wood products to electrical and optical equipment.”
  - No cost of moving production factors across sectors; taxation assumed lump sum (apart from pre-existing tariffs).
  - Free entry sets profits to zero in equilibrium.
- Data coverage and calibration
  - 43 countries (including all EU member states).
  - 56 industries from the 2014 World Input-Output Data (WIOD) aggregated into 15 traded industries plus services.
  - Bilateral applied import tariffs constructed based on UNCTAD TRAINS data (mainly from 2022).
  - Status quo for export subsidies and industrial Pigouvian subsidies assumed to be zero.

### Table A1 — Industry-level trade elasticities and external scale parameters (calibration inputs)
- Agriculture and Mining
  - Trade Elasticity: 6.2 (2.3)
  - External Scale Parameter: 0.1 (0.1)
- Food
  - Trade Elasticity: 2.3 (0.8)
  - External Scale Parameter: 0.4 (0.1)
- Textiles, Leather & Footwear
  - Trade Elasticity: 3.4 (0.4)
  - External Scale Parameter: 0.2 (0.02)
- Wood
  - Trade Elasticity: 3.9 (1.9)
  - External Scale Parameter: 0.2 (0.1)
- Paper
  - Trade Elasticity: 2.6 (1.1)
  - External Scale Parameter: 0.3 (0.1)
- Petroleum
  - Trade Elasticity: 0.6 (0.5)
  - External Scale Parameter: 1.2 (0.9)
- Chemicals
  - Trade Elasticity: 4 (0.4)
  - External Scale Parameter: 0.2 (0.02)
- Rubber & Plastic
  - Trade Elasticity: 5.2 (1.2)
  - External Scale Parameter: 0.1 (0.03)
- Minerals
  - Trade Elasticity: 5.3 (1.7)
  - External Scale Parameter: 0.2 (0.06)
- Metals
  - Trade Elasticity: 3 (0.5)
  - External Scale Parameter: 0.2 (0.03)
- Machinery
  - Trade Elasticity: 7.8 (1.3)
  - External Scale Parameter: 0.1 (0.02)
- Electrical & Optical Equipment
  - Trade Elasticity: 1.2 (0.3)
  - External Scale Parameter: 0.6 (0.1)
- Transport Equipment
  - Trade Elasticity: 2.8 (0.9)
  - External Scale Parameter: 0.1 (0.04)
- Recycling and others
  - Trade Elasticity: 6.2 (1)
  - External Scale Parameter: 0.2 (0.02)
- Services (model calibration note)
  - Trade elasticity: 11
  - External scale parameter: 0

### Key mechanisms by which industrial policy affects other countries
- Production relocation externality: subsidized expansion in one country can shrink production in trading partners in the same industry, reducing overall productivity in those partners.
- Terms of trade effects: reallocating labor toward industries with higher scale economies can lower export prices; when foreign demand elasticities are low, increased volumes may not offset price declines, harming welfare.
- Interaction and sensitivity: net welfare impacts depend on balance of efficiency gains, relocation externalities, and terms-of-trade changes; highly sensitive to 휎푘 (foreign demand elasticity) and 휇푘 (scale elasticity).

### Simulation summaries and quantitative findings
- Simulation #1 — “Bad” industrial policy (illustrative)
  - Policy: 20 percent subsidy to services production in Germany (services exhibit zero economies of scale in the calibration).
  - Germany domestic effects:
    - Employment share of the subsidized services sector increases by 15 percentage points.
    - Sectoral TFP in the subsidized services sector does not change.
    - TFP decreases in all other sectors (which exhibit economies of scale).
    - Aggregate TFP in Germany decreases — policy is production inefficient domestically.
  - Spillovers to France:
    - French services sector contracts; aggregated French TFP increases slightly due to expansion in other French sectors with economies of scale.
- Simulation #2 — “Good” industrial policy (illustrative)
  - Policy: production subsidies directed at Electrical and Optical Equipment in Germany (large external scale economies).
  - Germany domestic effects:
    - Employment share of the subsidized industry increases by around 15 percentage points.
    - TFP of the Electrical and Optical Equipment sector increases by about 5.5 percent, weighted by its share in total output.
    - All other sectors lose employment share and their weighted TFP decreases by 2.5 percent.
    - Overall weighted average TFP of the German economy rises by 3 percent.
    - Model-calibrated note: scale economies in electrical and optical equipment manufacturing need a subsidy of 52 percent of production costs to be fully exploited.
  - Spillovers to France:
    - Weighted TFP of the French economy falls by 0.4 percent due to contraction in Electrical and Optical Equipment.
- Coordination case
  - If all European countries implement the same production subsidy in Electrical and Optical Equipment simultaneously:
    - Weighted TFP in France increases by 0.8 percent.
    - Weighted TFP in Germany increases by 2.4 percent.
  - Interpretation: coordinated Pigouvian subsidies avoid production relocation externalities and raise production efficiency across countries relative to unilateral subsidies.
- Simulation #3 — Unilateral industrial policy (general)
  - One country eliminates market failures by exploiting scale effects across sectors while others remain status quo.
  - Repeated for each of the 43 countries: overall net welfare losses for most countries.
  - Losses particularly large for smaller, more open economies (in export/import shares of GDP).
- Simulation #4 — Minimal cooperation via simultaneous implementation
  - All EU countries implement subsidies simultaneously to correct market failures.
  - Simultaneous implementation minimizes production relocation externality; aggregated welfare loss is smaller than when each country acts unilaterally.
- Simulation #5 — EU integration and global cooperation
  - Treating the 27 EU countries as a single country (free relocation of labor and firms) reduces fiscal cost and welfare losses relative to unilateral action.
  - Adding simultaneous action by other large partners (e.g., UK and USA) yields further welfare benefits for EU countries.
- Multilateral implication
  - If all countries set industrial policy unilaterally but simultaneously, aggregate impact is positive in welfare terms on a GDP-weighted basis; multilateral cooperation is likely worthwhile.

