## Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations? (WP/24/111)

## Source details

**Canonical URL:** [Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations? (WP/24/111)](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024111-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024111-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024111-print-pdf.pdf.json)

---

### Introduction and research question
- Research question: Whether higher shares of renewable energy in a country's energy mix reduce the sensitivity of domestic energy inflation to international fossil fuel price changes (the "divine coincidence" hypothesis).
- Empirical scope:
  - panel data spanning 50 years (from 1973 to 2022) covering 75 countries, including the three price-spike episodes 1973, 1979, and 2021/2.
- Motivation points:
  - Inflation surged in 2021 and 2022; increases in oil and gas prices were a dominant factor.
  - Schnabel (2022) — “energy accounted for more than 50% of headline inflation in the euro area, mainly reflecting the sharp increases in oil and gas prices”.
  - Renewable electricity production has marginal costs close to 0 (Hogan, 2022), suggesting a potential decoupling from international fossil energy prices.

### Theoretical framework
- Aggregate inflation decomposition and identity preserved:
  - π = Σ_i w_i π_i, where π is CPI inflation, w_i is item weight, p_t^i is price of item i in period t, and π_i = (p_t^i − p_{t−1}^i)/p_{t−1}^i.
- Country comparison formulation and decomposition:
  - Δπ_RE − Δπ_F = Σ_i (w_RE_i Δπ_RE_i − w_F_i Δπ_F_i)
  - Re-written to separate two channels:
    - Δπ_RE − Δπ_F = Σ_i [ (w_i (Δπ_RE_i − Δπ_F_i)) ︸ price effect [P]  +  ((w_RE_i − w_F_i) Δπ_F_i) ︸ weight effect [W] ]
- Channel interpretations:
  - Price channel [P]: with identical weights w_i, lower inflation in a high-RE country requires Δπ_RE_i < Δπ_F_i for dominant items.
  - Weight channel [W]: with identical item impacts Δπ_i, lower aggregate impact in high-RE country requires lower weights on high-inflation items (e.g., larger weight on power vs. car fuels).
- Factors affecting channel importance: trade openness, net energy import status, energy market design, energy subsidies and price controls, sources of renewable energy (electricity, biofuels, biomass), and economic weight.

### Data and empirical approach (overview)
- Dependent variable: energy inflation rates (rate of change of the energy consumer price index).
- Core explanatory variables: changes in international fossil energy prices; interaction terms between fossil-price changes and the share of renewable energy.
- Country-specific fossil energy price index construction (equation (4)):
  - F^LC_{i,t} = h_{s oil_{i,t}} O^USD_t + (1− s_{oil_{i,t}}) G^USD_{i,t} i ER^{USD−LC}_{i,t}
  - Interpretation preserved:
    - F^LC_{i,t} is the local-currency fossil energy price index for country i in year t.
    - s_{oil_{i,t}} and (1− s_{oil_{i,t}}) are the relative shares of oil and gas in the oil-and-gas consumption of that country.
    - O^USD_t is the USD oil price index in year t.
    - G^USD_{i,t} is the USD gas price index in year t, where i indexes the geographically closest gas hub for country i.
    - ER^{USD−LC}_{i,t} is the USD exchange rate index of country i’s currency in year t.
  - Outcome: country-specific fossil energy price change = year-on-year percent change of F^LC_{i,t}.
- Data sources summarized:
  - Inflation series from Ha, Kose, & Ohnsorge (2023) covering 1970–2022.
  - Renewable energy shares from Ritchie et al. (2023b); renewables defined as solar, wind and hydroelectric; LC = renewables + nuclear.
  - International energy prices from World Bank Commodity Price Database; exchange rates from BIS.
  - Fossil fuel subsidies from the Fossil Fuel Subsidy Tracker (OECD and IISD) for 2010–2021 (IMF “explicit” consumption subsidy estimates only).
  - Net energy import metrics from World Bank; energy import composition from IEA World Energy Balance.

### Main empirical findings (summary)
- Primary result:
  - Empirical results are inconsistent with the divine coincidence hypothesis: no evidence that increased renewable energy adoption reduces the impact of fossil fuel price changes on energy inflation rates.
- Robustness:
  - Finding is robust to sub-periods, country sub-samples, alternative metrics of inflation, fossil prices, and renewable energy.
- Suggested explanations for the counter-intuitive result:
  - Lack of country-level data on detailed energy policies beyond fossil-fuel subsidies (e.g., price controls) that could alter pass-through.
  - Threshold effects in electricity markets: the marginal producer (often coal or gas-fired) can set wholesale prices for the entire market.
  - Trade linkage spillovers: domestic decarbonization may not insulate from international fossil-price-driven inflation via cross-border trade and price transmission.

### Regression results and key estimated coefficients (baseline, lags, and interactions)
- Baseline (Table 3) key coefficients and notes (exact figures preserved):
  - ∆E (βE): 0.114∗∗∗, 0.121∗∗∗, 0.117∗∗∗, 0.118∗∗∗, 0.123∗∗∗ (standard errors: (0.009), (0.018), (0.019), (0.010), (0.012))
  - X (βX): 0.109, -0.019, 0.006, 0.132 (standard errors: (0.068), (0.056), (0.022), (0.118))
  - ∆E×X (βE·X): 0.001, 0.001, 0.000∗, -0.014∗∗∗ (standard errors: (0.001), (0.001), (0.000), (0.003))
  - Output gap (βg): 0.088∗, 0.133, 0.125, 0.088, 0.084 (standard errors: (0.035), (0.067), (0.067), (0.046), (0.047))
  - Inflation expectation (βEπ): 0.608∗∗∗, 0.689∗∗∗, 0.684∗∗∗, 0.598∗∗∗, 0.461∗ (standard errors: (0.052), (0.058), (0.058), (0.054), (0.217))
  - Sample metrics: R2: 0.49, 0.62, 0.62, 0.51, 0.45; Country-Year observations: 5,911; 2,472; 2,472; 4,705; 1,424; Countries: 170; 69; 69; 128; 151; Start years: 1973 (except column (5) start 2010); End year: 2022.
- Introducing lags (Table 4, column (1)) — interpretation preserved:
  - ∆E (βE): 0.109∗∗∗ (0.009)
  - L.∆E (βE1): 0.073∗∗∗ (0.006)
  - L2.∆E (βE2): 0.026∗∗∗ (0.005)
  - Interpretation: an increase in local-currency fossil energy prices of 1% is associated with an increase of energy inflation of 0.109% in the same year, 0.073% in the next year and 0.026% in the year after that.
- Interaction and modulating-variable findings with lags:
  - Share of renewables as X: contemporaneous and one-lag ∆E·X not significant; two-lag ∆E·X statistically significant but economically weak.
  - Fossil fuel subsidies as X: subsidies associated with a lower co-movement of international fossil energy prices and domestic energy inflation in the year of the price increase and the subsequent year.
- Two-modulating variables (Table 5) — preserved patterns:
  - ∆E coefficients generally positive and significant across columns; ∆E×X and lagged interactions mixed in significance and generally economically small.
  - Example: some columns show ∆E×X = 0.002∗∗ or L2.∆E×X = -0.002∗∗, but overall interpretation is that interaction terms are not both statistically and economically significant to corroborate divine coincidence.

