## A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change

## Source details

**Canonical URL:** [A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change](https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026003-source-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2026/english/wpiea2026003-source-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2026/english/wpiea2026003-source-pdf.pdf.json)

---

### Key findings and quantitative impacts
- The 2022 energy price shock is estimated to reduce euro area potential GDP by 0.8 percent by 2027 under the baseline price scenario.
- Energy efficiency is projected to rise by approximately three percent by 2027 relative to the no-shock counterfactual.
- Without the observed improvement in energy efficiency, the decline in potential output would have been two-thirds larger (implying a potential output loss around 1.3 percent by 2027).
- Alternative scenario outcomes:
  - Baseline (actual and futures-based 2022–27 path): potential output loss of 0.8 percent by 2027.
  - Severe shock (price expectations during winter 2023): loss in 2027 is 0.5 percentage points larger than baseline (implying 1.3 percent).
  - Extreme shock (futures pricing at the peak of the 2022 crisis): drop of 2.2 percent by 2027.
- Cross-country heterogeneity (baseline scenario, drop in potential output by 2027):
  - Italy: around 1.2 percent.
  - Germany: around 0.9 percent.
  - Spain: around 0.6 percent.
  - France: around 0.4 percent.
- Price dynamics and persistence:
  - European natural gas prices briefly surged to more than ten times their early 2021 levels in summer 2022.
  - By mid-2025, gas prices remained twice their historical average.
  - Futures pricing as of mid-2025 implies a composite index of European fossil fuel prices will return to its pre-pandemic average by 2027 after a multi-year deviation to the upside.
- Elasticities of substitution and their dynamics:
  - Short-run elasticity of substitution between energy and other inputs for European countries estimated at 0.04-0.06.
  - The elasticity of substitution increases to 0.3 in the medium term.
- Sensitivity to energy-efficiency responsiveness:
  - If responsiveness of energy efficiency to price shocks had been 80 percent lower, the euro area’s potential output loss would have been around 1.3 percent by 2027 (about two-thirds larger than observed).

### Model intuition, mechanisms, and calibration
- Framework:
  - An endogenous directed technical change model where firms reallocate investment from non-energy productivity to energy productivity when energy prices rise.
- Mechanism:
  - Higher energy prices prompt a shift in investment toward energy efficiency, raising energy productivity but reducing capital-labor productivity and generating transition frictions.
  - In the short- to medium-term, energy efficiency gains mitigate but do not fully offset the adverse price effect on output; potential output falls relative to a no-shock counterfactual.
  - Over the long run, the economy returns to its balanced growth path, but the temporary deviation implies a permanent output level loss.
- Production technology (CES-type aggregation used in the model):
  - yt = [ (1−γ)(At ktα lt1−α)ε−1/ε + γ(Ae,t et)ε−1/ε ]ε/(ε−1)
  - At denotes capital-labor productivity; Ae,t denotes energy productivity; α is the labor share; γ is the share parameter; ε denotes the elasticity of substitution between the capital-labor bundle and energy.
- R&D allocation and technology constraint:
  - At+1/At ≡ f(nt)
  - Ae,t+1/Ae,t ≡ fe(1−nt)
  - Technology constraint G[f(nt), fe(1−nt)] = 0 imposes a trade-off between improving capital-labor productivity and energy productivity.
- Calibration highlights:
  - β (discount factor) = 0.985.
  - δ (depreciation rate) = 0.05.
  - σ (risk aversion) = 1.
  - Parameters α and γ calibrated from EU KLEMS and BP data (country-specific).
- Bayesian estimation key results:
  - Elasticity of substitution ε: posterior mean ~0.061; posterior sd ~0.046.
  - R&D efficiency and other parameters (posterior means for EA): B ≈ 0.012 (sd 0.002), Be ≈ 0.170 (sd 0.024), φ ≈ 0.904 (sd 0.014), ρq ≈ 0.196 (sd 0.098).
- Empirical reduced-form estimates:
  - Short-term elasticity ε ≈ 0.04 (significant).
  - Medium-term elasticity ε ≈ 0.30 (significant).
  - Technical bias coefficient τe − τk,l ≈ 0.02 (positive and significant), indicating technical change favors energy-saving technologies.