### Fiscal costs, sensitivity, and welfare metrics
- Fiscal cost measure
  - Fiscal costs are production subsidies that fully correct for scale or agglomeration externalities; computed in steady state and interpreted approximately as cumulative cost of eliminating the market failure, in percent of GDP.
- Estimates reported
  - For a country implementing full corrective subsidies unilaterally while others maintain status quo: fiscal cost is 13.2 percent of GDP on average across EU countries (GDP-weighted basis).
  - If the EU is treated as a single country (full integration): fiscal cost declines to 10.9 percent of GDP.
- Welfare metric and sensitivity
  - Welfare simulations combine production efficiency changes and distributional effects from relative market price changes (terms of trade).
  - Model scale for welfare simulations: 43 countries (including all EU member states); also considered case where all EU countries act as a single country.
  - Key sensitivities: foreign demand elasticity, scale economies elasticity, and degree of openness.
  - Sensitivity analysis (parsimonious small open economy):
    - Three key parameters: Scale Externality (흁), Openness (1-휷), Elasticity of Foreign Demand (흈푇).
    - If foreign demand elasticity is high (or economy closed), full internalization of scale externalities via subsidies is welfare-improving.
    - When economy is more open and foreign demand inelastic, subsidies can depress export prices and cause welfare losses.
    - In fully closed economy, subsidies always yield welfare gains that increase with size of scale externalities.

### Limitations, caveats, and parameter uncertainty
- Model omissions and limitations
  - No intermediate goods: cannot fully capture supply-chain development, coordination failures, or onshoring dynamics.
  - No modeling of rent-seeking, moral hazard, or long-run fiscal sustainability under distortionary taxation (model uses lump-sum taxation).
  - Static framework: cannot analyze endogenous market-structure evolution or dynamic interaction between industrial and competition policies.
- Parameter uncertainty emphasized
  - Quantitative outcomes hinge on calibration of 휎푘 and 휇푘; these elasticities are empirically uncertain.
- Political economy and governance caveats
  - Model does not capture governance failures, political-economy influences, or implementation risks (rent-seeking, retaliatory trade policy).

### Case studies and empirical illustrations
- Airbus consortium
  - Timeframe and scale:
    - Airbus launched in 1970 and became profitable in the 1990s; fiscal support was critical during this period.
    - Fiscal support amounted to 0.35 percent of 1994 GDP of the main Airbus consortium countries (Klepper, 1994).
  - Employment and market position:
    - By 1987, Airbus directly employed over 30,000 workers across Western Europe, representing 30 percent of the total workforce of the three main industrial partners.
    - Airbus commanded around 40 percent market share in the wide-body jet market.
  - Exports and returns:
    - In the last two decades, EU aerospace industry exports have been about 2 percent of total EU exports.
    - By 2019, total exports of the EU aircraft manufacturing industry reached about 104 billion USD.
    - Unit cost was approximately 17 percent higher at the same scale than Boeing (Klepper 1994).
    - Airbus generated a 6 to 11 percent earnings return and positive knowledge spillovers.
  - Mechanisms: public support reduced financing costs, mitigated business-failure risk, facilitated knowledge spillovers, supply-chain expansion, and worker training.
- Germany photovoltaic (PV) industry and KfW
  - KfW provided low-cost loans starting in 2000, offering subsidized fixed-interest and long-maturity (up to 20 years) loans and was the single largest source of financing for PV adoption, providing at least 400 million euro of support every year in the early 2000s.
  - Outcomes: domestic solar installation rose steeply; Germany became a key EU solar panel exporter; Chinese imports increased from 2004 onward.
  - Limits and decline: by early 2010s, lack of domestic innovation, foreign competition, and inadequate manufacturing capacity undermined the German PV industry; many firms exited post-2008 financial crisis.

### Policy implications and recommendations (synthesized)
- Targeting and guardrails
  - Ensure state aid is well-targeted at correcting market failures (scale/agglomeration externalities, knowledge spillovers) rather than rent-seeking.
  - Use State Aid rules, data collection, and transparency as guardrails to minimize misuse and adverse spillovers.
- Coordination and institutional design
  - Minimum coordination of production subsidies across EU countries can raise production efficiency and avoid production-relocation distortions.
  - Suggested institutional measures: strengthen ex-ante coordination, streamline approval processes, consider a singular EU decision-making body for strategic priorities, common EU funds for collective actions, harmonized regulations.
- Integration and fiscal capacity
  - Greater European integration and full internal mobility of firms and labor increase gains from coordination.
  - Deeper integration may require an EU-wide central fiscal capacity or larger EU budget for centralized projects, complemented by capital market union efforts to mobilize private cross-border investment.
  - Strengthening public finances and rebuilding fiscal buffers remains important given high fiscal costs estimated for full corrective subsidies.
- Multilateral governance
  - Global cooperation on industrial subsidies through WTO framework is recommended to avoid harmful retaliation and reconcile differing approaches.