### Illustrative sensitivity example (Figure 10, Table 5 column (1) regression)
- Facing ∆E = 20% in year 1:
  - Hypothetical country with net energy imports of 70% and renewable share of 10%:
    - energy inflation rise by 2.6% at end of year 1, 4.6% at end of year 2 and 4.8% at end of year 3, all else equal.
  - Same country with renewables share of 20% instead of 10%:
    - would not see its energy inflation sensitivity to ∆E change in a statistically significant way.
- Note: Given βE·X is not significant, no level of renewables would lead to a statistically significant divergence of energy inflation paths in this example.

### Robustness checks (Section 4.3 and Appendix B)
- Robustness settings and representative reported coefficients (exact values preserved):
  - Clustering at year-level (Table 8): ∆E 0.109 ∗∗∗ (0.012); L.∆E 0.074 ∗∗∗ (0.012); L2.∆E 0.026 ∗ (0.012); R2 0.50; Country-Year observations 5,895; Countries 170; Start 1973; End 2022.
  - Dropping outliers (Table 9): ∆E 0.097 ∗∗∗ (0.007); L.∆E 0.064 ∗∗∗ (0.006); L2.∆E 0.029 ∗∗∗ (0.004); R2 0.36.
  - Winsorizing at 150% (Table 10): ∆E 0.118 ∗∗∗ (0.011); R2 0.51.
  - Alternative ∆E definitions (Tables 11–13): ∆E ranges reported exactly (e.g., Table 11: ∆E 0.110 ∗∗∗ (0.009); Table 12: ∆E 0.068 ∗∗∗ (0.008) with ∆E×X 0.002 ∗∗∗ (0.001); Table 13: ∆E 0.231 ∗∗∗ (0.030)).
  - RE and LC defined as fraction of electricity production (Table 14): ∆E 0.109 ∗∗∗ (0.009); ∆E×X -0.000 (0.000).
  - Sub-periods and subsamples (Tables 15–18): examples include Table 15 (1985–2019) ∆E 0.092 ∗∗∗ (0.010); Table 16 (2000–2022) ∆E 0.077 ∗∗∗ (0.007) with ∆E×X 0.002 ∗∗ (0.001); Table 17 (AEs) ∆E 0.149 ∗∗∗ (0.014); Table 18 (EMDEs) ∆E 0.094 ∗∗∗ (0.010) with L2.∆E×X -0.005 ∗∗ (0.001).
  - PPI as dependent variable (Table 19): ∆E 0.178 ∗∗∗ (0.014); R2 0.64.
- Robustness-pattern summary:
  - Instantaneous coefficient on ∆E is positive and statistically significant across multiple specifications and robustness checks, with reported values ranging from 0.068 ∗∗∗ to 0.231 ∗∗∗ depending on specification and sample.
  - Lagged coefficients L.∆E and L2.∆E are often positive and statistically significant in many specifications (examples preserved from tables).
  - Interaction terms (e.g., ∆E×X, L.∆E×X, L2.∆E×X) are usually small in magnitude; some statistically significant but economically negligible in several checks (except fossil subsidies which show a clearer dampening association).

### Thresholds and electricity-price–specific analysis (Section on marginal pricing)
- Motivation: marginal pricing in liberalized wholesale electricity markets can keep wholesale prices tied to marginal producers (e.g., gas-fired plants), creating threshold effects.
- Electricity-price regression (equation (8)) and threshold tests:
  - Continuous regressions (share of renewables in electricity production) show interaction terms not economically or statistically significant.
  - Threshold regression (RE > 85%, Table 6 column (6)) reported coefficients:
    - ∆E (βE): 0.125∗∗∗ (0.024)
    - L.∆E (βE1): 0.106∗∗∗ (0.020)
    - L2.∆E (βE2): 0.015 (0.016)
    - X (βX): -7.708 (4.584)
    - L.∆E×X (βE·X1): -0.133∗ (0.051)
  - Interpretation preserved: for countries with very high shares of renewables in electricity production (≥85%) electricity prices rise in the year of an international fossil energy price shock but do not rise in the second year while they do for other countries — indicating a possible threshold-formulation of the divine coincidence hypothesis with a delayed and limited shielding effect.
- Limitations noted:
  - Shielding effect is relatively weak, materializes with a one-year delay only, and is limited to one item (electricity) in the energy CPI basket.
  - Results obtained only on a sub-sample of European countries; further research on threshold effects needed.

### Explanations considered for the empirical result and research priorities
- Explanations considered (Section 5.1):
  - National energy policies (market structure, long-term price agreements, price caps/rules) not fully captured by fossil-fuel subsidies data may mask a divine coincidence signal.
  - Marginal pricing and threshold effects: unless renewables dominate marginal hours, marginal-price setting by gas/coal keeps prices linked to fossil inputs.
  - Trade linkages and spillovers: country-level decoupling can be undermined by cross-border price transmission and international demand for renewables.
- Research priorities and policy implications (Section 5.2):
  - More structured datasets needed on country-level energy policies (price controls, market design), sources of renewables by use (electricity vs. transport vs. heating), and cross-border trade linkages to better identify channels and thresholds.
  - If divine coincidence were confirmed, implications would follow for monetary policy conduct and the greening of the monetary policy toolkit; current evidence does not support that renewable expansion has already reduced fossil-fuel price pass-through to domestic energy inflation at the country level.
  - Policy stance: not a critique of renewable development; renewables reduce greenhouse gas emissions and local air pollution. The analysis does not advocate subsidizing fossil fuels.