### Simulation scenarios and comparative dynamics
- Simulation setup:
  - Energy price paths over 2021-2027 relative to the 2016/19 average and deflated by the euro area GDP deflator.
  - Three scenarios: "baseline" (large) shock, "severe" shock, and "extreme" shock corresponding to different vintages of actual and expected energy price developments.
- Main simulation insights:
  - Energy price hikes spur firms to increase investment in energy-saving technologies; model-predicted Ae comoves closely with energy prices when other shocks are shut down.
  - Energy efficiency gains partially buffer losses to potential output, but abrupt large spikes cause larger medium-term potential output losses due to transition costs in reallocating R&D and investment toward energy-saving technologies.
  - A sudden, large energy price spike induces larger deviations from the balanced growth path than a gradual price increase; the impact on potential output is more than four times smaller with a gradual price increase that reaches the same endpoint by 2027.
  - Movements in the energy price basket below 10 percent that last only a year or two have negligible effects on euro area potential output.
  - The baseline versus severe comparison implies that sustained higher prices (about 30 percent higher each year over five years) produce materially larger output losses.
- Level versus growth effects:
  - Potential growth effects are large in the first years after the shock but are essentially zero by the end of the period (the shock ultimately has a level, not a growth, effect on potential output).
  - Despite a persistent increase in energy prices, the economy returns to the balanced growth path, so the shock produces a level offset rather than a permanent growth difference.

### Stylized facts and empirical context
- Historical and recent price patterns:
  - GDP and energy consumption grew at a similar speed prior to the oil price shocks of the 1970s; energy consumption flat-lined thereafter while GDP continued to grow.
  - Following the Russian invasion of Ukraine, real wholesale natural gas prices spiked: the average natural gas price (Dutch TTF) in 2022 was about six times the 2016-19 average and the euro area fossil fuel price basket was more than three times the pre-shock average.
  - Natural gas prices in 2025 are still around 90 percent above pre-shock levels; the fossil fuel price index in 2025 is around 30 percent above pre-shock levels.
- Observed reductions in energy use and interpretation:
  - Broad-based reduction in natural gas consumption across all sectors and months in the EU in 2022–2023.
  - Energy productivity (gas per unit of output) increased shortly after gas prices started rising and stabilized after gas prices dropped back significantly.
  - Aggregate energy intensity fell in 2022 and further in 2023, with declines for natural gas and other fossil fuels; renewables’ energy intensity was broadly stable.
  - Taken at face value, the improvement in energy efficiency by 2023 was as much as 10 percent relative to the pre-shock trend (with the caveat that compositional effects likely imply the "true" improvement was noticeably less than 10 percent).
- Firm-level and survey evidence:
  - A 2022 German IFO survey showed many firms adjusted production processes without reducing output in response to high prices.
  - The 2024 European Investment Bank investment survey reports that 50 percent of European firms are investing in energy efficiency; on average, energy-efficiency investment accounts for 12 percent of investment by EU firms.

