## 1. Descriptive Statistics

## Source details

**Canonical URL:** [1. Descriptive Statistics](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021089-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021089-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021089-print-pdf.pdf.json)

---

### I. INTRODUCTION
- Daily emissions in early April 2020 fell by about 17 percent compared with 2019 levels.
- Recent studies forecast a reduction in annual emissions of about 4 to 7 percent in 2020, and that emissions will rebound in 2021.
- UNEP (2019) estimates that emissions need to decline by 2.7 percent annually in order to reach the 2°C goal by 2030.
- The COVID-19 crisis may both impede and accelerate the transition to a low-carbon economy:
  - Impediments: heightened economic uncertainty, corporate balance sheet vulnerabilities, reduced investment and research in long-horizon, capital-intensive green projects, and rescue packages/subsidies that support firms or activities not compatible with long-term climate mitigation.
  - Accelerants: structural shifts in consumer and investor preferences toward environmentally friendly goods and services; increased awareness of long-term disaster prevention facilitating green policy measures such as carbon taxes and “green recovery” packages.
- Oil price effects on firms’ incentives are theoretically ambiguous:
  - Lower oil prices may relax firms’ financial constraints but reduce incentives to improve energy efficiency and invest in renewables.
  - Lower oil prices can also hurt oil-sector profitability, reducing fossil-fuel investment and easing competition for clean energy firms.
- Net effect of an oil price shock depends on origin:
  - Demand-driven negative oil price shocks (e.g., reducing economic activity and oil demand) could lead to lower corporate environmental performance via delayed investments and tighter financial conditions.
  - Oil supply shocks that lower oil prices can increase global economic activity, relax firms’ financial constraints, and allow firms to improve environmental performance.
- Key research questions:
  1. How has the COVID-19 crisis affected green investments so far?
  2. What can be learned from past economic crises about firm behavior regarding the greening of the economy?
  3. What do developments in the oil market imply about firms’ environmental performance?
- Core empirical findings preview:
  - The COVID-19 crisis has not led to a sustained decline in green financing so far; flows into sustainable funds and performance of sustainable assets have been robust.
  - Tighter financial constraints and economic downturns are associated with weaker environmental performance and lower levels of green investments by firms.
  - Predominantly demand-driven negative oil price shocks (as during COVID-19) are likely detrimental for firms’ environmental performance; oil supply shocks that lower oil prices are associated with higher environmental performance.
  - Adverse financial and economic shocks that limit activity and tighten financial constraints could reverse progress in corporate environmental performance by several years.
  - Policy implication: public policies and recovery packages that boost green investments are warranted. Fostering sustainable finance through better disclosures, development of green taxonomies, and product standardization may help mobilize green investments.

### II. THE COVID-19 CRISIS AND FINANCING THE ENERGY TRANSITION
- Green corporate bond issuance:
  - Trended up over the past decade; declined in March 2020 during market turmoil but stayed within the range of historical deviations.
  - Issuance picked up thereafter, with the share of green bonds in total corporate bond issuance more than doubling between March and June 2020.
- Syndicated loan market:
  - Loans to firms with an above-median environmental performance score have increased over the past decade relative to loans to firms with a below-median score.
  - Lending to both types of firms dropped slightly in the first quarter of 2020.
- Investment funds:
  - Sustainable and environmental-focused funds, especially fixed-income funds, continued to attract investment throughout the crisis with only a small drop in aggregate inflows in some asset classes.
  - A possible driver: relatively high returns that green investments experienced during the crisis.
- Overall assessment:
  - The impact of the COVID-19 crisis on environmental finance has been modest and short-lived to date.
  - Given the persistence and severity of the shock—decline in output, potential scarring, and heightened uncertainty—it is uncertain whether these trends will continue and what the overall impact will be on firms’ environmental performance and their contribution to global mitigation efforts.
  - The paper examines firm environmental performance during previous episodes of financial and economic stress to draw implications for the pandemic context.