### Key statistics and factual points (preserved exactly)
- Time span of dataset: 50 years (from 1973 to 2022).
- Number of countries in sample: 75.
- Episodes included: 1973, 1979 and 2021/2.
- Quotation preserved: Schnabel (2022) — “energy accounted for more than 50% of headline inflation in the euro area, mainly reflecting the sharp increases in oil and gas prices”.
- Model identity preserved: π = Σ_i w_i π_i.
- Selected summary statistics (Table 1; units [%] preserved exactly):
  - Headline inflation: mean 44.88, sd 37.7, min -72.7, p25 2.3, p50 5.6, p75 11.8, max 65,374.1, n 9,220
  - Energy inflation: mean 38.6, sd 1,209.4, min -98.7, p25 1.3, p50 4.6, p75 11.3, max 94,802.4, n 6,744
  - Energy price change: mean 161.0, sd 8,324.0, min -99.9, p25 -3.3, p50 7.0, p75 28.1, max 747,182.0, n 10,466
  - RE in consumption: mean 11.0, sd 12.8, min 0.0, p25 2.1, p50 6.1, p75 15.9, max 74.3, n 3,382
  - RE in electricity: mean 30.7, sd 32.3, min 0.0, p25 2.1, p50 17.5, p75 54.5, max 100.0, n 5,480
  - LC in consumption: mean 15.0, sd 14.9, min 0.0, p25 3.6, p50 10.8, p75 21.3, max 74.3, n 3,382
  - Net energy import: mean -75.2, sd 559.6, min -17,632.8, p25 -23.3, p50 23.0, p75 60.6, max 100.0, n 5,997
  - Output gap: mean 0.3, sd 37.8, min -1,270.0, p25 -4.1, p50 -0.1, p75 3.8, max 2,208.5, n 9,711
  - Fossil subsidy: mean 1.4, sd 2.9, min 0.0, p25 0.0, p50 0.3, p75 1.3, max 31.7, n 2,109
- Correlations preserved (Table 2 entries):
  - Headline inflation with Energy inflation: 0.96
  - Headline inflation with Energy price change: 0.79
  - Energy inflation with Energy price change: 0.89
  - RE in consumption with Fossil subsidy: -0.27
  - LC in consumption with Fossil subsidy: -0.30
  - Net energy import with Fossil subsidy: -0.25

*Prepared by Laurent Millischer, Chenxu Fu, Ulrich Volz, and John Beirne; IMF Working Paper WP/24/111 — Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations?*

### 2022. In the wake of recently increased oil and gas prices leading to a surge in inflation, the notion of a “divine

### Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations?

### Introduction and research question
- Research question: Whether higher shares of renewable energy in a country's energy mix reduce the sensitivity of domestic energy inflation to international fossil fuel price changes (the "divine coincidence" hypothesis).
- Empirical scope: panel data spanning 50 years (from 1973 to 2022) covering 75 countries, including the three price-spike episodes 1973, 1979, and 2021/2.
- Motivation:
  - Inflation surged in 2021 and 2022; increases in oil and gas prices were a dominant factor.
  - Schnabel (2022) noted that in February 2022 “energy accounted for more than 50% of headline inflation in the euro area, mainly reflecting the sharp increases in oil and gas prices”.
  - Renewable electricity production has marginal costs close to 0 (Hogan, 2022), suggesting a potential decoupling from international fossil energy prices.

### Theoretical framework (simple model of energy prices and inflation)
- Aggregate inflation decomposition:
  - π = Σ_i w_i π_i, where π is CPI inflation, w_i is item weight, p_t^i is price of item i in period t, and π_i = (p_t^i − p_{t−1}^i)/p_{t−1}^i.
- Country comparison formulation:
  - Difference in energy-inflation impacts between a high-renewables country (RE) and a low-renewables country (F) can be written as:
    - Δπ_RE − Δπ_F = Σ_i (w_RE_i Δπ_RE_i − w_F_i Δπ_F_i)
  - Re-written to separate two channels:
    - Δπ_RE − Δπ_F = Σ_i [ (w_i (Δπ_RE_i − Δπ_F_i)) ︸ price effect [P]  +  ((w_RE_i − w_F_i) Δπ_F_i) ︸ weight effect [W] ]
- Channel interpretations:
  - Price channel [P]: With identical weights w_i, a lower inflation impact in RE requires Δπ_RE_i < Δπ_F_i for dominant items (i.e., renewable sources lower the price response of affected items).
  - Weight channel [W]: With identical item impacts Δπ_i, a lower aggregate impact in RE requires that high-inflation items have lower weights in RE’s consumption basket (e.g., larger weight on power vs. car fuels).
- Influencing factors that affect channel importance:
  - Trade openness, net energy import status, energy market design, energy subsidies and price controls, sources of renewable energy (electricity, biofuels, biomass), and economic weight.

### Data and empirical approach (overview from source)
- Data sources and aggregation described in Section 3 (details not reproduced here).
- Key variables:
  - Dependent variable: energy inflation rates (rate of change of the energy consumer price index).
  - Core explanatory variables: changes in international fossil energy prices; interaction terms between fossil-price changes and the share of renewable energy.
- Empirical strategy: regress energy inflation rates on macroeconomic controls and changes in international fossil energy prices; include interaction of fossil-price changes with renewable share to test whether renewables dampen the pass-through.

### Main empirical findings
- Primary result:
  - Empirical results are inconsistent with the divine coincidence hypothesis: no evidence that increased renewable energy adoption reduces the impact of fossil fuel price changes on energy inflation rates.
- Robustness:
  - The finding is robust to sub-periods, country sub-samples, and alternative metrics of inflation, fossil prices, and renewable energy.
- Suggested explanations for the counter-intuitive empirical result:
  - Lack of country-level data on detailed energy policies beyond fossil-fuel subsidies (e.g., price controls) that could alter pass-through.
  - Potential threshold effects, especially in electricity markets where the marginal producer (often coal or gas-fired plants) sets wholesale prices for the entire market.
  - Trade linkage spillovers: domestic decarbonization may not insulate a country from international fossil-price-driven inflation via cross-border trade and price transmission.

### Implications and research priorities
- Policy implications if divine coincidence were confirmed (conditional statement from source):
  - If renewable expansion both mitigates fossil-fuel-induced inflation volatility and fights climate change, this would have bearing on the conduct of monetary policy and on efforts to "green" the monetary policy toolkit.
- Given current empirical evidence:
  - The study does not find support for arguing that renewable expansion has already reduced fossil-fuel price pass-through to domestic energy inflation at the country level.
- Research priorities highlighted:
  - More work is required to understand how the energy transition will affect inflation dynamics at both country and global levels.
  - Need for structured datasets capturing country-level energy policies (price controls, market design), the sources of renewables by use (electricity vs. transport vs. heating), and cross-border trade linkages to better identify channels and possible threshold effects.

