## ch3onlineannex

## Source details

**Canonical URL:** [ch3onlineannex](https://www.imf.org/-/media/files/publications/gfsr/2023/october/english/ch3onlineannex.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/gfsr/2023/october/english/ch3onlineannex.pdf.md)
- [Structured JSON version](/-/media/files/publications/gfsr/2023/october/english/ch3onlineannex.pdf.json)

---

### 1. Electricity system — key figures and cross-organizational estimates
- Renewable-based electricity generation, storage and distribution total: 2.419 (IEA), 2.088 (IRENA)
  - Of which storage and distribution: 1.212 (IEA)
  - Additional storage and distribution breakdown: 0.582 (IEA), 0.397 (IEA)
- Electrification in end-use sectors and Energy efficiency: 1.334 (IEA), 2.227 (IRENA)
  - Related storage/distribution: 0.636 (IEA)
- CCUS and other negative emission technologies (including BECCS): 0.205 (IEA), 0.102 (IRENA)
  - Related storage/distribution: N/A
- Others (hydrogen production and its derivatives, geothermal, solar thermal, marine, fossil fuels in existing oil and gas fields, nuclear) total: 1.025 (IEA), 0.163 (IRENA)
  - Related storage/distribution: 0.139 (IEA)
- Cross-organizational/calculated estimates:
  - IPCC (2022) (using IEA and IRENA data with IAMs and the IPCC AR6 Scenarios Database): total investments in mitigation need to increase by around three and six times with significant gaps across sectors and regions, amounting to US$4.5 trillion (averaged until 2030).
  - Songwe, Stern, and Bhattacharya (2022): EMDEs other than China will need to spend around $1 trillion per year by 2025, and around $2.4 trillion per year by 2030, on transformation of the energy system, adaptation and resilience, and sustainable agriculture.
  - McKinsey (2022): capital spending on physical assets for energy and land-use systems in the net-zero transition between 2021 and 2050 would amount to about $275 trillion, or $9.2 trillion per year on average.

### Online Annex 3.2 — Methodology for projecting the private sector share in climate finance
- Data sources used:
  - Climate Policy Initiative, CPI (2022) for current climate finance flows.
  - IEA (2023) for projected investment needs for 2030 in EMDEs.
  - IMF Investment and Capital Stock Dataset, 1960-2019, IMF (2021a) for calculating general government investment as a percent of GDP till 2019.
  - IMF WEO (2023) for the growth rate of general government net acquisition of nonfinancial assets and GDP.
- Baseline (2020) CPI-based private sector shares for EMDEs:
  - Total EMDEs climate flows in 2020: US$0.37 trillion with a private sector share of about 40 percent.
  - EMDEs excluding China total flows in 2020: US$0.15 trillion, of which 43 percent was from the private sector.
- Projection approach to 2030:
  - Determine public climate finance share as public climate investment as a percent of total public investment using CPI 2020 data.
  - Project total public investment to 2030 using IMF Investment and Capital Stock Dataset and IMF WEO growth rates for government net acquisition of nonfinancial assets and GDP.
  - Compute public climate investment in 2030 from the projected public-investment share; residual of investment needs not covered by public finance is attributed to the private sector.
- Important methodological notes:
  - All 2030 estimates are in real PPP (USD 2020) terms, using a WEO-derived deflator (nominal GDP / real GDP in PPP terms) rebased to 2020 and applied to both total public investment and GDP projections.
  - The WEO has projections till 2028; for 2029 and 2030 the average growth rate of the last 10 years was applied to extend series.
  - The baseline for calculating the increase in climate mitigation investment by 2030 is based on CPI (2022) climate finance flows; climate investment needs can also be met by regular finance beyond CPI-tracked flows.
  - “Private finance” is defined as financial flows not related to the public sector (governments, multilateral development banks, national development banks, state-owned banks, and other state-owned entities are public sector sources).