### Caveats, confounders, and measurement nuances
- Confounding factors:
  - Fuel switching and sectoral reallocation can confound estimates of genuine within-sector energy efficiency gains; compositional effects may overstate aggregate energy-efficiency improvements.
  - Behavioral changes (e.g., households reducing thermostat settings) and lower demand due to reduced output can contribute to observed reductions in energy consumption and are difficult to fully disentangle from efficiency gains in aggregate data.
- Substitution between fossil fuels and renewables:
  - Literature documents high elasticities of substitution between fossil fuels and renewables of around 2.
  - Empirical data for Europe show energy intensity dropped for both fossil fuels and other sources of energy (which includes renewables).
  - Prior work finds incorporating renewables does not qualitatively change model implications, though energy efficiency investments are directed towards renewables.
- Structural transformation and sectoral composition:
  - Aggregate model does not capture sectoral or firm-level dynamics; compositional effects could matter.
  - Shift-share decomposition (IEA data) shows historically within-sector energy efficiency gains have been the main driver in all countries (in no country does reallocation account for more than fifty percent).
  - For a broad group of euro area countries in 2022:
    - Compositional effects reduced energy consumption in industry and services by 5.9 percent.
    - Lower energy intensity lowered consumption by 4.4 percent.
    - Higher output increased consumption by 3.3 percent.
  - For a subgroup that accounts for around 40 percent of the broader group’s energy consumption and for which 2023 data are available, the historical developments are very similar (correlation of 0.92). 2023 data for this subgroup suggest energy savings came exclusively from lower energy intensity, with compositional effects slightly increasing consumption.
- Measurement cautions:
  - Aggregate estimates of up to a 10 percent improvement in energy efficiency by 2023 should be interpreted cautiously because compositional effects imply the true within-sector or within-firm improvement likely was noticeably less than 10 percent.

### Policy implications and recommendations
- Avoid policies that suppress energy price movements through subsidies, as these can dampen incentives for innovation and directed technical change.
- Prioritize policies that:
  - Reduce underlying energy price volatility.
  - Support long-term investment in energy-saving technologies.
- Address the productivity tradeoff (energy efficiency gains come at the cost of non-energy productivity) with complementary reforms:
  - National-level structural reforms.
  - Deeper integration of the EU single market.
  - Coordinated public investment at the EU level to revitalize long-term productivity growth.
- Additional policy-relevant takeaways:
  - Energy price signals matter; persistently suppressing them can hinder improvements in energy efficiency, which acts as a buffer against future adverse energy price shocks.
  - Abrupt price spikes are harmful; policies that reduce wholesale price volatility (for example, further integration of the European energy system) are desirable.
  - Policy should aim to lower price volatility by removing economic distortions (e.g., artificial fragmentation of European energy markets across national lines) rather than introducing new distortions (e.g., price subsidies).
  - The productivity tradeoff highlights the need to focus on capital-labor productivity in Europe; ambitious policy action is needed, including national structural reforms, deepening the EU single market, and joint provision of key public goods at the EU level.
- Broader implication:
  - Productivity growth should improve from the low of 2022-23 as the energy shock’s impact fades, and the temporary nature of the energy price drag on productivity means energy prices are unlikely to be the core reason behind Europe’s ongoing, longer-lasting productivity malaise.

*Source: IMF Working Paper "A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change" — Executive Summary and selected sections.*

### Executive Summary ......................................................................................................

### Executive Summary

### Key findings and quantitative impacts
- The 2022 energy price shock is estimated to reduce euro area potential GDP by 0.8 percent by 2027 under the baseline price scenario.
- Energy efficiency is projected to rise by approximately three percent by 2027 relative to the no-shock counterfactual.
- Without the observed improvement in energy efficiency, the decline in potential output would have been two-thirds larger (implying a potential output loss around 1.3 percent by 2027).
- Alternative scenario outcomes:
  - Baseline (actual and futures-based 2022–27 path): potential output loss of 0.8 percent by 2027.
  - Severe shock (price expectations during winter 2023): loss in 2027 is 0.5 percentage points larger than baseline (implying 1.3 percent).
  - Extreme shock (futures pricing at the peak of the 2022 crisis): drop of 2.2 percent by 2027.
- Cross-country heterogeneity (baseline scenario, drop in potential output by 2027):
  - Italy: around 1.2 percent.
  - Germany: around 0.9 percent.
  - Spain: around 0.6 percent.
  - France: around 0.4 percent.
- Price dynamics and persistence:
  - European natural gas prices briefly surged to more than ten times their early 2021 levels in summer 2022.
  - By mid-2025, gas prices remained twice their historical average.
  - Futures pricing as of mid-2025 implies a composite index of European fossil fuel prices will return to its pre-pandemic average by 2027 after a multi-year deviation to the upside.
- Elasticities of substitution and their dynamics:
  - Short-run elasticity of substitution between energy and other inputs for European countries estimated at 0.04-0.06.
  - The elasticity of substitution increases to 0.3 in the medium term.
- Sensitivity to energy-efficiency responsiveness:
  - If responsiveness of energy efficiency to price shocks had been 80 percent lower, the euro area’s potential output loss would have been around 1.3 percent by 2027 (about two-thirds larger than observed).