### Key statistics and factual points (preserved exactly)
- Time span of dataset: 50 years (from 1973 to 2022).
- Number of countries in sample: 75.
- Episodes included: 1973, 1979 and 2021/2.
- Quotation preserved: Schnabel (2022) — “energy accounted for more than 50% of headline inflation in the euro area, mainly reflecting the sharp increases in oil and gas prices”.
- Model identity preserved: π = Σ_i w_i π_i and related equations as presented in the source.

*Prepared by Laurent Millischer, Chenxu Fu, Ulrich Volz, and John Beirne; IMF Working Paper WP/24/111 — Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations?*

### Section 3.2 discusses how we compute the change of fossil-fuel prices for each country and Section 3.3

### Section 3.2–3.3: Change in International Fossil Energy Prices and Stylized Facts

### Data sources (Section 3.1)
- Inflation
  - Measures drawn from Ha, Kose, & Ohnsorge (2023) covering 37 advanced economies (AEs) and 159 emerging markets and developing economies (EMDEs) for 1970-2022.
  - Database contains CPI inflation, energy CPI (ECPI), and PPI inflation. Analysis focuses on ECPI and its response to international fossil energy prices.
  - OECD definition of energy CPI inflation followed; data drawn from OECD.Stat, UNdata, Consumer Price Index, Eurostat, FRED, and country-specific sources (up to 52 countries).
- Renewable energy
  - Renewable energy share from Ritchie et al. (2023b), combining BP, IEA and others.
  - Renewable energy defined as solar, wind and hydroelectric energy; measured as share in total primary energy consumption or share in electricity production.
  - Share of nuclear energy also used; “low-carbon” (LC) = sum of renewables and nuclear.
- Energy price
  - International energy prices from the World Bank Commodity Price Database (the “Pink Sheet”).
  - Database contains USD-denominated price index of crude oil, several natural gas hubs and coal.
  - Exchange rates from the Bank of International Settlement (BIS) used to convert international fossil energy prices into local currency for each country.
- Energy imports
  - Share of renewable and fossil in total energy import drawn from the World Energy Balance by the IEA.
  - Net energy import as percentage of energy use from the World Bank used to categorize countries into high energy importer, low importer, or exporter.
  - Net energy import data was discontinued in 2015, so the latest observation is carried forward.
- Fossil fuel subsidies
  - Data from the Fossil Fuel Subsidy Tracker (OECD and IISD) available for 185 countries from 2010 to 2021.
  - Data picks up IMF “explicit” consumption subsidy estimates only.
  - Total subsidy as a percentage of GDP computed for each country and used as a proxy for price rigidity policy.

### Construction of country-specific fossil energy price index (Section 3.2)
- Rationale
  - International fossil energy price “seen” by a country differs from the international market due to exchange rate, energy imports, and domestic energy mix.
  - Construct a country-specific international fossil energy price index capturing those factors.
- Formula (equation (4))
  - F^LC_{i,t} = h_{s oil_{i,t}} O^USD_t + (1− s_{oil_{i,t}}) G^USD_{i,t} i ER^{USD−LC}_{i,t}
  - Described in text as:
    - F^LC_{i,t} is the local-currency fossil energy price index for country i in year t.
    - s_{oil_{i,t}} and (1− s_{oil_{i,t}}) are the relative shares of oil and gas in the oil-and-gas consumption of that country.
    - O^USD_t is the USD oil price index in year t.
    - G^USD_{i,t} is the USD gas price index in year t, where i indexes the geographically closest gas hub for country i.
    - ER^{USD−LC}_{i,t} is the USD exchange rate index of country i’s currency in year t.
  - For gas prices, the geographically closest available gas exchange price is used (three gas hub prices available: Henry Hub in the US, Dutch TTF in the EU and the Tokyo hub). Adjustment made to account for divergence in gas prices in 2021/22.
  - For countries without energy mix data, the international average share is used.
- Outcome
  - The country-specific fossil energy price change is the year-on-year percent change of the local-currency fossil energy index defined in equation (4).
  - Country-specific fossil energy price changes broadly follow international fossil energy price dynamics but with larger distributions across countries due to weighting and exchange rate fluctuations.

### Stylized facts (Section 3.3)
- Summary statistics (Table 1; all units are [%])
  - Headline inflation: mean 44.88, sd 37.7, min -72.7, p25 2.3, p50 5.6, p75 11.8, max 65,374.1, n 9,220
  - Energy inflation: mean 38.6, sd 1,209.4, min -98.7, p25 1.3, p50 4.6, p75 11.3, max 94,802.4, n 6,744
  - Energy price change: mean 161.0, sd 8,324.0, min -99.9, p25 -3.3, p50 7.0, p75 28.1, max 747,182.0, n 10,466
  - RE in consumption: mean 11.0, sd 12.8, min 0.0, p25 2.1, p50 6.1, p75 15.9, max 74.3, n 3,382
  - RE in electricity: mean 30.7, sd 32.3, min 0.0, p25 2.1, p50 17.5, p75 54.5, max 100.0, n 5,480
  - LC in consumption: mean 15.0, sd 14.9, min 0.0, p25 3.6, p50 10.8, p75 21.3, max 74.3, n 3,382
  - Net energy import: mean -75.2, sd 559.6, min -17,632.8, p25 -23.3, p50 23.0, p75 60.6, max 100.0, n 5,997
  - Output gap: mean 0.3, sd 37.8, min -1,270.0, p25 -4.1, p50 -0.1, p75 3.8, max 2,208.5, n 9,711
  - Fossil subsidy: mean 1.4, sd 2.9, min 0.0, p25 0.0, p50 0.3, p75 1.3, max 31.7, n 2,109
- Correlations (Table 2)
  - Key correlations (matrix entries preserved as shown):
    - Headline inflation with Energy inflation: 0.96
    - Headline inflation with Energy price change: 0.79
    - Energy inflation with Energy price change: 0.89
    - RE in consumption with Fossil subsidy: -0.27
    - LC in consumption with Fossil subsidy: -0.30
    - Net energy import with Fossil subsidy: -0.25
  - Noted patterns:
    - High correlation between headline and energy inflation.
    - High correlation between RE and LC shares.
    - Fossil subsidies tend to be higher for countries with a low share of renewables and low-carbon energy (-27% and -30% correlation respectively) and higher for countries that have lower energy imports (-25% correlation).
- Additional stylized facts and visual patterns
  - Fossil fuels dominate energy imports
    - In the sample, all countries imported at least 97% fossil of total energy import in 1980.
    - Fossil import share decreased over four decades but still accounts for over 92% of total energy imports in 75% of countries in 2020.
    - Renewable energy imports are below 2% in most countries despite increases in recent decades.
  - Energy inflation over time
    - Two clear hikes in energy inflation: during the oil crises in the 1970s and the energy price shock after 2021.
  - Renewables share over time
    - Median share of renewables hovered just below 5% from 1970 to 2005, then gradually increased to above 10%.
  - Country example — Denmark
    - Share of renewables rose steadily from close to 0% in 1990 to over 40% in 2022.
    - Scatter plot and regression for Denmark suggest recent increases in fossil energy prices (when share of renewables was higher) are not associated with lower energy inflation than past increases, indicating a higher share of renewables in Denmark might not have shielded energy inflation from changes in international fossil energy prices.