### Scenarios for public sector share of total climate investment in 2030 and worked example
- Scenarios for public sector share in 2030 (relative to 2020 public share of total public investment):
  - Lower bound: public-sector investment share doubles in 2030 from its 2020 share.
  - Baseline: public climate investment share in total public investment is 1.5 times its 2020 value.
  - Upper bound: public-sector investment share stays the same as in 2020.
- Worked calculation steps (as presented in the source):
  - Step 1: Use CPI 2020 flows and calculate gross public investment series.
  - Step 2: climate investment share in gross public investment in 2020 = (Public share * Total climate finance flow)/Total gross public investment = (1-40.5%)*0.372/3.144 = 7.05%
  - Step 3: Baseline 2030 climate investment share of gross public investment = 7.05% * 1.5 = 10.57%
  - Step 4: public sector share in 2030 = 10.57% * $3.385/$1.99 = 18%
  - Step 5: Private sector share in 2030 = 100 % – 18% = 82%
  - Step 6: Modify Step 3 assumptions for lower and upper bounds and repeat Steps 3–5.

### EMDE adjustments and sensitivity analysis results
- EMDE (including China) numeric changes relative to EMDEs excluding China:
  - Climate finance flow increased from US$ 0.146 trillion to US$ 0.372 trillion in 2020.
  - Gross public investment in 2020 increased from US$ 1.636 trillion to US$ 3.143 trillion in 2020.
- Sensitivity of Private Sector Share in total Climate Investment needs by 2030:
  - EMDEs (incl. China) — Lower bound / Baseline / Upper bound
    - 2020 flows (Current): 76.0% / 82.0% / 88.0%
    - 2019 flows: 78.4% / 83.8% / 89.2%
    - 2019 & 2020 (Average flows): 77.2% / 82.9% / 88.6%
  - EMDEs (excl. China) — Lower bound / Baseline / Upper bound
    - 2020 flows (Current): 85% / 89% / 93%
    - 2019 flows: 92% / 94% / 96%
    - 2019 & 2020 (Average flows): 89% / 92% / 94%
- Interpretation provided: estimates derived from 2019 and the average of 2019–20 flows indicate results are not very sensitive to the flows data.

### 2030 — GDP-weighted average of general government investment as a percent of GDP in 2019 and projection method
- Data and projection method:
  - 2019 GDP-weighted average of general government investment as a percent of GDP taken from IMF Investment and Capital Stock Dataset (1960–2019).
  - Project public investment share forward using projected growth rate of government net acquisition of nonfinancial assets from IMF WEO until 2030.
  - Multiply the projected ratio for 2030 by inflation-adjusted GDP projection for 2030 (IMF WEO) to calculate gross public investment.
  - All 2030 estimates expressed in real PPP (USD 2020) terms using a WEO-derived deflator; same deflator applied to total public investment series and GDP projections.
- Footnote methodological points preserved:
  - WEO projections extended to 2029 and 2030 by applying the average growth rate of the last 10 years.
  - Baseline climate mitigation investment increases to 2030 based on CPI (2022) climate finance flows.
  - Definition of “private finance” reiterated (financial flows not related to public sector institutions).

### Investment in Oil and Gas Sector Exploration and Production — empirical findings on sustainable fund flows and labels
- Major findings:
  - Greener funds by label attract more flows.
  - Climate impact funds domiciled in Europe have attracted the highest investments to date.
  - All types of sustainable investment funds experienced inflows significantly larger than conventional funds; among sustainable funds, only those targeting investors in advanced economies experienced higher inflows.
  - Funds with similar labels domiciled in emerging markets have not systematically benefited from higher inflows.
  - EMDEs face challenges attracting green investments due to systematically lower ESG scores (IMF, 2022b).
  - ESG funds (not climate impact funds) allocate smaller portions of investments to EMDEs compared to conventional funds.
  - Funds reclassified as “dark green” attract higher inflows compared to funds without Article 8 or Article 9 products; no statistically supported inflow effect for funds downgraded from Article 9 to Article 8.
- Empirical specification (panel-data regression):
  - Dependent variable: Flow_{i,t} (net capital flow into the fund).
  - Main regression: Flow_{i,t} = β1 Flow_{i,t-1}^{miss} + β2 Return_{i,t-1}^{miss} + β3 Labels_{i,t} + β4 MS Rating_{i,t} + β5 MS Sust Rating_{i,t} + β6 CRS_{i,t} + β7 Controls_{i,t} + δk,t + δj,t + ε_{i,t}
  - Controls and fixed effects: category-year fixed effects, domicile-year fixed effects; standard errors clustered at the fund level.
  - Independent variables include Labels_{i,t}, MS Rating_{i,t} (1–5 stars), MS Sust Rating_{i,t} (1–5 globes), CRS_{i,t} (Sustainalytics carbon-risk), and fund-level controls (lagged flows, lagged return, log fund size, expense ratio, log fund age).