### Model intuition and mechanisms
- Framework: an endogenous directed technical change model where firms reallocate investment from non-energy productivity to energy productivity when energy prices rise.
- Mechanism:
  - Higher energy prices prompt a shift in investment toward energy efficiency, raising energy productivity but reducing capital-labor productivity and generating transition frictions.
  - In the short- to medium-term, energy efficiency gains mitigate but do not fully offset the adverse price effect on output; potential output falls relative to a no-shock counterfactual.
  - Over the long run, the economy returns to its balanced growth path, but the temporary deviation implies a permanent output level loss.
- Reconciling substitution estimates:
  - Very low short-run substitution (elasticity 0.04-0.06) explains large immediate impacts of price spikes.
  - Endogenous investment increases the elasticity over the medium term (to 0.3), limiting longer-run output losses.

### Scenario comparisons and volatility effects
- A gradual, exogenous increase in energy prices results in a much smaller potential-output loss than a sharp exogenous price spike like that observed in 2022.
- Movements in the energy price basket below 10 percent that last only a year or two have negligible effects on euro area potential output.
- The baseline versus severe comparison implies that sustained higher prices (about 30 percent higher each year over five years) produce materially larger output losses.

### Policy implications and recommendations
- Avoid policies that suppress energy price movements through subsidies, as these can dampen incentives for innovation and directed technical change.
- Prioritize policies that:
  - Reduce underlying energy price volatility.
  - Support long-term investment in energy-saving technologies.
- Address the productivity tradeoff (energy efficiency gains come at the cost of non-energy productivity) with complementary reforms:
  - National-level structural reforms.
  - Deeper integration of the EU single market.
  - Coordinated public investment at the EU level to revitalize long-term productivity growth.

*Source: IMF Working Paper — Executive Summary.*

### 0.6 and 0.4 percent, respectively, due to a different energy mix (and hence price shock) and different estimated

### A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change

### Key findings on energy price shocks and potential output
- A sudden, large energy price spike induces larger deviations from the balanced growth path than a gradual price increase; the impact on potential output is more than four times smaller with a gradual price increase that reaches the same endpoint by 2027.
- Short-term versus long-run divergence:
  - In the medium term, lower volatility of energy prices and absence of sharp price spikes reduce output losses because abrupt switching of investment from capital-labor saving to energy-saving technologies entails transition costs.
  - In the long run, the level of potential output converges between abrupt and gradual price-shock scenarios if the steady-state energy prices are the same.

### Stylized facts and empirical evidence on energy efficiency responses
- Historical and recent patterns:
  - GDP and energy consumption grew at a similar speed prior to the oil price shocks of the 1970s; energy consumption flat-lined thereafter while GDP continued to grow.
  - Following the Russian invasion of Ukraine, real wholesale natural gas prices spiked: the average natural gas price (Dutch TTF) in 2022 was about six times the 2016-19 average and the euro area fossil fuel price basket was more than three times the pre-shock average.
  - Natural gas prices in 2025 are still around 90 percent above pre-shock levels; the fossil fuel price index in 2025 is around 30 percent above pre-shock levels.
- Observed reductions in energy use and their interpretation:
  - Broad-based reduction in natural gas consumption across all sectors and months in the EU in 2022–2023.
  - Energy productivity (gas per unit of output) increased shortly after gas prices started rising and stabilized after gas prices dropped back significantly.
  - Aggregate energy intensity fell in 2022 and further in 2023, with declines for natural gas and other fossil fuels; renewables’ energy intensity was broadly stable.
  - Taken at face value, the improvement in energy efficiency by 2023 was as much as 10 percent relative to the pre-shock trend (with the caveat that compositional effects likely imply the "true" improvement was noticeably less than 10 percent).
- Firm-level responses and survey evidence:
  - A 2022 German IFO survey showed many firms adjusted production processes without reducing output in response to high prices.
  - The 2024 European Investment Bank investment survey reports that 50 percent of European firms are investing in energy efficiency; on average, energy-efficiency investment accounts for 12 percent of investment by EU firms.