*International Monetary Fund — wpiea2024111-print-pdf*

### Section 3.2.

### Section 3.2.

### Regression results: baseline (Table 3)
- Computed as the percentage gap of real GDP to its Hodrick-Prescott filtered trend; controlling for the output gap allows for a better identification in the case that both countries’ energy inflation and international fossil fuel prices are driven by a global demand shock.
- The expected inflation is proxied by the average headline inflation of years t−1, t−2 and t−3.
- Key estimated coefficients (Table 3):
  - ∆E (βE): 0.114∗∗∗, 0.121∗∗∗, 0.117∗∗∗, 0.118∗∗∗, 0.123∗∗∗ (standard errors: (0.009), (0.018), (0.019), (0.010), (0.012))
  - X (βX): 0.109, -0.019, 0.006, 0.132 (standard errors: (0.068), (0.056), (0.022), (0.118))
  - ∆E×X (βE·X): 0.001, 0.001, 0.000∗, -0.014∗∗∗ (standard errors: (0.001), (0.001), (0.000), (0.003))
  - Output gap (βg): 0.088∗, 0.133, 0.125, 0.088, 0.084 (standard errors: (0.035), (0.067), (0.067), (0.046), (0.047))
  - Inflation expectation (βEπ): 0.608∗∗∗, 0.689∗∗∗, 0.684∗∗∗, 0.598∗∗∗, 0.461∗ (standard errors: (0.052), (0.058), (0.058), (0.054), (0.217))
- Sample and specification notes (Table 3):
  - Country FE: YYYYY
  - Cluster: cty
  - R2: 0.49, 0.62, 0.62, 0.51, 0.45
  - Country-Year observations: 5,911; 2,472; 2,472; 4,705; 1,424
  - Countries: 170; 69; 69; 128; 151 (breakdowns by AE, EM, LIC provided)
  - Start years: 1973, 1973, 1973, 1973, 2010
  - End year: 2022 (all columns)
- Interpretation:
  - Low-carbon energy (renewables + nuclear) as X is not associated with a statistically significant βE·X ≠ 0 in the baseline; higher consumption of low-carbon energy is not associated with a higher or lower energy inflation in the face of an increase in international energy prices.
  - Net energy imports as X (column (4)) yields βE·X = 0.000 with weak statistical significance and no economic significance.
  - Fossil fuel subsidies as a share of GDP (column (5)): ∆E remains significant and the interaction ∆E·X is significant; higher fossil fuel subsidies are associated with a lower impact on energy inflation of an increase in international fossil energy prices.

### Introducing lags (equation (6) and Table 4)
- Rationale: wholesale and retail contracts lock in prices and adjust at set dates; feed-through from international prices to domestic energy inflation takes time.
- Regression specification (6): πe i,t = βX Xi,t + sum_{k=0}^{N_X} [βE k ∆E i,t−k + βE·X k ∆E i,t−k · Xi,t] + ...
- Key lagged estimates (Table 4, column (1)):
  - ∆E (βE): 0.109∗∗∗ (0.009)
  - L.∆E (βE1): 0.073∗∗∗ (0.006)
  - L2.∆E (βE2): 0.026∗∗∗ (0.005)
  - Interpretation: an increase in local-currency fossil energy prices of 1% is associated with an increase of energy inflation of 0.109% in the same year, 0.073% in the next year and 0.026% in the year after that. The third lag was not significant and excluded.
- Interaction and modulating-variable findings with lags (Table 4):
  - Using share of renewables as X (column (2)): βX weakly significant and positive; contemporaneous and one-lag ∆E·X not significant; two-lag ∆E·X statistically significant but economically weak.
  - Using low-carbon share as X (column (3)): results similar to renewables.
  - Using net imports as X (column (4)): contemporaneous interaction weakly significant but β∆E·X = 0.000 (no economic significance).
  - Using fossil fuel subsidies as X (column (5)): subsidies associated with a lower co-movement of international fossil energy prices and domestic energy inflation in the year of the price increase and the subsequent year.
- Macro controls (output gap, inflation expectation) remain significant with intuitive signs.

### Two modulating variables (equation (7) and Table 5)
- Specification (7): πe i,t = βE ∆E i,t + βX Xi,t + βE·X ∆E i,t·Xi,t + βY Yi,t + βE·Y ∆E i,t·Yi,t + ...
- Combinations studied: X = share of renewables (RE) or share of low-carbon energy (LC); Y = net imports (IMP), fossil fuel subsidies (SUB), GDP per capita (GDP).
- Key results (Table 5 columns (1)–(6)):
  - ∆E (βE): e.g., 0.110∗∗∗, 0.092∗∗∗, 0.165, 0.109∗∗∗, 0.093∗∗∗, 0.177∗ (standard errors vary by column)
  - L.∆E (βE1) and L2.∆E (βE2) remain often positive and significant in many specifications.
  - ∆E×X (βE·X): mixed significance—some columns show small positive and significant coefficients (e.g., 0.002∗∗, 0.002∗).
  - L2.∆E×X (βE·X2): several negative and statistically significant coefficients (e.g., -0.002∗∗, -0.002∗).
  - ∆E×Y (βE·Y0): small or negative coefficients; some statistically significant (e.g., -0.008∗∗).
- Interpretation:
  - None of the interaction terms are both statistically and economically significant in a way that corroborates the divine coincidence hypothesis.
  - Example (column (1) of Table 5): higher consumption of renewable energy is not associated with a weaker co-movement of energy inflation and international energy prices when controlling for energy imports. The statistically significant second-lag interaction is economically negligible.