### III. LESSONS FROM PAST ECONOMIC CRISES FOR THE ENERGY TRANSITION DURING THE COVID-19 CRISIS

#### A. Financial Constraints and Firms’ Environmental Performance
- Baseline empirical specification:
  - EE_{i,s,c,t} = α_s + γ_c + δ_t + β_1 FINCONSTR_{i,t−1} + μ′ X_{i,t−1} + ε_{i,s,c,t}
  - EE_{i,s,c,t}: environmental score for firm i in sector s, country c, time t.
  - α_s, γ_c, δ_t: sector, country, and time (year)-fixed effects.
  - X_{i,t−1}: firm-level controls (e.g., logarithm of total assets, earnings before interest and taxes).
  - FINCONSTR_{i,t−1}: firm-level financial constraint measure (several alternative definitions used).
- Financial constraint measures:
  - Firm size: logarithm of firm’s total assets (in estimations the sign is reversed so higher values indicate smaller firms).
  - Dividends: dummy = 1 if firm does not pay dividends, 0 otherwise.
  - Ratings: dummy = 1 if a firm with positive debt-to-asset ratio is not rated according to Standard and Poor’s, 0 otherwise.
  - Interest Coverage Ratio (ICR): dummy = 1 if earnings before interest and taxes are below interest expenses, 0 otherwise.
  - KZ score: dummy = 1 if Kaplan-Zingales score is above the median, 0 otherwise.
- Data coverage:
  - Sample period: 2002 to 2019.
  - Sample composition: about 7,000 listed firms, 69 industries (Global Industry Classification Standard), 62 economies.
  - Data frequency: annual.
- Environmental scores:
  - Obtained from Refinitiv.
  - Based on 68 metrics across three environmental categories: Resource use, Emissions, and Innovation.
  - Category scores calculated using a rank scoring methodology; overall scores are weighted averages with industry-varying weights.
  - Proprietary environmental aggregate scores range between 0 and 100.
- Key estimation results (Table 2, panel A):
  - Tighter financial constraints are associated with worse environmental performance.
  - Specific magnitudes:
    - Environmental performance falls by 10 points when firm size drops from the median to the 25th percentile of the firm size distribution.
    - When a firm does not pay dividends, its environmental score is 4 points lower.
    - When a firm is not rated, its environmental score is 3 points lower.
- Caveats:
  - Data cover only publicly listed firms.
  - Lack of standardization and transparency across data providers; some scores are self-reported.

#### B. Financial constraints and firms’ investments in green technologies
- Hypothesis: Financial constraints reduce firms’ investments in green technologies and intangibles.
- Empirical approach: Probit model for probability of environmental investment P(I_{i,s,c,t}=1) = Փ(α_s + γ_c + δ_t + β_1 CONSTRAINT_{i,t-1} + μ′ X_{i,t-1}).
- Firm-level findings (Table 2, panel B):
  - The probability that a firm will make an environmental investment falls by 6 percentage points when firm size drops from the median to the 25th percentile of the firm size distribution.
  - A firm unable to pay dividends has a 3 percentage points lower likelihood of making a green investment.
- Robustness: Results robust to alternative financial constraint definitions, inclusion of macro controls (lagged output gaps, lagged log WTI, lagged VIX), and balanced panel from 2010.