### Directed technical change mechanism and model structure
- Production technology and directed technical change:
  - Aggregate domestic output yt is produced using capital kt, labor lt, and energy et via a CES-type aggregation:
    yt = [ (1−γ)(At ktα lt1−α)ε−1/ε + γ(Ae,t et)ε−1/ε ]ε/(ε−1)
  - At denotes capital-labor productivity; Ae,t denotes energy productivity (energy efficiency); α is the labor share in the Cobb-Douglas composite; γ is the share parameter; ε denotes the elasticity of substitution between the capital-labor bundle and energy.
  - R&D allocation n t determines growth of At and Ae,t via:
    At+1/At ≡ f(nt)
    Ae,t+1/Ae,t ≡ fe(1−nt)
  - Technology constraint G[f(nt), fe(1−nt)] = 0 imposes a trade-off between improving capital-labor productivity and energy productivity.
- Planner’s results and dynamic intuition:
  - With an interior solution for technology allocation, Akα and Ae et grow at a common rate g.
  - The allocation of technological effort n and output growth rate g are jointly determined by f(n)σ/(1−α) = gσ.
  - The long-run share of energy in output depends on the relative cost of improving energy efficiency (i.e., foregone productivity gains in capital and labor).
  - In a decentralized setting, firms face externalities: firms do not internalize dynamic spillovers of current technology choices on future outcomes, causing potential inefficiency.

### Calibration, estimation, and empirical parameter estimates
- Data and sample:
  - Annual data for 1995-2021 for main advanced European countries (Germany, Italy, France, Spain) and the Euro Area (EA).
  - Energy consumption and fossil fuel prices from BP Statistical Review of World Energy; prices deflated by country-level GDP deflators in 2015 EUR per million Btu.
  - Output y is measured as GDP minus net exports of fossil fuels in 2015 EUR.
- Calibration highlights (Table 1):
  - β (discount factor) = 0.985.
  - δ (depreciation rate) = 0.05.
  - σ (risk aversion) = 1.
  - α and γ calibrated from EU KLEMS and BP data (country-specific).
- Estimation and key parameter estimates (Table 2 and text):
  - The model introduces three shocks: (i) zA (labor/capital productivity), (ii) zAe (energy-saving technology), (iii) zpe,t (energy price).
  - Bayesian estimation yields an elasticity of substitution ε estimated at around 0.06 for the euro area (posterior mean ~0.061; posterior sd ~0.046).
  - Parameters governing R&D efficiency (B, Be) and φ are estimated; reported posterior means for EA: B ≈ 0.012 (sd 0.002), Be ≈ 0.170 (sd 0.024), φ ≈ 0.904 (sd 0.014), ρq ≈ 0.196 (sd 0.098).
- Empirical estimates of substitution and technical bias (Table 3 and text):
  - Panel regression (reduced form) yields:
    - Short-term elasticity of substitution ε ≈ 0.04 (significant).
    - Medium-term elasticity of substitution ε ≈ 0.30 (significant).
    - Technical bias coefficient τe − τk,l ≈ 0.02 (positive and significant), indicating technical change favors energy-saving technologies (price effect dominates short-run).
  - Interpretation: energy generally behaves as a complement to capital and labor in the short run (very low substitutability), but substitutability increases in the medium run, consistent with improvements in energy efficiency.