### Online Annex 3.6 — Constructed firm-level climate impact scores (methodology, datapoints, and implications)
- Purpose and departures from Refinitiv:
  - Use only datapoints directly related to climate impact (subset of Refinitiv E datapoints).
  - Score calculation emphasizes continuous datapoints over binary datapoints.
  - Scores are not calculated relative to firms in the same sector; cross-sector comparisons allowed.
- Datapoints used (16 total; 10 continuous):
  - Continuous Negative polarity: Total CO2 and CO2 equivalents direct emission (scope 1) intensity (tons/revenue in millions of US$).
  - Continuous Negative polarity: Total CO2 and CO2 Scope 3 equivalent emission intensity (same units).
  - Continuous Positive polarity: Internal price on carbon per ton of CO2 equivalent (US$).
  - Continuous Positive polarity: Total primary renewable energy purchased and produced divided by total energy use.
  - Continuous Positive polarity: Total recycled and reused waste produced divided by total waste produced.
  - Binary Positive polarity: policy to improve energy efficiency? (Yes/No)
  - Binary Positive polarity: targets/objectives to be achieved on emission reduction? (Yes/No)
  - Binary Positive polarity: targets/objectives on energy efficiency? (Yes/No)
  - Continuous Positive polarity: Environmental R&D costs divided by revenue.
  - Continuous Negative polarity: NOx emissions (tons/revenue in millions of US$).
  - Continuous Negative polarity: SOx emissions (tons/revenue in millions of US$).
  - Continuous Negative polarity: Direct flaring/venting of natural gas emissions (tons/revenue in millions of US$).
  - Continuous Negative polarity: CO2 emissions in tons per ton of cement produced.
  - Binary Positive polarity: financial company public commitment to divest from fossil fuel? (Yes/No)
  - Binary Positive polarity: reports on initiatives to reduce/reuse/recycle/substitute/phase out SOx or NOx? (Yes/No)
  - Binary Positive polarity: reports on take-back/recycling programs or product features/services promoting responsible environmental use? (Yes/No)
- Scoring formulas:
  - Continuous positive polarity P: Score_{d,i}^{P} = ((Value_{d,i} - Value_{d,1%}) / (Value_{d,99%} - Value_{d,1%})) × 100, min 0, max 100.
  - Continuous negative polarity N: Score_{d,i}^{N} = ((Value_{d,99%} - Value_{d,i}) / (Value_{d,99%} - Value_{d,1%})) × 100, min 0, max 100.
  - Percentile cutoffs: 1st percentile (Value_{d,1%}) and 99th percentile (Value_{d,99%}) across reporting firms to limit outlier influence.
  - Binary scoring (positive polarity): Score_{b,i} = ((# firms with a lower value + # firms with equal value / 2) / Total applicable firms) × 100.
  - Aggregation: Impact Score_i = ∑_{d or b} ω_d/b × Score_{d/b,i} with ω_continuous = 5 × ω_binary and subject to ∑ ω = 1.
  - Missing-data treatment: firms not reporting a datapoint receive zero for that datapoint; only firms with at least three reported datapoints receive an overall score.
- Properties versus Refinitiv E(SG) scores:
  - Impact scores differ significantly from Refinitiv E (and E pillar) scores despite overlapping datapoints.
  - ESG and E scores are close to normally distributed; carbon intensities and absolute emissions are highly skewed.
  - Climate impact scores show a large share of firms with low scores due to poor performance and scarcity of climate-impact data (missing-data penalty).
- Portfolio implications:
  - Using impact scores instead of E scores materially changes portfolio allocations under exclusion/negative screening, best-in-class, and ESG integration strategies.
  - Firms in the best 5 percent by impact score often rank outside the top 5 percent by E score.
    - Best 5 percent corresponds to 400 firms in the sample.
    - Overlap between top 5 percent by impact score and top 5 percent by E score is about 20 percent.
  - Rank correlation between top firms by impact score and rank by E score: 0.16.

### Regional and country classification notes
- Online Annex 3.7 lists region and country groupings used; examples given for Emerging market, Developing economies, and High income groups.
- Exact sample composition varies across empirical analyses based on data availability.