### Illustrative sensitivity example (Figure 10)
- Using regression from Table 5 column (1):
  - Facing an increase of international fossil energy prices (∆E) of 20% in year 1:
    - A hypothetical country with net energy imports of 70% and renewable share of 10% would see energy inflation rise by 2.6% at the end of the first year, 4.6% at the end of the second year and 4.8% at the end of the third year, all else equal.
    - That same country with renewables share of 20% instead of 10% would not see its energy inflation sensitivity to ∆E change in a statistically significant way.
  - Note: Given βE·X is not significant, no level of renewables would lead to a statistically significant divergence of energy inflation paths.

### Robustness checks (Section 4.3)
- The empirical inconsistency with the divine coincidence hypothesis is robust to:
  - Clustering standard errors at the year instead of country level (Table 8).
  - Dropping outliers instead of winsorizing (Table 9) or winsorizing at 150% (Table 10).
  - Alternative definitions of fossil energy price changes (Table 11, Table 12, Table 13).
  - Using shares of renewables and nuclear in electricity production rather than in total energy consumption (Table 14).
  - Different sub-periods (Table 15 and Table 16) and sub-regions (Table 17 and Table 18).
  - Using producer price index (PPI) inflation as the dependent variable (Table 19).

### Thresholds and electricity-price–specific analysis (equation (8) and Table 6)
- Motivation: marginal pricing in liberalized wholesale electricity markets can keep wholesale prices tied to marginal producers (e.g., gas-fired plants), creating a threshold effect where renewables only decouple prices once they dominate as marginal producers.
- Electricity-price regression (8): πelec i,t = βX Xelec i,t + sum_{k=0}^{2} [βE k ∆E i,t−k + βE·X k ∆E i,t−k · Xelec i,t] + ...
- Continuous and threshold dummies tested for X = share of renewables in electricity production (thresholds: >50%, >60%, >70%, >80%, >85%).
- Key electricity-price results (Table 6):
  - Continuous regression (column (1)): interaction terms not economically or statistically significant.
  - Threshold regression (column (6), RE > 85%):
    - ∆E (βE): 0.125∗∗∗ (0.024)
    - L.∆E (βE1): 0.106∗∗∗ (0.020)
    - L2.∆E (βE2): 0.015 (0.016)
    - X (βX): -7.708 (4.584)
    - L.∆E×X (βE·X1): -0.133∗ (0.051)
    - Interpretation: for countries with very high shares of renewables in electricity production (≥85%) electricity prices rise in the year of an international fossil energy price shock but do not rise in the second year while they do for other countries—indicating a possible threshold-formulation of the divine coincidence hypothesis with a delayed and limited shielding effect.
- Limitations:
  - The shielding effect is relatively weak, materializes with a one-year delay only, and is limited to one item (electricity) in the energy CPI basket.
  - Results are obtained only on a sub-sample of European countries; further research on the threshold effect is needed.

### Explanations considered for counter-intuitive empirical result (Section 5.1)
- National energy policies:
  - Variation in energy policies (market structure, long-term price agreements, price caps/rules such as the “Iberian exception”) may not be fully captured by fossil fuel subsidies data and could drown a divine coincidence signal.
- Marginal pricing and threshold effects:
  - In liberalized wholesale power markets the marginal producer sets the price; if marginal producer is gas-fired, wholesale (and thus retail-indexed) prices move with gas regardless of renewables share, implying a threshold effect where only very high renewables shares (approaching 100%) change marginal hours sufficiently to decouple prices.
- Trade linkages and spillovers:
  - A country fully covered by renewables could see international demand for renewable energy products rise if trading partners face fossil shocks, raising domestic prices and defeating country-level decoupling; the divine coincidence might hold only at global level, not individual-country level.

### Conclusions and implications (Section 5.2)
- Empirical finding: results are inconsistent with the divine coincidence hypothesis across specifications, samples, and robustness checks.
- Policy-relevant takeaways:
  - This should not be interpreted as a critique of renewable energy development; renewables deliver established benefits (e.g., reducing greenhouse gas emissions, mitigating climate change, reducing local air pollution).
  - The analysis does not advocate subsidizing fossil fuels, which distort prices, fail to address externalities, and accelerate climate change.
  - As the global economy shifts to low-carbon energy, further research should explicitly account for national energy policies, threshold effects, and trade linkages.
  - Confirming a robust divine coincidence would have implications for monetary policy conduct and the greening of the monetary policy toolkit (NGFS, 2021).

*IMF Working Paper — Section 3.2.*

### References

### wpiea2024111-print-pdf - References

### Oil prices, pass-through, and inflation dynamics
- Baumeister, C., & Peersman, G. (2013). Time-Varying Effects of Oil Supply Shocks on the US Economy. American Economic Journal: Macroeconomics, 5(4), 1-28. doi: 10.1257/mac.5.4.1
- Baba, C., & Lee, J. (2022). Second-round effects of oil price shocks – Implications for Europe’s inflation outlook. IMF Working Papers.
- Binder, C., & Makridis, C. (2022). Stuck in the Seventies: Gas Prices and Consumer Sentiment. The Review of Economics and Statistics, 104(2), 293-305. doi: 10.1162/rest a00944
- Binder, C. C. (2018). Inflation expectations and the price at the pump. Journal of Macroeconomics, 58, 1-18. doi: 10.1016/j.jmacro.2018.08.006
- Castro, C., & Jiménez-Rodríguez, R. (2017). Oil price pass-through along the price chain in the euro area. Energy Economics, 64, 24-30. doi: 10.1016/j.eneco.2017.03.012
- Chen, S.-S. (2009). Revisiting the inflationary effects of oil prices. The Energy Journal, 30(4), 141–154. doi: 10.5547/ISSN0195-6574-EJ-Vol30-No4-5
- Clark, T. E., & Terry, S. J. (2010). Time variation in the inflation passthrough of energy prices. Journal of Money, Credit and Banking, 42. doi: 10.1111/j.1538-4616.2010.00347.x
- Conflitti, C., & Luciani, M. (2019). Oil price pass-through into core inflation. Energy Journal,

### Macroeconomic theory and real wage/inflation mechanisms
- Blanchard, O., & Galí, J. (2007). Real Wage Rigidities and the New Keynesian Model. Journal of Money, Credit and Banking, 39(s1), 35-65. doi: 10.1111/j.1538-4616.2007.00015.x
- Bernanke, B., & Blanchard, O. (2023). What Caused the U.S. Pandemic-Era Inflation? Hutchins Center Working Papers.

### Energy, climate policy, and renewable energy interactions with macroeconomics
- Akan, T. (2023). Can renewable energy mitigate the impacts of inflation and policy interest on climate change? Renewable Energy, 214, 255-289. doi: 10.1016/j.renene.2023.05.023
- Birol, F. (2022). A call to clean energy. Finance & Development.