### Model simulations and interpretation (simulated shock scenarios)
- Simulation setup:
  - Simulations use energy price paths over 2021-2027 relative to the 2016/19 average and deflated by the euro area GDP deflator.
  - Three scenarios: "baseline" (large) shock, "severe" shock, and "extreme" shock corresponding to different vintages of actual and expected energy price developments.
  - The shock path referred to as the 2022 energy price shock covers the entire price path until 2027 (implemented starting in 2021 for model/data consistency).
- Qualitative simulation findings:
  - Energy price hikes spur firms to increase investment in energy-saving technologies; model-predicted energy-saving productivity Ae comoves closely with energy prices when other shocks are shut down.
  - Energy efficiency gains partially buffer losses to potential output, but abrupt large spikes cause larger medium-term potential output losses due to transition costs in reallocating R&D and investment toward energy-saving technologies.
  - The paper reports simulated impacts on euro area potential output in 2027 across the three scenarios; the simulated impact ranges from a loss of 0.8 percent to higher losses depending on scenario severity (simulation details and extreme-scenario magnitudes are presented in the main results section).

### Caveats and interpretation nuances
- Confounding factors:
  - Fuel switching and sectoral reallocation can confound estimates of genuine within-sector energy efficiency gains; compositional effects may overstate aggregate energy-efficiency improvements.
  - Behavioral changes (e.g., households reducing thermostat settings) and lower demand due to reduced output can contribute to observed reductions in energy consumption and are difficult to fully disentangle from efficiency gains in aggregate data.
- Policy environment:
  - Results are conditional on a policy environment where energy efficiency was a stated policy goal in the European Union (for example, the EU’s directive on energy efficiency).
- Measurement cautions:
  - Aggregate estimates of up to a 10 percent improvement in energy efficiency by 2023 should be interpreted cautiously because compositional effects imply the true within-sector or within-firm improvement likely was noticeably less than 10 percent.

*IMF Working Paper: A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change*

### 2.2 percent.

### A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change

### Impact on Euro Area Potential Output
- Given ex-post price developments, the authors take a decline of somewhat below one percent in euro area potential output as a plausible estimate of the actual effect of the 2022 energy price shock; the text opens with "2.2 percent."
- Incorporating the endogenous response of firms, the authors find that as a consequence of the energy shock, European potential output in 2027 was reduced by about 0.8 percent relative to a baseline without shocks.
- Under the baseline vs severe shock comparison:
  - The loss in potential output in 2027 is 0.5 percentage points larger under the severe shock than under the baseline.
  - The severe shock corresponds to an energy price path about 30 percent higher each year over a five-year period.
- Counterfactual on responsiveness of energy efficiency:
  - If the baseline shock hit an economy with 80 percent lower responsiveness of energy efficiency to prices, the loss in potential output would have been about 1.3 percent by 2027.
  - This implies a lower degree of energy efficiency response would have reduced potential output by an additional 0.5 percentage point (relative to the baseline scenario with observed responsiveness).

### Energy Efficiency Response and Magnitude
- For the baseline shock scenario, the estimated improvement in energy efficiency is 3 percent relative to a no-shock counterfactual.
- The implied energy productivity gain of about 3 percent was crucial to limit the impact of the price shock; without this reaction, the output loss would have been around two-thirds higher.

### Dynamics: Level vs Growth Effects
- Potential growth effects are large in the first years after the shock but are essentially zero by the end of the period (the shock ultimately has a level, not a growth, effect on potential output).
- Despite a persistent increase in energy prices, the economy returns to the balanced growth path, so the shock produces a level offset rather than a permanent growth difference.

### Cross-Country Heterogeneity
- Under the baseline shock scenario, estimated potential output deviations in 2027 (deviation from no-shocks scenario) include:
  - Italy: around 1.2 percent loss
  - Germany: 0.9 percent loss
  - Spain: 0.6 percent loss
  - France: 0.4 percent loss
- These cross-country differences are driven mainly by:
  - Variation in the response of energy efficiency and elasticity of substitution parameters across countries.
  - Heterogeneous energy shares making the price shock not homogeneous across countries.