#### C. Firms’ environmental performance and macro-financial shocks
- Shocks analyzed:
  - Global financial stress proxied by the VIX.
  - Real economic activity shock defined as change in the annual output gap (WEO).
- Dynamic-response model: E_{i,s,c,t+h} = α_{s}^h + γ_{c}^h + β^h CONSTRAINT_{c,t}^{...} + δ′^h FIRM_CONSTRAINT_{i,t-1} + γ′^h MACRO_CONTROLS_{t-1} + ε_{i,s,c,t+h}.
- Key results:
  - A sudden jump in the VIX comparable to that observed during the first half of 2020 would lead to a persistent drop in firms’ environmental performance by up to 5 points, with the pre-shock performance level not attained for at least three years after the shock.
  - Absent policy actions and behavioral changes, average corporate environmental performance would return to the levels that prevailed in 2006.
  - Interaction with financial constraints:
    - The adverse effect of global financial shocks is magnified for financially constrained firms.
    - For firms not paying dividends or for unrated firms in 2019, the global financial stress shock observed thus far in 2020 is estimated to lower environmental performance by 1 and 2 additional points, respectively, compared with dividend-paying or rated firms.
  - Output-gap shock magnitude and effects:
    - A large decline in the output gap (10 percentage points, about 50 percent larger than that observed in the Group of Seven (G7) economies during the global financial crisis) would lead to a 3 point decline in firms’ environmental performance in the medium term.
    - Firms’ carbon intensity (total carbon emissions relative to revenue) could increase by up to 8.5 percent in the medium term after such a decline in the output gap.
- Controls: firm-level controls include log total assets and EBIT; macro controls include log WTI, country-specific output gaps, and VIX.

#### D. Firms’ environmental performance and oil-market shocks
- Oil shock identification:
  - Supply and demand shocks from the structural VAR of Baumeister and Hamilton (2019); median of the posterior distribution aggregated from monthly to annual by averaging.
  - Oil inventory demand (speculative demand) shock excluded.
- Findings on oil-price-source heterogeneity (Table 4, panel B):
  - When oil prices fall due to demand-side factors, environmental corporate performance has historically been weaker.
  - When oil prices decline due to an oil supply shock, environmental performance of firms has generally improved.
  - The COVID-19-induced oil price shock is largely demand-driven; therefore, firms’ environmental performance is likely to suffer significantly.
- Robustness: Including world economic activity shocks from Baumeister and Hamilton (2019) does not qualitatively change results.

### Overall synthesis of empirical evidence
- Tighter financial constraints are associated with weaker corporate environmental performance.
- Adverse global financial and output shocks that increase uncertainty and amplify firms’ financial constraints weigh significantly on firms’ environmental performance.
- A reduction in oil prices amid a decline in global economic activity is unlikely on its own to lift corporate environmental performance.
- Absent strong supportive policy actions, tighter financial constraints and weaker economic activity related to the COVID-19 crisis may act as a drag on firms’ environmental performance in the future.

### Policy implications and recommendations
- Public policies and green recovery packages to support firms’ environmental performance during the COVID-19 crisis are warranted.
- To alleviate firms’ financial constraints and aid green investment, key policy measures include support for the sustainable finance sector, specifically:
  - Better disclosure standards.
  - Development of green taxonomies.
  - Product standardization.

### Annex A.1. — Robustness and supplementary results
- Logarithm transformation of environmental scores yields qualitatively unchanged results.
- Comparative differences:
  - Environmental score is 1 point lower when a firm’s interest coverage ratio is below 1.
  - Environmental score is 1 point lower when the Kaplan-Zingales index is above the median.
  - Firms that do not pay dividends and unrated firms have environmental scores multiple points lower than dividend-paying and rated firms.
- Sub-category (emissions and resource use) results:
  - For a firm that does not pay dividends, its emissions and resource use scores drop by 6 points.
  - For an unrated firm, its emissions and resource use scores drop by 3 and 4 points, respectively, compared with 3 points for the overall environmental score.
- Robustness to fixed effects and alternative financial-constraint measures, balanced panel from 2010, and alternative macro controls confirmed.
- Appendix material and data sources:
  - Country-level macro/financial variables from Refinitiv Datastream, Haver Analytics, IMF WEO, Federal Reserve Bank of St. Louis, U.S. Energy Information Administration, and updated Baumeister and Hamilton (2019) series.
  - Firm-level variables from Refinitiv Datastream and Standard & Poor’s.
  - Appendix tables report alternative specifications, balanced panel results, and replacement of country fixed effects with environmental policy or World Economic Forum indices.
- Figures and tables summarized:
  - Figure: Reduction in daily CO2 emissions in 2020 vs 2019 (Percent); carbon futures prices (Euros per metric ton).
  - COVID-19 and green investments: series on green bond issuance, syndicated loans by environmental-score groups, sustainable fund flows (2003: Q1–2020: Q1), and cumulative returns of green vs conventional equity indices (Percent).
  - VIX shock defined as a 16.3-point increase (difference in average VIX 2020 up to July 31, 2020 vs 2019) and local projection firm-level responses.
  - Output-gap shock scaled as a 10 percentage points drop; corresponding responses in environmental score and log CO2 emissions relative to revenue (Percent).
  - Table summaries: descriptive statistics; regression analyses for environmental performance and financial constraints; alternative environmental measures; oil-market shocks; standard errors clustered at firm level with ***, **, * for 1%, 5%, 10% significance.