_Source: wpiea2024249-print-pdf — Annex I. The Model and Data_

### Annex I. The Model and Data _______________________________________________________________________ 26

### Annex I. The Model and Data _______________________________________________________________________ 26

### I. Introduction
- Global fragmentation, heightened by Russia’s war in Ukraine and intensified US and Chinese industrial support, compounds Europe’s pre-existing low growth and limited fiscal space challenges.
- Industrial policy usage in the EU has risen markedly:
  - State aid in the EU is about 1.5 percent of GDP (as of 2022), compared to around 0.5 percent of GDP a decade ago.
  - State aid reached 2.4 percent of GDP as governments mobilized spending in response to the COVID pandemic.
- Current state aid composition and orientation:
  - Most recent state aid measures are predominantly related to environmental protection and energy efficiency; this trend is likely to grow with the green transition.
  - New EU-level industrial policy initiatives include the European Green Deal, European CHIPS Act, and the Digital Europe Program.
- Economic rationale and risks of industrial policy:
  - “Good” industrial policies narrowly targeted at correcting domestic market failures (e.g., scale externalities, knowledge spillovers) can increase production efficiency and correct market failures.
  - “Bad” industrial policies driven by rent-seeking misallocate resources toward low-productivity industries and generate fiscal costs.
  - Structural modeling (multi-country, Krugman-style model of Lashkaripour and Lugovskyy (2023)):
    - Well-targeted unilateral policies can nonetheless produce international spillovers that shrink industries in trading partners via “production relocation” (Ossa, 2011), lowering overall productivity in affected economies.
    - Unilateral industrial policy may not be welfare-enhancing for the implementing country due to adverse terms-of-trade and export price effects; effects depend crucially on foreign demand elasticities.
- Coordination benefits:
  - Coordinated, well-targeted industrial policies within the EU and with non-EU partners avoid distortion away from comparative advantages, increase gains from specialization, and mitigate adverse terms-of-trade effects.
  - Existing EU state aid rules act as guardrails to monitor policies and incentivize targeting of market failures.
  - Greater European integration and full internal mobility of firms and labor increase gains from coordination.
  - Practical governance suggestions include a singular EU decision-making body for strategic priorities, streamlined programs, common EU funds for collective actions, and harmonized regulations and approval processes.
- Organization of the paper: stylized facts (Section II), structural model and simulations (Section III), illustrative case studies (Section IV), conclusions (Section V).

### II. Industrial Policy in the EU Single Market: The Current State of Play
- Single Market context and trade patterns:
  - The Single European Act (SEA, 1987) guarantees free movement of goods, services, capital, and people across member states and is complemented by State Aid regulations prohibiting assistance that distorts competition.
  - Intra-EU trade accounts for about 60 percent of European trade; exports to non-EU partners account for the remaining 40 percent of EU trade or over 15 percent of EU GDP.
  - EU27 total exports (imports) totaled €4.0 (€4.1) trillion in 2023.
  - China’s share of EU trade has roughly doubled over the past two decades.
- Evolution and scale of state aid:
  - State aid composition changed markedly during crises:
    - COVID-related measures increased overall state aid to 2.4 percent of GDP at peak.
    - By 2022, state aid remained about 1.5 percent of GDP.
  - Distributional concentration:
    - The four largest EU member states—Germany, France, Italy, and Spain—account for 70 percent of total state aid expenditure.
    - The average EU country spends over 1 percent of GDP on industrial policies; some smaller EU countries spend up to 2 percent of GDP (e.g., Hungary).
  - Instruments and fiscal implications:
    - Direct subsidy grants and interest rate subsidies accounted for 57.8 percent of total expenditure in 2022 for non-crisis state aid.
    - Tax incentives accounted for 30.7 percent of total aid in 2022.
- Regulatory and institutional changes enabling policy flexibility:
  - Temporary Crisis Framework (TCF) and its successor:
    - The Temporary Crisis and Transition Framework (TCTF) replaced the TCF in March 2023 and allows state aid for the transition to a net-zero economy in line with the Green Deal Industrial Plan.
    - The TCTF allows member states until end-December 2025 to grant aid for the transition to a net-zero economy, including accelerating roll-out of renewable energy, storage, renewable heat relevant for REPowerEU, decarbonizing industrial production processes, and accelerating investments for manufacturing strategic equipment such as batteries, solar panels, wind turbines, and heat-pumps.
    - Extension of TCTF until end-December 2025 is expected to support state aid for environmental protection.
  - Block exemption changes:
    - Revision and modernization of the General Block Exemption Regulation (GBER) in 2014 reduced the need for prior notification when conditions are met.
    - State aid measures provided under GBER increased to about two-thirds of all active measures in 2022 from about 40 percent in 2014.
    - By expenditure size in 2022, notified aid declined to about 70 percent and GBER aid increased to about a quarter of total state aid expenditure.
    - The Green Deal GBER amendment endorsed in March 2023 further streamlined state aid rules for green and digital transitions and may increase spending classified under GBER and IPCEI projects.
- Governance trade-offs and coordination gaps:
  - State Aid control framework has improved assessment of market failures and proportionality but modernization has introduced lighter ex ante scrutiny of aid schemes, increasing flexibility while potentially reducing incentives for cross-national coordination.
  - IPCEI initiatives, enhanced in 2022, can support cross-border innovation and infrastructure projects, but further steps are needed to strengthen coordination and transparency.
  - Recent policy reviews (Draghi and Letta reports) recommend strengthening coordination among member states and expediting decision making on EU strategic priorities.
- Implication for policy focus:
  - Need to ensure state aid is well-targeted at correcting market failures to avoid fiscal costs and cross-border efficiency losses.
  - Strengthened ex-ante coordination, transparency, streamlined decision-making, and possible use of common EU funds can help realize collective gains from industrial policy while minimizing adverse spillovers.