### Price Volatility vs Gradual Price Increases
- A volatile sharp price spike (the "extreme" shock) that reaches the same endpoint by 2027 as a smooth, monotonic, linear increase produces a deviation from the balanced growth path that is more than four times larger than the gradual increase scenario (impact on potential output in 2027 more than four times smaller for the gradual shock).
- The larger cost of sharp spikes is attributed to:
  - A larger integral (overall size) of the energy shock when increases are sharp and partially reversed.
  - Transition frictions that make shifting investment from capital-labor to energy-saving technology costly.
- Over the long run (by 2027, when the two scenarios face the same energy prices), impacts on potential output converge.
- The exercise compares a volatile exogenous price increase to a steady exogenous price increase; it does not study policy responses such as temporary price subsidies (which impose costs, e.g., paid for by distortionary taxation).

### Possible Confounders: Fuel Substitution and Structural Change
- Substitution across fossil fuels: modeled energy aggregate could miss fuel switching, but authors are not overly concerned given evidence of notable gains in overall fossil fuel energy efficiency.
- Substitution between fossil fuels and renewables:
  - Literature documents high elasticities of substitution between fossil fuels and renewables of around 2.
  - Empirical data for Europe show energy intensity dropped for both fossil fuels and other sources of energy (which includes renewables).
  - Prior work (Hassler, Krusell, and Olovsson (2022)) finds incorporating renewables does not qualitatively change model implications, though energy efficiency investments are directed towards renewables.
- Structural transformation and sectoral composition:
  - Aggregate model does not capture sectoral or firm-level dynamics; compositional effects could matter.
  - Shift-share decomposition (IEA data) shows historically within-sector energy efficiency gains have been the main driver in all countries (in no country does reallocation account for more than fifty percent).
  - For a broad group of euro area countries in 2022:
    - Compositional effects reduced energy consumption in industry and services by 5.9 percent.
    - Lower energy intensity lowered consumption by 4.4 percent.
    - Higher output increased consumption by 3.3 percent.
  - For a subgroup that accounts for around 40 percent of the broader group’s energy consumption and for which 2023 data are available, the historical developments are very similar (correlation of 0.92). 2023 data for this subgroup suggest energy savings came exclusively from lower energy intensity, with compositional effects slightly increasing consumption.

### Policy Implications and Conclusions
- Main findings:
  - Firms redirected resources toward energy-saving technology in reaction to the energy price shock, significantly increasing energy productivity and buffering adverse output effects.
  - This reallocation involved a tradeoff: resources diverted to energy-saving reduced efforts to save on other inputs, contributing to lower labor productivity.
- Policy-relevant takeaways:
  - Energy price signals matter; persistently suppressing them can hinder improvements in energy efficiency, which acts as a buffer against future adverse energy price shocks.
  - Abrupt price spikes are harmful; policies that reduce wholesale price volatility (for example, further integration of the European energy system) are desirable.
  - Policy should aim to lower price volatility by removing economic distortions (e.g., artificial fragmentation of European energy markets across national lines) rather than introducing new distortions (e.g., price subsidies).
  - The productivity tradeoff highlights the need to focus on capital-labor productivity in Europe; ambitious policy action is needed, including national structural reforms, deepening the EU single market, and joint provision of key public goods at the EU level.
- Broader implication:
  - Productivity growth should improve from the low of 2022-23 as the energy shock’s impact fades, and the temporary nature of the energy price drag on productivity means energy prices are unlikely to be the core reason behind Europe’s ongoing, longer-lasting productivity malaise.

*Source: IMF Working Paper "A Silver Lining? The European Energy Crisis through the Lens of Directed Technical Change" (Working Paper No. WP/2026/003).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2026/english/wpiea2026003-source-pdf.pdf_