*Source: wpiea2021089-print-pdf — IMF staff figures, tables, and appendix material as provided in the content unit.*

### 1. Descriptive Statistics ___________________________________________________________ 18

### 1. Descriptive Statistics

### I. INTRODUCTION
- Daily emissions in early April 2020 fell by about 17 percent compared with 2019 levels.
- Recent studies forecast a reduction in annual emissions of about 4 to 7 percent in 2020, and that emissions will rebound in 2021.
- UNEP (2019) estimates that emissions need to decline by 2.7 percent annually in order to reach the 2°C goal by 2030.
- The COVID-19 crisis may both impede and accelerate the transition to a low-carbon economy:
  - Impediments: Heightened economic uncertainty, corporate balance sheet vulnerabilities, reduced investment and research in long-horizon, capital-intensive green projects, and rescue packages/subsidies that support firms or activities not compatible with long-term climate mitigation.
  - Accelerants: Structural shifts in consumer and investor preferences toward environmentally friendly goods and services; increased awareness of long-term disaster prevention facilitating green policy measures such as carbon taxes and “green recovery” packages.
- The large drop in oil prices coincident with the COVID-19 pandemic can affect firms’ incentives and financial constraints in theoretically ambiguous ways:
  - Lower oil prices may relax firms’ financial constraints but reduce incentives to improve energy efficiency and invest in renewables by making such investments relatively less profitable.
  - Alternatively, lower oil prices can hurt the profitability of the oil sector, reducing investment and production in fossil fuels and easing competition for clean energy firms.
- The net effect of an oil price shock depends on whether the shock is demand-driven or supply-driven:
  - Demand-driven negative oil price shocks (e.g., those reducing economic activity and oil demand) could lead to lower corporate environmental performance due to delayed investments and tighter financial conditions.
  - Oil supply shocks that lower oil prices can increase global economic activity, relax firms’ financial constraints, and allow firms to improve environmental performance.

- Key research questions addressed:
  1. How has the COVID-19 crisis affected green investments so far?
  2. What can be learned from past economic crises about firm behavior regarding the greening of the economy?
  3. What do developments in the oil market imply about firms’ environmental performance?

- Core empirical findings preview:
  - The COVID-19 crisis has not led to a sustained decline in green financing so far; flows into sustainable funds and performance of sustainable assets have been robust.
  - Tighter financial constraints and economic downturns are associated with weaker environmental performance and lower levels of green investments by firms.
  - Predominantly demand-driven negative oil price shocks (as during COVID-19) are likely detrimental for firms’ environmental performance; oil supply shocks that lower oil prices are associated with higher environmental performance.
  - Adverse financial and economic shocks that limit activity and tighten financial constraints could reverse progress in corporate environmental performance by several years.
  - Policy implication: public policies and recovery packages that boost green investments are warranted. Fostering sustainable finance through better disclosures, development of green taxonomies, and product standardization may help mobilize green investments.