*Source: wpiea2024249-print-pdf - Annex I. The Model and Data _______________________________________________________________________ 26*

### Box 1. EU-wide Initiatives

### Box 1. EU-wide Initiatives

### Objectives of the EU industrial policy strategy
- Boost economic resilience.
- Increase diversification.
- Enhance energy security.
- Preserve the competitiveness of the EU as a green investment location.
- Make the European Union (EU) climate-neutral by 2050.

### Key EU initiatives
- European Chips Act
  - Provides funding (similar to the US Chips Act) to bolster the EU’s competitiveness and resilience in semiconductor technologies and applications.
- Foreign Subsidies Regulation
  - Creates a new regime aimed at combating distortions of competition on the EU internal market caused by foreign subsidies.
- Foreign Direct Investment Regulation
  - Screens and regulates FDI into EU member states to safeguard critical infrastructure, technologies, and strategic assets of the EU countries from potential risks posed by foreign investors.
- Export Control Regulation
  - Monitors exports, deployment of technical assistance, transit and transfer of “dual-use” items.
- Green Deal Industrial Plan (GDIP) (including through REPowerEU funding)
  - Prioritizes streamlining of regulations and the promotion of R&D in clean technologies, to facilitate rapid funding, enhance skills, and strengthen supply chains.
  - Proposed in February 2023 as part of the European Green Deal, launched by the European Commission in 2020.

### Green Deal Industrial Plan (GDIP) — main components
- (i) Providing more flexibility on the use of the REPowerEU loans.
- (ii) The Temporary Crisis and Transition Framework that relaxes state aid rules and allows matching subsidies offered by non-EU countries in clean-tech industries.
- (iii) The Net Zero Industry Act, which defines industries that benefit from streamlined permitting and financing.
- (iv) The Critical Raw Materials Act, which seeks to boost EU production and diversify sourcing of critical resources for the green transition.

*Source: Box 1. EU-wide Initiatives, wpiea2024249-print-pdf*

### conclusions on the spillovers of “good” domestic industrial policies on production efficiency in other countries.

### Conclusions on the spillovers of “good” domestic industrial policies on production efficiency in other countries.

### Model setup and key assumptions
- Simulations are based on the long-run version of the model, where there is “free entry” and firms can transition across industries after paying a one-off entry cost.
- Additional fiscal instruments in the model include trade taxes, in the form of import tariffs and export subsidies. Any tax revenue not required for production subsidies is rebated to consumers via lump-sum transfers. There are no other distortionary fiscal instruments.
- The model omits intermediate goods (so it cannot fully characterize supply chains), is static, and uses a very stylized representation of fiscal policy.
- Calibration draws on empirical evidence (see Table A1, Annex 1) and retains key features of Lashkaripour and Lugovskyy (2023) in sensitivity analysis for a small economy.
- Key model parameters with high uncertainty: the elasticity of foreign demand and the elasticity of scale economies.

### Channels and mechanisms through which industrial policy affects other countries
- Production relocation externality: Increasing exports in a subsidized industry can shrink other countries’ production in that industry, reducing overall productivity in trading partner countries (as per Ossa (2011)).
- Terms of trade effects: Changing patterns of production and trade affect trade prices and thus countries’ terms of trade. Reallocating labor toward industries with higher scale economies can change export prices; if foreign demand elasticities are lower in subsidized industries, higher export volumes may not compensate for lower export prices, potentially deteriorating a country’s terms of trade.
- Interaction between channels: The net welfare impact depends on the balance of efficiency gains, production relocation externalities, and terms of trade changes; this balance is sensitive to the elasticities mentioned above and the degree of openness.

### Simulation #1 — “Bad” industrial policy (illustrative)
- Policy: 20 percent subsidy to services production in Germany (services exhibit zero economies of scale in the calibration).
- Domestic effects in Germany:
  - Employment share of the subsidized services sector increases by 15 percentage points.
  - Sectoral TFP in the subsidized services sector does not change as the sector expands (no economies of scale).
  - TFP decreases in all other sectors (which exhibit economies of scale).
  - Aggregate TFP in Germany decreases — the policy is unambiguously production inefficient domestically.
- Spillovers to France (illustrative trading partner):
  - French services sector contracts (mirror image of Germany’s response), despite services being relatively hard to trade under the calibration.
  - Contraction of French services is matched by expansion of other French sectors with economies of scale, yielding a small aggregate TFP increase in France.
- Interpretation: Poorly designed unilateral subsidies can cause large domestic production inefficiencies while producing small production efficiency gains in other countries.

### Simulation #2 — “Good” industrial policy (illustrative)
- Policy: Production subsidies directed at a manufacturing sector (Electrical and Optical Equipment) with large external scale economies, implemented unilaterally in Germany.
- Domestic effects in Germany:
  - Employment share of the subsidized industry increases by around 15 percentage points.
  - TFP of the Electrical and Optical Equipment sector increases by about 5.5 percent, weighted by its share in total output.
  - All other sectors lose employment share and their weighted TFP decreases by 2.5 percent.
  - Overall weighted average TFP of the German economy rises by 3 percent.
  - Note: Scale economies in electrical and optical equipment manufacturing need a large subsidy to be fully exploited, of 52 percent of production costs, according to the model’s evidence-based calibration.
- Spillovers to France:
  - Production relocation externality contracts French production in Electrical and Optical Equipment; weighted-TFP in that sector falls significantly.
  - Net effect: weighted TFP of the French economy falls by 0.4 percent.
- Interpretation: A unilateral “good” industrial policy can increase production efficiency at home but reduce productivity in trading partners by distorting international production patterns away from underlying comparative advantages.