_International Monetary Fund — ch3onlineannex (PDF chapter/section)._

### 1. Electricity system

### 1. Electricity system

### Renewable-based electricity generation, storage and distribution — key figures
- Total for renewable-based electricity generation, storage and distribution: 2.419 (IEA), 2.088 (IRENA)
- Of which storage and distribution: 1.212 (IEA)
- Additional storage and distribution breakdown: 0.582 (IEA), 0.397 (IEA)

### Electrification in end-use sectors and Energy efficiency — key figures
- Electrification in end-use sectors and Energy efficiency: 1.334 (IEA), 2.227 (IRENA)
- Related storage/distribution: 0.636 (IEA)

### CCUS and other negative emission technologies (including BECCS) — key figures
- CCUS and other negative emission technologies: 0.205 (IEA), 0.102 (IRENA)
- Related storage/distribution: N/A

### Others (hydrogen production and its derivatives, geothermal, solar thermal, marine, fossil fuels in existing oil and gas fields, nuclear) — key figures
- Others total: 1.025 (IEA), 0.163 (IRENA)
- Related storage/distribution: 0.139 (IEA)

### Alternative estimates and cross-organizational calculations
- IPCC (2022), using IEA and IRENA data with IAMs and the IPCC AR6 Scenarios Database, calculates that total investments in mitigation need to increase by around three and six times with significant gaps across sectors and regions, amounting to US$4.5 trillion (averaged until 2030).
- Songwe, Stern, and Bhattacharya (2022) estimate EMDEs other than China will need to spend around $1 trillion per year by 2025, and around $2.4 trillion per year by 2030, on three priorities: transformation of the energy system, adaptation and resilience, and sustainable agriculture. These estimates are based on Bhattacharya and others (2022).
- McKinsey (2022) estimates capital spending on physical assets for energy and land-use systems in the net-zero transition between 2021 and 2050 would amount to about $275 trillion, or $9.2 trillion per year on average. The methodology is based on the NGFS Net Zero 2050 scenario and covers the most carbon-intensive sectors (power, industry including steel and cement production, mobility, buildings, agriculture and food, and forestry and other land use, as well as fossil fuels). To estimate financing needs, the methodology analyzes effects of the transition and of climate change in 69 countries that represent the bulk of global GDP.

### Online Annex 3.2: Methodology for projecting the private sector share in climate finance
- Four data sources used:
  - Climate Policy Initiative, CPI (2022) for current climate finance flows.
  - IEA (2023) for projected investment needs for 2030 in EMDEs.
  - IMF Investment and Capital Stock Dataset, 1960-2019, IMF (2021a) for calculating general government investment as a percent of GDP till 2019.
  - IMF WEO (2023) for the growth rate of general government net acquisition of nonfinancial assets and GDP.
- Baseline private sector shares in 2020 for EMDEs (including and excluding China) are taken as given from CPI estimates:
  - Total EMDEs climate flows in 2020: US$0.37 trillion with a private sector share of about 40 percent.
  - EMDEs excluding China total flows in 2020: US$0.15 trillion, of which 43 percent was from the private sector.
- Calculation approach for private sector share in 2030:
  - Determine the current public climate finance share as public climate investment as a percent of total public investment using latest available CPI data (2020).
  - Project current public flows relative to total public investments forward to 2030 by projecting total public investment until 2030 using the IMF datasets and WEO growth rates.
  - The private sector share in 2030 is then calculated as the residual of investment not covered by the projected public sector share.

*Source: ch3onlineannex - 1. Electricity system (PDF chapter/section).*

### 2030. The GDP weighted average of general government investment as a percent of GDP in 2019

### 2030. The GDP weighted average of general government investment as a percent of GDP in 2019

### Data sources and projection method
- 2019 GDP weighted average of general government investment as a percent of GDP is taken from the IMF Investment and Capital Stock Dataset for 1960-2019.  
- To project the public investment share to GDP forward, the projected growth rate of government net acquisition of nonfinancial assets (IMF WEO) is applied until 2030.  
- The ratio for 2030 is multiplied by the (inflation adjusted) GDP projection for 2030 (IMF WEO) to calculate gross public investment.  
- All the 2030 estimates are in real PPP (USD 2020) terms, using a deflator from WEO calculated by dividing nominal GDP by real GDP (both in PPP terms), rebased to 2020. The same deflator is applied to the total public investment series and GDP projections to express figures in 2020 real terms.  
- Footnote methodological points preserved exactly:
  - The WEO has projections till 2028, for extending the GDP, General government net acquisition of nonfinancial assets, and total investment series, we applied the average growth rate of last 10 years to project the data for 2029 and 2030.
  - The baseline for calculating the increase in climate mitigation investment by 2030 are based on the climate finance flows given by CPI (2022). We would like to note that climate investment needs can also be met by regular finance, in addition to the climate finance flows tracked by CPI.
  - The term “private finance” refers to financial flows not related to the public sector. Public sector sources are public institutions such as governments (all levels), multilateral development banks, national development banks, state-owned banks, and other state-owned entities.