### II. THE COVID-19 CRISIS AND FINANCING THE ENERGY TRANSITION
- Green corporate bond issuance:
  - Trended up over the past decade; declined in March 2020 during market turmoil but stayed within the range of historical deviations.
  - Issuance picked up thereafter, with the share of green bonds in total corporate bond issuance more than doubling between March and June 2020.
- Syndicated loan market:
  - Loans to firms with an above-median environmental performance score have increased over the past decade relative to loans to firms with a below-median score.
  - Lending to both types of firms dropped slightly in the first quarter of 2020.
- Investment funds:
  - Sustainable and environmental-focused funds, especially fixed-income funds, continued to attract investment throughout the crisis with only a small drop in aggregate inflows in some asset classes.
  - A possible driver: relatively high returns that green investments experienced during the crisis.
- Overall assessment:
  - The impact of the COVID-19 crisis on environmental finance has been modest and short-lived to date.
  - Given the persistence and severity of the shock—decline in output, potential scarring, and heightened uncertainty—it is uncertain whether these trends will continue and what the overall impact will be on firms’ environmental performance and their contribution to global mitigation efforts.
  - The paper therefore examines firm environmental performance during previous episodes of financial and economic stress to draw implications for the pandemic context.

### III. LESSONS FROM PAST ECONOMIC CRISES FOR THE ENERGY TRANSITION DURING THE COVID-19 CRISIS
#### A. Financial Constraints and Firms’ Environmental Performance
- Empirical specification (baseline):
  - Environmental score equation: EE_{i,s,c,t} = α_s + γ_c + δ_t + β_1 FINCONSTR_{i,t−1} + μ′ X_{i,t−1} + ε_{i,s,c,t}
  - EE_{i,s,c,t}: environmental score for firm i in sector s, country c, time t.
  - α_s, γ_c, δ_t: sector, country, and time (year)-fixed effects.
  - X_{i,t−1}: firm-level controls (e.g., logarithm of total assets, earnings before interest and taxes).
  - FINCONSTR_{i,t−1}: firm-level financial constraint measure (several alternative definitions used).

- Financial constraint measures used:
  - Firm size: logarithm of firm’s total assets (large firms expected to be less constrained). In estimations the sign is reversed so higher values indicate smaller firms.
  - Dividends: dummy = 1 if firm does not pay dividends (considered financially constrained), 0 otherwise.
  - Ratings: dummy = 1 if a firm with positive debt-to-asset ratio is not rated according to Standard and Poor’s (indicating constrained), 0 otherwise.
  - Interest Coverage Ratio (ICR): dummy = 1 if earnings before interest and taxes are below interest expenses (indicating constrained), 0 otherwise.
  - KZ score: dummy = 1 if Kaplan-Zingales score is above the median (aggregate measure of financial constraints), 0 otherwise.

- Data coverage and variables:
  - Sample period: 2002 to 2019.
  - Sample composition: about 7,000 listed firms, 69 industries (Global Industry Classification Standard), 62 economies.
  - Data frequency: annual.
  - Environmental scores source and construction:
    - Obtained from Refinitiv.
    - Based on 68 metrics across three environmental categories: Resource use, Emissions, and Innovation.
    - Category scores calculated using a rank scoring methodology evaluating firms’ environmental performance relative to all other firms each year.
    - Overall environmental scores calculated from a weighted average of category scores where weights vary by industry.
    - Proprietary environmental aggregate scores range between 0 (low performance) and 100 (high performance).

- Key estimation results (Table 2, panel A):
  - Tighter financial constraints are associated with worse environmental performance.
  - Specific magnitudes:
    - Environmental performance falls by 10 points when firm size drops from the median to the 25th percentile of the firm size distribution.
    - When a firm does not pay dividends, its environmental score is 4 points lower.
    - When a firm is not rated, its environmental score is 3 points lower.
- Caveats on firm-level environmental, social, and corporate governance data:
  - Data cover only publicly listed firms; results may not extend to unlisted small- and medium-sized enterprises.
  - Lack of standardization and transparency across data providers; scores from different providers may capture different aspects of environmental performance.
  - Some scores are self-reported by firms; accuracy may vary across the sample.