### IMF work on fossil fuel subsidies, climate tools, and related data
- Black, S., Liu, A., & Ian Parry, N. V. (2023). IMF Fossil Fuel Subsidies Data: 2023 Update. IMF Working Papers.
- Black, S., Parry, I., Mylonas, V., Vernon, N., & Zhunussova, K. (2023). The IMF-World Bank Climate Policy Assessment Tool (CPAT): A Model to Help Countries Mitigate Climate Change. IMF Working Papers.

### Other methodological and topical contributions
- Blanchard, O., & Galí (2007) — see "Real Wage Rigidities and the New Keynesian Model" entry above for theoretical background.

*References section from wpiea2024111-print-pdf - References*

### 40.  doi:  10.5547/01956574.40.6.ccon

### Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations? Working Paper No. WP/2024/111

### References and cited literature
- Bibliographic citations and working-paper references listed (selection): Corbeau, A.-S., Farfan, J. C., & Orozco, S. (2023); Deka, A., Cavusoglu, B., & Dube, S. (2022); Deka, A., & Dube, S. (2021); Erdmann, G. (2017); Galí, J., & Monacelli, T. (2005); Gao, L., Kim, H., & Saba, R. (2014); Gugler, K., Haxhimusa, A., & Liebensteiner, M. (2018); Ha, J., Kose, M. A., & Ohnsorge, F. (2023); Heemskerk, I., Nerlich, C., & Parker, M. (2022); Hogan, W. W. (2022); Ielpo, F. (2018); IPCC (2022); Kilian, L. (2008, 2009); Kilian, L., & Zhou, X. (2022); Melodia, L., & Karlsson, K. (2022); NGFS (2021); Panetta, F. (2022); Peersman, G., & Van Robays, I. (2012); Pous, P. D., Patuleia, A., Brown, S., & Rosslowe, C. (2022); Ritchie, H., Roser, M., & Rosado, P. (2023a, 2023b); Schnabel, I. (2022); Wen, F., Zhang, K., & Gong, X. (2021); Wong, B. (2015).

### Appendix A — Country list (used in regressions)
- Countries and codes (excerpted as presented):  
  - Algeria 612 DZA EM; Argentina 213 ARG EM; Australia 193 AUS AE; Austria 122 AUT AE; Azerbaijan 912 AZE EM; Bangladesh 513 BGD LIC; Belarus 913 BLR EM; Belgium 124 BEL AE; Brazil 223 BRA EM; Bulgaria 918 BGR EM; Canada 156 CAN AE; Chile 228 CHL EM; China 924 CHN EM; Colombia 233 COL EM; Croatia 960 HRV EM; Czech Republic 935 CZE AE; Denmark 128 DNK AE; Egypt 469 EGY EM; Estonia 939 EST AE; Finland 172 FIN AE; France 132 FRA AE; Germany 134 DEU AE; Greece 174 GRC AE; Hungary 944 HUN EM; India 534 IND EM; Indonesia 536 IDN EM; Iran 429 IRN EM; Iraq 433 IRQ EM; Ireland 178 IRL AE; Israel 436 ISR AE; Italy 136 ITA AE; Japan 158 JPN AE; Kazakhstan 916 KAZ EM; Latvia 941 LVA AE; Lithuania 946 LTU AE; Luxembourg 137 LUX AE; Macedonia 962 MKD EM; Malaysia 548 MYS EM; Mexico 273 MEX EM; Morocco 686 MAR EM; Netherlands 138 NLD AE; New Zealand 196 NZL AE; Norway 142 NOR AE; Oman 449 OMN EM; Pakistan 564 PAK EM; Peru 293 PER EM; Philippines 566 PHL EM; Poland 964 POL EM; Portugal 182 PRT AE; Romania 968 ROU EM; Russia 922 RUS EM; Saudi Arabia 456 SAU EM; Singapore 576 SGP AE; Slovak Republic 936 SVK AE; Slovenia 961 SVN AE; South Africa 199 ZAF EM; South Korea 542 KOR AE; Spain 184 ESP AE; Sweden 144 SWE AE; Switzerland 146 CHE AE; Thailand 578 THA EM; Trinidad and Tobago 369 TTO EM; Turkey 186 TUR EM; Ukraine 926 UKR EM; United Kingdom 112 GBR AE; United States 111 USA AE; Uzbekistan 927 UZB LIC; Venezuela 299 VEN EM; Vietnam 582 VNM LIC.  
- Appendix table presents full country coverage and classification (AE, EM, LIC) used in the regression sample.

### Appendix B — Additional robustness checks: main empirical findings (regression coefficients and robustness outcomes)
- General pattern across robustness checks: the coefficient on ∆E (change in energy-related exchange-rate/fossil-energy metric) is consistently positive and statistically significant in the reported specifications; interaction terms with modulating variables (RE, LC, IMPSUB) are generally economically small and often statistically insignificant, with some exceptions noted in specific specifications.

- Table 8 (standard errors clustered at the year level):  
  - ∆E 0.109 ∗∗∗ (0.012)  
  - L.∆E 0.074 ∗∗∗ (0.012)  
  - L2.∆E 0.026 ∗ (0.012)  
  - X 0.152 (0.129)  
  - ∆E×X 0.001 (0.001)  
  - Output gap (g) 0.069 ∗ (0.029)  
  - Inflation expectation (Eπ) 0.539 ∗∗∗ (0.040)  
  - R2 0.50; Country-Year observations 5,895; Countries 170; Start 1973; End 2022.

- Table 9 (dropping outliers instead of winsorizing):  
  - ∆E 0.097 ∗∗∗ (0.007)  
  - L.∆E 0.064 ∗∗∗ (0.006)  
  - L2.∆E 0.029 ∗∗∗ (0.004)  
  - X 0.137 ∗ (0.061)  
  - ∆E×X 0.001 ∗ (0.001)  
  - Inflation expectation (Eπ) 0.347 ∗∗∗ (0.069)  
  - R2 0.36; Country-Year observations 5,473; Countries 169; Start 1973; End 2022.

- Table 10 (winsorizing at 150% instead of 100%):  
  - ∆E 0.118 ∗∗∗ (0.011)  
  - L.∆E 0.077 ∗∗∗ (0.008)  
  - L2.∆E 0.020 ∗∗ (0.007)  
  - X 0.183 ∗ (0.078)  
  - Inflation expectation (Eπ) 0.661 ∗∗∗ (0.073)  
  - R2 0.51; Country-Year observations 5,895; Countries 170; Start 1973; End 2022.