### Scenarios for public sector share of total climate investment in 2030
- The chapter considers three scenarios for the public sector share of total climate investment in 2030:
  - Lower bound of private sector scenario: assumes that the investment from the public sector as a share of total public investment doubles in 2030 from its 2020 share.  
  - Upper bound of private sector scenario: assumes that for the global estimate the investment from the public sector as a share of total public investment stays the same as compared with its 2020 share.  
  - Baseline scenario: assumes that the share of climate investment in total public investment is 1.5 times (midpoint of lower and upper bound) of its 2020 value.  
- Residual needs (after estimating public climate investment) are attributed to the private sector.

### Worked calculation steps (as presented)
- Step 1: Start with the CPI finance flows data (shown in black) for 2020 split between private and public sector, and calculate the series for gross public investment (shown in blue) as described above.  
- Step 2: Calculate climate investment share in gross public investment in 2020:
  - climate investment share in gross public investment in 2020 = (Public share * Total climate finance flow)/Total gross public investment
  - =(1-40.5%)*0.372/3.144 =7.05%
- Step 3: For the baseline in 2030:
  - climate investment share of gross public investment in 2030 (shown in red) = climate investment share in gross public investment in 2020 * Baseline public climate investment share in total public investment
  - =7.05% * 1.5 = 10.57%
- Step 4: Calculate the public sector share in 2030:
  - public sector share in 2030 = climate investment share of gross public investment in 2030 * Total gross public investment/ Investment needs estimate
  - = 10.57% * $3.385/$1.99  = 18%.
- Step 5: Calculate the Private sector share in 2030:
  - Private sector share in 2030 = (1 – Public sector share in 2030)
  - = 100 % – 18% = 82%
- Step 6: Change the assumptions in step 3 based on the scenarios and repeat steps 3-5 to calculate the upper and lower bounds.

### EMDE (including China) adjustments and notable numeric changes
- The main difference between the calculations for EMDEs, after including China are:
  - the climate finance flow increased from US$ 0.146 trillion to US$ 0.372 trillion in 2020.
  - the gross public investment in 2020 increased from US$ 1.636 trillion to US$ 3.143 trillion in 2020.

### Sensitivity analysis to climate finance flows (summary of results shown)
- Private Sector Share in total Climate Investment needs by 2030 - EMDEs (incl. China)
  - Lower bound / Baseline / Upper bound
  - 2020 flows (Current): 76.0% / 82.0% / 88.0%
  - 2019 flows: 78.4% / 83.8% / 89.2%
  - 2019 & 2020 (Average flows): 77.2% / 82.9% / 88.6%
- Private Sector Share in total Climate Investment needs by 2030 - EMDEs (excl. China)
  - Lower bound / Baseline / Upper bound
  - 2020 flows (Current): 85% / 89% / 93%
  - 2019 flows: 92% / 94% / 96%
  - 2019 & 2020 (Average flows): 89% / 92% / 94%
- Interpretation given in the source: these estimates derived from the 2019 and the average of 2019-20 flows show that the results are not very sensitive to the flows data.

### Implementation and usage notes embedded in methodology
- All 2030 estimates are expressed in real PPP (USD 2020) terms because the latest available data for climate finance flows by CPI are as of 2020.  
- The same WEO-derived deflator is applied to both GDP projections and total public investment series to express figures in 2020 real terms.  
- The approach back-calculates public climate investment from assumed shares of public investment devoted to climate, subtracts that amount from IEA investment needs estimates, and assigns the residual to private finance and investment.