### Coordination and comparative outcomes
- Coordinated policy across Europe:
  - If all European countries implement production subsidies in the Electrical and Optical Equipment sector simultaneously, employment and TFP of this sector in France increase (unlike the unilateral case).
  - Net effects under coordination:
    - Weighted TFP in France increases by 0.8 percent.
    - Weighted TFP in Germany increases by 2.4 percent.
  - Interpretation: Simultaneous implementation of production subsidies that correct market failures avoids distorting international production patterns, allowing gains from specialization across countries and raising production efficiency in all countries relative to unilateral subsidies.
- Policy implication: Minimum coordination of production subsidies can raise production efficiency across countries; in presence of external economies, sectoral productivities and comparative advantages are endogenous to policies, so coordinated policies are needed to fully exploit comparative advantages.

### Welfare, terms of trade, and sensitivity
- Welfare metric: Simulations combine production efficiency changes with distributional effects from changes in relative market prices across countries (terms of trade).
- Model scale for welfare simulations: 43 countries (including all EU member states); also considered case where all EU countries act as a single country.
- Key sensitivity: Overall welfare effects depend critically on foreign demand elasticity, scale economies elasticity, and the degree of openness of the economy.
- Important tension: There is a tension between using industrial policy to enhance production efficiency and using it as a trade policy tool; this arises when foreign demand is downward sloping (foreign demand not perfectly elastic), which generates terms of trade effects and production relocation externalities even for small countries.

### Limitations and caveats emphasized by the authors
- Omitted features that could affect conclusions:
  - No intermediate goods: cannot fully capture supply-chain development, coordination failures, or onshoring of critical inputs.
  - No modeling of rent-seeking behavior or moral hazard if industrial and trade policies shield firms from competition.
  - Static model: cannot analyze endogenous evolution of market structure, dynamic interaction between industrial and competition policies, or fiscal sustainability of large-scale industrial policy over time.
- Parameter uncertainty: Elasticities driving the three main channels are highly uncertain; quantitative results hinge on specific modeling assumptions and calibrations.

*Source: IMF Working Paper — conclusions on the spillovers of “good” domestic industrial policies on production efficiency in other countries.*

### Annex 1. Specifically, the model included 15 tradable goods sectors and a services sector. Among the

### wpiea2024249-print-pdf - Annex 1. Specifically, the model included 15 tradable goods sectors and a services sector. Among the

### Model structure
- Model included 15 tradable goods sectors and a services sector.
- Tradables sector composition explicitly includes agriculture, mining, and a range of manufacturing industries of differing levels of technological sophistication, "from wood products to electrical and optical equipment."
- Calibration captures an empirical relationship across industries between the extent of scale economies and demand elasticities.
- Model assumptions noted:
  - No cost of moving production factors across sectors.
  - Taxation is assumed to be lump sum in the model (apart from pre-existing tariffs).
  - The model is static; fiscal costs are computed in steady state and interpreted approximately as the cumulative cost of eliminating the market failure.

### Simulation #3 — Unilateral industrial policy
- Scenario: One country unilaterally implements industry-specific production subsidies that eliminate market failures by exploiting scale effects in all sectors, while other countries maintain status quo policies.
- Key mechanisms and effects:
  - Domestic production efficiency improves by expanding the share of the most productive sectors at the expense of others.
  - A production relocation externality distorts international patterns of production away from underlying comparative advantages and affects trade prices.
  - Domestic welfare benefit from improved production efficiency is partially offset by a drop in export prices because subsidies are largest in industries where demand elasticities are smaller.
  - Piecewise repetition of this simulation for each of the 43 countries in the sample, while all other countries maintain the status quo in each case, reveals overall net welfare losses for most countries.
- Distributional/focus observations:
  - Losses are particularly large for smaller economies that are more open to trade (in terms of exports and imports as a share of GDP), including most member states of the EU.
  - Domestic-optimal targeting can differ from the narrow correction of a single sector: "correcting external scale economies only in the Electrical and Optical Equipment is a 'good' policy" by the paper's definition, but not the best domestic policy, which would require correcting external economies in all other domestic sectors as well.

### Simulation #4 — Minimal cooperation via simultaneous implementation
- Scenario: All EU countries implement industrial subsidies simultaneously to correct market failures by internalizing scale effects.
- Key mechanisms and effects:
  - Simultaneous implementation minimizes the production relocation externality and avoids concentration of production in a single unilateral subsidizing country.
  - While subsidies may reduce export prices in industries with smaller foreign demand elasticities, trading partners’ own subsidies to industries with larger scale externalities may lower import prices and improve the first country’s terms of trade.
  - Aggregated welfare impact across all EU countries acting simultaneously indicates a smaller average welfare loss than in the case where each country acts entirely unilaterally (see bar in Figure 7 labeled ‘’EU countries acting simultaneously’).

### Simulation #5 — EU integration and global cooperation
- Scenarios considered:
  - Re-calibration treating the 27 countries of the EU as a single country, with labor and firms allowed to relocate freely across the EU (industrial policy set at the regional level).
  - Further extension where the EU as a whole, plus the UK and USA, all implement industrial subsidies simultaneously.
- Key mechanisms and effects:
  - Full factor mobility and regional integration maximize efficiency gains from exploiting scale effects and minimize losses from depressing export prices because the EU as a whole is less open to trade than many individual EU countries.
  - Aggregated outcomes: welfare outcome for "EU as an integrated region" shows smaller average welfare loss relative to unilateral action; adding simultaneous action by other large partners (e.g., UK and USA) produces further welfare benefits for EU countries.