- Table 11 (∆E defined using USD fossil-fuel prices and exchange rates, no composition effects):  
  - ∆E 0.110 ∗∗∗ (0.009)  
  - L.∆E 0.072 ∗∗∗ (0.006)  
  - L2.∆E 0.027 ∗∗∗ (0.005)  
  - X 0.158 ∗ (0.069)  
  - Inflation expectation (Eπ) 0.538 ∗∗∗ (0.054)  
  - R2 0.50; Country-Year observations 5,895; Countries 170.

- Table 12 (∆E defined using USD fossil energy prices only, same ∆E for all countries):  
  - ∆E 0.068 ∗∗∗ (0.008)  
  - L.∆E 0.056 ∗∗∗ (0.006)  
  - L2.∆E 0.020 ∗∗∗ (0.005)  
  - ∆E×X 0.002 ∗∗∗ (0.001) — positive, statistically significant but economically small; authors note this points to higher currency depreciation for high-renewables countries when fossil energy prices increase.  
  - Inflation expectation (Eπ) 0.666 ∗∗∗ (0.053)  
  - R2 0.47; Country-Year observations 6,025; Countries 172; Start 1973; End 2022.

- Table 13 (adding year fixed effects so ∆E captures exchange-rate changes only):  
  - ∆E 0.231 ∗∗∗ (0.030)  
  - L.∆E 0.120 ∗∗∗ (0.020)  
  - L2.∆E -0.012 (0.021)  
  - X 0.108 (0.085)  
  - Inflation expectation (Eπ) 0.445 ∗∗∗ (0.063)  
  - R2 0.54; Country-Year observations 5,895; Countries 170.

- Table 14 (RE and LC defined as fraction of electricity production rather than of total energy consumption):  
  - ∆E 0.109 ∗∗∗ (0.009)  
  - L.∆E 0.074 ∗∗∗ (0.006)  
  - L2.∆E 0.026 ∗∗∗ (0.005)  
  - X 0.049 ∗ (0.021)  
  - ∆E×X -0.000 (0.000)  
  - Inflation expectation (Eπ) 0.539 ∗∗∗ (0.054)  
  - R2 0.50; Country-Year observations 5,895.

- Table 15 (sub-period 1985–2019 excluding oil price and COVID shocks):  
  - ∆E 0.092 ∗∗∗ (0.010)  
  - L.∆E 0.050 ∗∗∗ (0.007)  
  - L2.∆E 0.018 ∗∗ (0.006)  
  - X -0.046 (0.071)  
  - Inflation expectation (Eπ) 0.511 ∗∗∗ (0.051)  
  - R2 0.50; Country-Year observations 4,537; Countries 169; Start 1985; End 2019.  
  - Note reported change: SUB no longer significant in this sub-period.

- Table 16 (sub-period 2000–2022 where renewable shares increased):  
  - ∆E 0.077 ∗∗∗ (0.007)  
  - L.∆E 0.043 ∗∗∗ (0.005)  
  - L2.∆E 0.011 (0.006)  
  - X 0.396 ∗∗∗ (0.089)  
  - ∆E×X 0.002 ∗∗ (0.001)  
  - Inflation expectation (Eπ) 0.626 ∗∗∗ (0.150)  
  - R2 0.38; Country-Year observations 3,443; Countries 170; Start 2000; End 2022.

- Table 17 (sub-sample: advanced economies):  
  - ∆E 0.149 ∗∗∗ (0.014)  
  - L.∆E 0.103 ∗∗∗ (0.010)  
  - L2.∆E 0.015 ∗ (0.007)  
  - X 0.303 ∗∗∗ (0.072)  
  - Inflation expectation (Eπ) 0.545 ∗∗∗ (0.041)  
  - R2 0.61; Country-Year observations 1,571; Countries 35; Start 1973; End 2022.

- Table 18 (sub-sample: emerging markets and low-income countries):  
  - ∆E 0.094 ∗∗∗ (0.010)  
  - L.∆E 0.064 ∗∗∗ (0.007)  
  - L2.∆E 0.029 ∗∗∗ (0.007)  
  - X -0.283 (0.186)  
  - L2.∆E×X -0.005 ∗∗ (0.001) — marginal economic significance: higher shares in renewables and low-carbon energy lead to a somewhat lower co-movement with a two-year lag.  
  - Inflation expectation (Eπ) 0.542 ∗∗∗ (0.062)  
  - R2 0.49; Country-Year observations 4,324; Countries 135; Start 1973; End 2022.

- Table 19 (using producer price index (PPI) inflation as dependent variable):  
  - ∆E 0.178 ∗∗∗ (0.014)  
  - L.∆E 0.081 ∗∗∗ (0.007)  
  - L2.∆E 0.013 (0.009)  
  - X 0.193 ∗∗ (0.071)  
  - Inflation expectation (Eπ) 0.454 ∗∗∗ (0.039)  
  - R2 0.64; Country-Year observations 3,005; Countries 109; Start 1973; End 2022.

### Robustness-pattern summary (empirical outcomes preserved)
- The instantaneous coefficient on ∆E is positive and statistically significant across multiple specifications and robustness checks, with reported values ranging (exactly as reported) from 0.068 ∗∗∗ to 0.231 ∗∗∗ depending on specification and sample.  
- Lagged coefficients L.∆E and L2.∆E are often positive and statistically significant in many specifications (examples include 0.074 ∗∗∗, 0.056 ∗∗∗, 0.120 ∗∗∗ for L.∆E; and 0.026 ∗, 0.020 ∗∗∗, -0.012 for L2.∆E in different tables).  
- Interaction terms (e.g., ∆E×X, L.∆E×X, L2.∆E×X) are usually small in magnitude; some are statistically significant but described in the tables as economically insignificant in several checks (for example, Table 9 notes the interaction terms "exhibit statistical significance but no economic significance (except for fossil subsidies)").  
- Results are broadly robust to clustering standard errors at the year level, alternative treatments of outliers (dropping vs. winsorizing), alternative winsorization thresholds (150%), alternative ∆E definitions (USD fossil energy prices only, inclusion/exclusion of exchange rates), adding year fixed effects, alternative modulating variable definitions (fraction of electricity production vs. total energy consumption), alternative sample periods (1985–2019; 2000–2022), and subsamples (advanced economies; emerging markets and low-income countries).  
- PPI-based regressions (Table 19) show qualitatively similar results to CPI-based specifications.

*Source: IMF Working Paper — Do Renewables Shield Inflation from Fossil Fuel-Price Fluctuations? Working Paper No. WP/2024/111*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024111-print-pdf.pdf_