*International Monetary Fund — ch3onlineannex (PDF chapter) — October 2023*

### 1. Investment in the Oil and Gas Sector Exploration and Production by

### 1. Investment in the Oil and Gas Sector Exploration and Production by

### Major empirical findings on sustainable fund flows and labels
- Greener funds by label attract more flows.
- Climate impact funds domiciled in Europe have thus far attracted the highest investments.
- While all types of sustainable investment funds experienced inflows significantly larger than those for conventional funds, among the sustainable funds only those targeting investors in advanced economies experienced higher inflows.
- Funds with similar labels domiciled in emerging markets have not systematically benefitted from higher inflows.
- EMDEs face challenges in attracting green investments due to their systematically lower ESG scores (IMF, 2022b).
- ESG funds (though not climate impact funds) allocate smaller portions of their investments to EMDEs compared to conventional funds, further contributing to the disadvantage these economies experience in accessing green finance.
- Initial analysis validates that funds reclassified as “dark green” attract higher inflows compared to funds without any Article 8 or Article 9 products. The same effect is not statistically supported for funds downgraded from Article 9 to Article 8.
- Sources: Morningstar, and IMF staff calculations.

### Empirical specification used to examine fund flows and labels
- Dependent variable:
  - Flow_{i,t}: net capital flow into the fund.
- Main regression equation (panel-data specification):
  - Flow_{i,t} = β1 Flow_{i,t-1}^{miss} + β2 Return_{i,t-1}^{miss} + β3 Labels_{i,t} + β4 MS Rating_{i,t} + β5 MS Sust Rating_{i,t} + β6 CRS_{i,t} + β7 Controls_{i,t} + δk,t + δj,t + ε_{i,t}
  - Fixed effects: category-year fixed effects and domicile-year fixed effects.
  - Standard errors clustered at the fund level.
- Independent variables:
  1. Labels_{i,t}: vector of dummies capturing presence of specific sustainability labels (ESG, Sustainability themed, climate action).
  2. MS Rating_{i,t}: Morningstar rating (1 to 5 stars).
  3. MS Sust Rating_{i,t}: Morningstar sustainability rating (1 to 5 globes).
  4. CRS_{i,t}: Portfolio carbon risk score (asset-weighted Sustainalytics carbon-risk rating).
  5. Controls_{i,t}: fund-level controls such as lagged flows (1 quarter), lagged return (1 quarter), log fund size, expense ratio, log fund age.

### Reclassification under EU SFDR and effects on flows
- Alternative specification replaces Labels dummies with dummies indicating:
  - Upgrade from light green to dark green.
  - Downgrade from dark green to light green.
- Data on reclassified funds limited; initial analysis indicates:
  - Funds reclassified as “dark green” attract higher inflows compared to funds without any Article 8 or Article 9 products.
  - No statistically supported inflow effect for funds downgraded from Article 9 to Article 8.
- Figure references in source: Chapter Figure 3.5, panel 2; Figure 3.5.2, panel 2.

### Online Annex 3.6 — Constructed firm-level climate impact scores versus corporate E(SG) scores
- Purpose:
  - Construct example impact scores using Refinitiv ESG datapoints but departing from Refinitiv methodology in three ways:
    1. Use only datapoints directly related to climate impact (subset of Refinitiv E datapoints).
    2. Score calculation reflects continuous nature of impact and gives more weight to continuous over binary datapoints.
    3. Scores are not calculated relative to firms in the same sector; scores take into account all firms, allowing cross-sector comparison.

- Climate impact-related datapoints (subset of Refinitiv ESG KPIs) include:
  - Continuous Negative polarity: Total CO2 and CO2 equivalents direct emission (scope 1) intensity (in tons divided by revenue in millions of US dollars).
  - Continuous Negative polarity: Total CO2 and CO2 Scope 3 equivalent emission intensity (unit as above).
  - Continuous Positive polarity: The internal price on carbon per ton of CO2 equivalent emissions (converted into US dollars).
  - Continuous Positive polarity: Total primary renewable energy purchased and produced divided by total energy use.
  - Continuous Positive polarity: Total recycled and reused waste produced divided by total waste produced.
  - Binary Positive polarity: “Does the company have a policy to improve its energy efficiency?”
  - Binary Positive polarity: “Has the company set targets or objectives to be achieved on emission reduction?”
  - Binary Positive polarity: “Has the company set targets or objectives to be achieved on energy efficiency?”
  - Continuous Positive polarity: Total amount of environmental R&D costs divided by revenue.
  - Continuous Negative polarity: Total amount of NOx emissions emitted (in tons divided by revenue in millions of US dollars).
  - Continuous Negative polarity: Total amount of SOx emissions emitted (in tons divided by revenue in millions of US dollars).
  - Continuous Negative polarity: Total direct flaring or venting of natural gas emissions (in tons divided by revenue in millions of US dollars).
  - Continuous Negative polarity: Total CO2 and CO2 equivalents emission in tons per ton of cement produced.
  - Binary Positive polarity: “Does the financial company have a public commitment to divest from fossil fuel?”
  - Binary Positive polarity: “Does the company report on initiatives to reduce, reuse, recycle, substitute, or phase out SOx (sulfur oxides) or NOx (nitrogen oxides) emissions?”
  - Binary Positive polarity: “Does the company report about take-back procedures and recycling programs to reduce the potential risks of products entering the environment or does the company report about product features or services that will promote responsible and environmentally preferable use?”
- Notes on datapoints:
  - The impact scores consider 16 data points in total, of which 10 data points are continuous measures.
  - These data points are a subset of the 68 datapoints used for the Refinitiv E score (with up to 29 continuous data points), and selection for regular E (and ESG) scores depends on industry.
  - The last 8 data points are not collected for firms in all industries and are considered only if they enter the Refinitiv ESG score calculation for a firm in a given industry.