### Multilateral implications and general findings
- Multilateral cooperation on design and implementation of industrial policy is likely to be worthwhile, even at the global level.
- If all countries set industrial policy unilaterally but simultaneously, this avoids some production relocation externalities and reductions in terms of trade; in that case the aggregate impact of industrial policy is positive in welfare terms on a GDP-weighted basis.
- Large emerging market economies that are less open to trade can achieve welfare gains from unilateral industrial policy because the cost of expanding production in industries with lower demand elasticities is less important and only partially offsets benefits of exploiting scale effects.
- Political economy and implementation caveats noted: the model does not capture governance failures, political-economy factors, or fiscal-sustainability impacts of industrial policy beyond the static subsidy cost calculations.

### Fiscal costs and fiscal risks
- Fiscal cost measure: fiscal costs are limited to production subsidies that fully correct for scale or agglomeration externalities; computed in the model’s steady state and interpreted approximately as the cumulative cost of eliminating the market failure, in percent of GDP.
- Estimates:
  - For a country implementing these subsidies unilaterally while others maintain status quo, fiscal cost is 13.2 percent of GDP on average across EU countries, on a GDP-weighted basis.
  - If the EU is treated as a single country (full integration), this fiscal cost declines to 10.9 percent of GDP.
- Implications:
  - These estimates raise concerns about fiscal sustainability, especially when fiscal space is limited in many EU countries.
  - The model does not account for potential distortionary effects of taxation needed to fund subsidies, since taxation is lump sum in the model.
  - Static modeling approach limits ability to gauge impacts on fiscal sustainability over time.

### Case-study introduction (context for broader market failures and trade-offs)
- The paper complements structural-model findings with case studies (Airbus and the German solar industry) to illustrate:
  - Broader market failures not captured in the structural model, including spillovers from innovation and technology diffusion, and informational asymmetries.
  - Implementation risks such as rent-seeking, high fiscal costs, potential for retaliatory trade policy, and the need for state and firms to adapt to changing economic and business conditions.
- Example mechanisms from Airbus (briefly):
  - Cross-border coordination among France, the UK, and West Germany in 1970 to enter commercial aerospace; Airbus transitioned from an intergovernmental consortium to a joint-stock venture in 2001.
  - Fiscal support forms included equity participation, launch aid, and guarantees against losses from exchange rate fluctuations.
  - From 1970 to 1994, the estimated size of accumulated subsidies ranged from 0.2 to

*Source: IMF Working Papers — Industrial Policy in Europe: A Single Market Perspective (Annex 1).*

### 0.35 percent of 1994 GDP of the main Airbus consortium countries (Klepper, 1994). This support was critical

### wpiea2024249-print-pdf - 0.35 percent of 1994 GDP of the main Airbus consortium countries (Klepper, 1994). This support was critical

### Airbus consortium: trajectory, impact, and quantitative outcomes
- Timeframe and scale
  - Airbus launched in 1970 and became profitable in the 1990s, a 20-year period during which fiscal support was critical.
  - Fiscal support amounted to 0.35 percent of 1994 GDP of the main Airbus consortium countries (Klepper, 1994).
  - Airbus secured orders from US airlines from 1978.
- Employment and market position
  - By 1987, Airbus directly employed over 30,000 workers across Western Europe.
  - This represented 30 percent of the total workforce of the three main industrial partners.
  - Airbus commanded around 40 percent of market share in the wide-body jet market.
- Exports and industry scale
  - In the last two decades, EU aerospace industry exports have been about 2 percent of total EU exports.
  - By 2019, total exports of the EU aircraft manufacturing industry reached about 104 billion USD.
- Efficiency, costs, and returns
  - The unit cost for Airbus was approximately 17 percent higher at the same scale of production than that of Boeing (Klepper 1994).
  - Airbus generated a 6 to 11 percent earnings return and created positive knowledge spillovers to other non-consortium member firms (IMF 2024c).
  - Airbus boosted its R&D expense in 2020 from 4.8 to 5.7 percent of total sales revenue and has maintained it at over 5 percent over the past few years.
- Mechanisms of impact
  - Public support reduced financing costs and mitigated high business-failure risk and delayed profitability during the firm’s early decades.
  - Support facilitated knowledge spillovers, continuous improvement in production techniques, product performance, expansion of European supply chains, and worker training.
  - Advanced technologies in new aircraft reduced fuel consumption and maintenance costs, improving competitiveness during the late 1970s global oil crisis era.

### Germany’s photovoltaic (PV) industry and KfW’s role
- Pre-2000 context and KfW intervention
  - Germany had provided strong and longstanding R&D support for PV technology since the 1970s, but policy and demand uncertainties constrained financing and investment.
  - Starting in 2000, KfW provided low-cost loans to mobilize private investment into PV manufacturing and to enable adaptation to government demand incentives (notably a feed-in-tariff scheme).
  - KfW entered contracts with financial intermediaries who bore credit risk, offering subsidized fixed-interest and long-maturity (up to 20 years) loans through intermediaries that could provide up to 100 percent of a generator’s financing needs.
  - In the early 2000s, KfW was the single largest source of financing for PV adoption, providing at least 400 million euro of support every year.
- Outcomes and trade patterns
  - Domestic solar panel installation rose steeply and Germany emerged as a key player in solar panel exports in the EU.
  - Chinese imports increased starting in 2004 to meet rising demand for solar panels.
  - Chinese factories purchased equipment from Germany, sent technicians for service and training, assembled panels domestically, and exported them to Europe as early as the mid-2000s (Zhu, He and Gu, 2021).
- Limits, risks, and eventual decline
  - By the early 2010s, lack of domestic innovation, foreign competition, and inadequate manufacturing capacity undermined the German PV industry.
  - German PV manufacturers struggled to keep pace with technological improvements and scaling by Chinese competitors who achieved lower unit costs.
  - From 2008 onward, the global financial crisis reduced access to capital and caused negative profit margins, driving many German PV manufacturers out of the market.
  - The German case illustrates that single-country industrial policy can overcome market failures (e.g., financial constraints) in the short run but may fail to sustain competitiveness without cross-border coordination, scale, and continuous innovation.