### Calculation of impact scores — formulas and weighting
- Continuous datapoint scoring:
  - For a continuous data point with positive polarity P:
    - Score_{d,i}^{P} = ( (Value_{d,i} - Value_{d,1%}) / (Value_{d,99%} - Value_{d,1%}) ) × 100, with min 0 and max 100.
  - For a continuous data point with negative polarity N:
    - Score_{d,i}^{N} = ( (Value_{d,99%} - Value_{d,i}) / (Value_{d,99%} - Value_{d,1%}) ) × 100, with min 0 and max 100.
  - Cutoffs used: 1st percentile (Value_{d,1%}) and 99th percentile (Value_{d,99%}) across all firms reporting the data point to limit influence of outliers.
- Binary datapoint scoring (positive polarity, 1 = Yes, 0 = No):
  - Score_{b,i} = ((# firms with a lower value + # firms with equal value / 2) / Total number of applicable firms) × 100
  - Calculated across the entire sample of firms for which the data point is collected.
- Aggregation to overall impact score:
  - Impact Score_i = (∑_{d or b} ω_d/b × Score_{d/b,i}) where ω continuous measures are weighted higher:
    - ω_continuous = 5 × ω_binary, subject to ∑ ω = 1.
  - Firms where individual data points are not reported receive a zero score for those datapoints.
  - Only firms with at least three reported datapoints receive an overall score.
  - The impact score embeds a penalty for non-reported continuous datapoints (zero assigned), differing from Refinitiv which excludes missing continuous datapoints from calculation.

### Properties and distributional differences versus Refinitiv E(SG) scores
- The resulting impact scores are significantly different from Refinitiv E (and E pillar) scores even though all underlying datapoints of the impact scores also feed into the E pillar score.
- ESG and E scores (Refinitiv and many providers) are close to normally distributed because they are ordinal rankings within industries aggregated across many datapoints.
- Carbon intensities and absolute carbon emissions are highly skewed with a very long tail at high emission intensity levels; a normal distribution of ESG scores does not reflect this skew.
- The climate impact scores have a relatively large share of firms with low scores (larger probability mass to the right of the graph after axis reversal for carbon intensity), driven by:
  1. A relatively large share of firms performing poorly on climate impact-related datapoints.
  2. Scarcity of climate impact-related data, which reduces scores by construction due to missing-data penalty.

### Implications for portfolio allocations and ESG investment strategies
- Common ESG investment strategies:
  1. Exclusion / negative screening: exclude or underweight firms not meeting a threshold.
  2. Best-in-class: focus on firms with the best ESG characteristics within a peer group.
  3. ESG integration: incorporate ESG risks/opportunities into financial analysis and investment decisions.
- Using impact scores versus E scores leads to very different portfolio allocations under each strategy:
  - Under “negative screening” or “best-in-class”, very different firms would be excluded or overweighted when using impact scores versus E scores.
  - Firms ranking within the best 5 percent using impact scores, in most cases rank outside the top 5 percent using the E score.
    - Best 5 percent corresponds to 400 firms in the sample described.
    - The overlap between top 5 percent by impact score and top 5 percent by E score is about 20 percent.
  - Rank correlation between top firms by impact score and rank by E score is positive but low: 0.16.