### Model-based analysis and broader conclusions
- General findings on industrial policy effectiveness
  - Industrial policies motivated by well-identified market failures (not rent-seeking or arbitrary “picking winners”) can be welfare-enhancing.
  - Well-designed policies can exploit untapped scale externalities at the industry level (“Marshallian externalities”), foster international competition, and solve coordination problems in creating domestic value chains.
- Limits of unilateral industrial policy
  - Unilateral industrial policy can generate welfare losses domestically and negative productivity effects in other countries by not internalizing production externalities in foreign sectors.
  - Production-efficiency enhancing industrial policies can deteriorate a country’s terms of trade because of lower demand elasticities in industries with higher scale economies.
  - For relatively small, open economies (like European countries), terms-of-trade deterioration can translate into welfare losses absent other integration (e.g., labor market integration).
  - Unilateral actions can lead to sub-optimal sectoral investment and inefficient production relocation away from underlying comparative advantages.
- Benefits of coordination and EU-level approaches
  - Policy coordination helps achieve efficient production levels across countries and avoid inefficient relocation effects.
  - Coordinated policies can attenuate adverse terms-of-trade effects (both export and import prices may fall), creating self-reinforcing incentives for efficient industrial policies.
  - State-aid rules, data collection, and transparency serve as useful guardrails to minimize policy misuse and adverse spillovers.
  - Ongoing EU efforts (e.g., Green Deal Industrial Plan, Important Projects of Common European Interests (IPCEI)) can be steps forward if they target market failures and enable true cross-border coordination for R&D and first industrial deployment to achieve scale and exploit comparative advantages.
  - A singular EU decision-making body could facilitate more coherent coordination amid overlapping programs and instruments.
  - Recent moves to increase flexibility in state aid approvals for critical areas should be balanced with preserving information sharing and transparency.
- Implications for deeper integration and fiscal capacity
  - A larger and more integrated single market could amplify gains from well-designed industrial policies by enabling consolidation and growth in scale in compliance with EU competition rules.
  - Closer integration would facilitate relocation of firms and workers across member states, harmonization of taxes and subsidies, and development of infrastructure networks, energy grids, and interconnectivity.
  - Deeper EU integration may require an EU-wide central fiscal capacity or a more ambitious EU budget for centralized projects of mutual interest, complemented by efforts to mobilize private cross-border investment via capital market union enhancements.
  - Strengthening public finances and rebuilding fiscal buffers, in line with EU rules, remains important to increase resilience against future shocks.
- Global multilateral governance
  - Global cooperation on industrial subsidies through the WTO framework is the preferred route to avoid harmful retaliatory policies and to reconcile differing industrial policy approaches across countries.

*Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024249-print-pdf.pdf*

### Annex I. The Model and Data

### Annex I. The Model and Data

### Model overview
- Model used: multi-industry, multi-country Krugman-style model of Lashkaripour and Lugovskyy (2023).
- Key features:
  - Industries differ by scale economies and elasticity of trade volumes to prices.
  - Countries 푖∈퐶̅ with population 퐿푖; labor is the sole factor of production and is supplied inelastically (one unit per individual).
  - Workers are perfectly mobile across industries within countries but immobile internationally; wages are country-specific 푤푖.
  - Representative consumer in country 푖 chooses industry-level product bundles for each k∈퐾̅; each industry bundle has price index 푃̃푖,푘 and aggregates country-specific and firm-level varieties via a nested CES aggregator.
  - Parameters:
    - 휎푘: determines elasticity of substitution between country-specific varieties; determines degree of price-elasticity of foreign demand in industry 푘.
    - 훾푘 (>1): determines firm-level market power and “love-of-variety”; firm-level mark-up equals 1/(훾푘−1).
  - Firms:
    - Mass of monopolistically competitive firms M푖,푘=Ω푖,푘 in country i, industry k; entry cost 푤푖푓푘푒.
    - Productivity draws z(ω)≥1 from distribution 퐺푖,푘(z).
    - Marginal cost for delivering to destination j is τ푖푗,푘 푤푖 z(ω)/, where τ푖푗,푘 is iceberg transport cost.
  - Scale elasticity:
    - Elasticity of the Producer Price Index (PPI) of composite good in industry k produced in country i with respect to number of firms is −휇푘 (<0); absolute value 휇푘 is industry-level scale elasticity.
    - 휇푘 equals the elasticity by which variety-adjusted total factor productivity increases with industry-level employment and is exactly equal to the constant firm-level mark-up 1/(훾푘−1) in this Krugman-style model.
  - Free entry assumption: profits driven to zero (restricted entry would raise average profit margins where sales are high in high mark-up industries).

### Policy instruments and efficiency benchmarks
- Policy instrument in simulations:
  - Industry-specific production subsidy applied to industry k’s output produced in country i, irrespective of destination of sales.
  - Subsidy financed via lump sum taxes to consumers.
- Global efficiency:
  - Social planner maximizes global welfare via good-specific taxes and lump-sum international transfers.
  - Taxes restore marginal cost pricing; lump-sum transfers allow international redistribution.
  - Globally efficient policy: zero trade taxes and Pigouvian industrial production subsidies that restore marginal cost pricing.
- Market-determined equilibrium:
  - Absent a planner, allocation is not globally efficient due to cross-country heterogeneity in markups (restricted entry) or scale elasticities (free entry).
- Relationship to other policy instruments:
  - Model also considers import tariffs and export subsidies in its broader literature context (references cited).

### Welfare implications of industrial policy (key qualitative mechanisms)
- Main quantitative determinants: scale elasticity 휇푘 and price-elasticity of foreign demand 휎푘.
- Empirical correlation:
  - Lashkaripour and Lugovskyy (2023) find 휎푘 and 휇푘 are negatively correlated empirically.
- Mechanisms of unilateral production subsidies:
  - Positive: correct misallocation by expanding output in high returns-to-scale (high 휇푘) industries.
  - Negative: worsen terms of trade by expanding exports in industries with low price-elasticity of foreign demand (low 휎푘), potentially offsetting allocative efficiency gains — an “immiserizing” welfare effect.
- Quantitative outcome:
  - Welfare gains or losses from unilateral Pigouvian industrial subsidies depend on calibration of 휎푘 and 휇푘 and other parameters.

### Table A1 — Industry-level trade elasticities and external scale parameters
Econometric estimation results used for model calibration (standard errors in parentheses)

- Agriculture and Mining
  - Trade Elasticity: foreign demand elasticity minus one (export prices to output): 6.2 (2.3)
  - External Scale Parameter: 0.1 (0.1)
- Food
  - Trade Elasticity: 2.3 (0.8)
  - External Scale Parameter: 0.4 (0.1)
- Textiles, Leather & Footwear
  - Trade Elasticity: 3.4 (0.4)
  - External Scale Parameter: 0.2 (0.02)
- Wood
  - Trade Elasticity: 3.9 (1.9)
  - External Scale Parameter: 0.2 (0.1)
- Paper
  - Trade Elasticity: 2.6 (1.1)
  - External Scale Parameter: 0.3 (0.1)
- Petroleum
  - Trade Elasticity: 0.6 (0.5)
  - External Scale Parameter: 1.2 (0.9)
- Chemicals
  - Trade Elasticity: 4 (0.4)
  - External Scale Parameter: 0.2 (0.02)
- Rubber & Plastic
  - Trade Elasticity: 5.2 (1.2)
  - External Scale Parameter: 0.1 (0.03)
- Minerals
  - Trade Elasticity: 5.3 (1.7)
  - External Scale Parameter: 0.2 (0.06)
- Metals
  - Trade Elasticity: 3 (0.5)
  - External Scale Parameter: 0.2 (0.03)
- Machinery
  - Trade Elasticity: 7.8 (1.3)
  - External Scale Parameter: 0.1 (0.02)
- Elec. & Optical Equip.
  - Trade Elasticity: 1.2 (0.3)
  - External Scale Parameter: 0.6 (0.1)
- Transport Equip.
  - Trade Elasticity: 2.8 (0.9)
  - External Scale Parameter: 0.1 (0.04)
- Recycling and others
  - Trade Elasticity: 6.2 (1)
  - External Scale Parameter: 0.2 (0.02)
- Services (model calibration note)
  - Trade elasticity: 11
  - External scale parameter: 0

- Source of estimates: Lashkaripour and Lugovskyy (2023), based on Colombian trade data at level of Harmonized System 10-digit product category, 2007-13.

### Data used for calibration and simulations
- Coverage:
  - 43 countries (including all EU member states).
  - 56 industries from the 2014 World Input-Output Data (WIOD) (Timmer and others, 2015).
- Aggregation:
  - Following Costinot and Rodriguez-Clare (2014), the 56 WIOD industries are aggregated into 15 traded industries (for which Lashkaripour and Lugovskyy (2023) have estimated 휎푘 and 휇푘) plus a services sector (assumed 휇푘=0 and 휎푘=11).
- Tariffs and subsidies:
  - Bilateral applied import tariffs (status quo) constructed following Kucheryavyy, Lyn and Rodriguez-Clare (2023), based on UNCTAD TRAINS data, using latest data available (mainly from 2022).
  - Status quo for export subsidies and industrial Pigouvian subsidies assumed to be zero (following Lashkaripour and Lugovskyy (2023)).

### Annex II — Sensitivity analysis (parsimonious model insights)
- Simplified small open economy model:
  - Domestic tradable good (consumption and exports), domestic non-tradable good, and foreign tradable good (imported).
  - Economy is small (takes import price as given) but domestically-produced tradable good is differentiated so export prices respond to production.
- Three key parameters determining welfare impact of production subsidies:
  - Scale Externality (흁): impact of subsidies on productivity in tradable sector.
  - Openness (1-휷): share of domestically-produced tradable good consumed domestically; smaller (1-휷) implies more open economy.
  - Elasticity of Foreign Demand (흈푇): elasticity of foreign demand; higher elasticity means foreign market absorbs increased production with smaller price decline.
- Main sensitivity conclusions:
  - If foreign demand elasticity is high (or economy closed), production subsidies that fully internalize scale externalities are welfare-improving.
  - When the economy is more open and foreign demand is inelastic, subsidies can depress export prices, harming terms of trade and potentially causing overall welfare losses.
  - For a given scale externality, low trade elasticity can generate welfare losses that worsen with greater openness.
  - In a fully closed economy, production relocation externality and foreign demand elasticity are irrelevant; subsidies always yield welfare gains that increase with the size of scale externalities.
- Overall implication:
  - Industrial policy via production subsidies corrects scale or agglomeration externalities and is always welfare-improving in closed economies or when foreign demand is perfectly elastic; otherwise, a trade-off arises between allocative efficiency gains and adverse terms-of-trade effects, which can make industrial policy welfare-reducing.

*Source: Annex I and Annex II, "The Model and Data" (from wpiea2024249-print-pdf).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024249-print-pdf.pdf_