### Regional and country classification (relevant for EMDE analysis and coverage)
- Online Annex 3.7 provides region and country groupings used in analyses; examples include:
  - Emerging market: Argentina, Brazil, Chile, China, Colombia, Egypt, Hungary, India, Indonesia, Iran, Malaysia, Mexico, Philippines, Poland, Russia, Saudi Arabia, South Africa, Thailand, Türkiye, United Arab Emirates.
  - Developing economies: long list including Afghanistan, Albania, Algeria, Angola, Antigua and Barbuda, Armenia, etc.
  - High income: Australia, Austria, Belgium, Canada, Chile, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Latvia, Lithuania, Luxembourg, Malta, Mauritius, Netherlands, New Zealand, Norway, Oman, Panama, Poland, Portugal, Qatar, Romania, Saudi Arabia, Singapore, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Trinidad and Tobago, United Arab Emirates, United Kingdom, United States, Uruguay.
- Note: Exact sample composition varies across empirical analyses based on data availability.

*Source: Morningstar; Refinitiv; IMF staff calculations; text and tables from the IMF GLOBAL FINANCIAL STABILITY REPORT — FINANCIAL SECTOR POLICIES TO UNLOCK PRIVATE CLIMATE FINANCE IN EMERGING MARKETS AND DEVELOPING ECONOMIES (October 2023).*

### References

### ch3onlineannex - References

### References

- Bhattacharya, Amar, Meagan Dooley, Homi Kharas, Charlotte Taylor. 2022. “Financing a big investment push in emerging markets and developing economies for sustainable, resilient and inclusive recovery and growth.” Grantham Research Institute on Climate Change and the Environment, London School of Economics and Political Science, and Brookings Institution. London, United Kingdom and Washington, DC.  
- Capotă, Laura-Dona, Margherita Giuzio, Sujit Kapadia, Dilyara Salakhova. 2022. “Are ethical and green investment funds more resilient?” Working Paper Series 2747, European Central Bank. Frankfurt, Germany.  
- Financial Stability Board (FSB). 2016. 2016 List of Global Systemically Important Insurers.  
- Financial Stability Board (FSB). 2022. 2022 List of Global Systemically Important Banks.  
- International Energy Agency (IEA). 2021. “Net Zero by 2050.” Paris, France.  
- International Energy Agency (IEA). 2023. “Scaling up Private Finance for Clean Energy in Emerging and Developing Economies.” Paris, France.  
- International Monetary Fund (IMF). 2021a, “Investment and Capital Stock Dataset, 1960-2019.” Washington, DC.  
- International Monetary Fund (IMF). 2021. “Investment funds: Fostering the transition to a green economy,” IMF Global Financial Stability Report, Chapter 3, October. Washington, DC.  
- International Monetary Fund (IMF). 2023. World Economic Outlook Update, July. Washington, DC.  
- International Renewable Energy Agency (IRENA). 2023. “World Energy Transition Outlook.” Abu Dhabi, United Arab Emirates.  
- Kreibiehl, S, T Yong Jung, S Battiston, P E Carvajal, C Clapp, D Dasgupta, N Dube, R Jachnik, K Morita, N Samargandi, M Williams. 2022. “Investment and finance.” In IPCC, 2022: “Climate Change 2022: Mitigation of Climate Change.” Contribution of Working Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [P R Shukla, J Skea, R Slade, A Al Khourdajie, R van Diemen, D McCollum, M Pathak, S Some, P Vyas, R Fradera, M Belkacemi, A Hasija, G Lisboa, S Luz, J Malley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY.  
- McKinsey & Company. 2022. “The net-zero transition. What it would cost, what it would bring.” McKinsey Global Institute. Washington, DC.  
- Naran, Baysa, Jake Connolly, Paul Rosane, Dharshan Wignarajah, Githungo Wakaba, and Barbara Buchner. 2022. “Global Landscape of Climate Finance: A Decade of Data 2011-2020.” Climate Policy Initiative, Washington, DC.  
- Songwe, Vera, Nicholas Stern, Amar Bhattacharya. 2022. “Finance for climate action: Scaling up investment for climate and development.” Report, Grantham Research Institute on Climate Change and the Environment and London School of Economics and Political Science. London, United Kingdom.

*Source: ch3onlineannex - References (PDF).*

---


_Source: https://www.imf.org/-/media/files/publications/gfsr/2023/october/english/ch3onlineannex.pdf_