- Synthesis of findings:
  - Tighter financial constraints and economic downturns are associated with weaker environmental performance and lower green investment by firms.
  - Oil market shocks have ambiguous effects: demand-driven negative shocks (like COVID-19) likely harmful to environmental performance; supply-driven shocks that lower oil prices can relax constraints and improve environmental performance.
  - Policy implication reiterated: public policies and recovery packages that boost green investments are warranted.
  - Measures to foster sustainable finance: better disclosures, development of green taxonomies, and product standardization to mobilize green investments.

*Italic source attribution line.*

### Annex A.1.

### Annex A.1.

### A. Robustness of environmental score estimates and main cross-sectional findings
- Using a logarithm transformation of the environmental scores in the estimations leads to qualitatively unchanged results compared with using the raw environmental scores.
- Comparative differences in environmental scores by financial characteristics:
  - The environmental score is 1 point lower when a firm’s interest coverage ratio is below 1.
  - The environmental score is 1 point lower when the Kaplan-Zingales index is above the median of the sample distribution.
  - Firms that do not pay dividends and unrated firms have environmental scores that are multiple points lower than dividend-paying and rated firms (text indicates larger point differences in surrounding discussion).
- Sub-category results related to climate change (emissions and resource use):
  - For a firm that does not pay dividends, its emissions and resource use scores drop by 6 points.
  - For a firm that is not rated, its emissions and resource use scores drop by 3 and 4 points, respectively, compared with 3 points for the overall environmental score.
- Notes on fixed effects and controls:
  - Firm fixed effects and country-time fixed effects are included to control for time-invariant firm characteristics and macroeconomic conditions; original conclusions are robust to these changes (see columns (6) and (7) of panel A, Table 2).
  - Results are robust to alternative definitions of financial constraint variables (e.g., firm size by total assets median cutoff, KZ score continuous, long-term issuer rating definitions, firm age), and to using a balanced panel starting from 2010.

### B. Financial constraints and firms’ investments in green technologies
- Hypothesis: Financial constraints reduce firms’ investments in green technologies and intangibles needed to improve environmental performance.
- Empirical approach: Probit model for the probability that a firm undertakes environmental investments P(I_{i,s,c,t}=1) = Փ(α_s + γ_c + δ_t + β_1 CONSTRAINT_{i,t-1} + μ′ X_{i,t-1}), where Փ is the cumulative normal and I_{i,s,c,t} is a binary indicator of environmental investment.
- Firm-level findings (Table 2, panel B):
  - The probability that a firm will make an environmental investment falls by 6 percentage points when firm size drops from the median to the 25th percentile of the firm size distribution.
  - A firm unable to pay dividends has a 3 percentage points lower likelihood of making a green investment.
- Robustness checks:
  - Alternative definitions of financial constraint variables, replacement of country fixed effects by climate policy indicators, inclusion of macroeconomic and financial control variables (lagged country-specific output gaps, lagged price of oil (logarithm of the WTI), lagged VIX), and use of a balanced panel all leave original conclusions robust.

### C. Firms’ environmental performance and macro-financial shocks
- Shocks analyzed:
  - Global financial stress shock proxied by the Chicago Board Options Exchange Volatility Index (VIX).
  - Real economic activity shock defined as the change in the annual output gap obtained from the World Economic Outlook database.
- Empirical model for dynamic responses of environmental scores: E_{i,s,c,t+h} = α_{s}^h + γ_{c}^h + β^h CONSTRAINT_{c,t}^{...} + δ′^h FIRM_CONSTRAINT_{i,t-1} + γ′^h MACRO_CONTROLS_{t-1} + ε_{i,s,c,t+h}.
- Key results:
  - A sudden jump in the VIX comparable to that observed during the first half of 2020 would lead to a persistent drop in firms’ environmental performance by up to 5 points, with the pre-shock performance level not attained for at least three years after the shock.
  - Absent policy actions and behavioral changes, this would imply that average corporate environmental performance would return to the levels that prevailed in 2006.
  - Interaction with financial constraints:
    - The adverse effect of global financial shocks on environmental performance is magnified when firms are financially constrained (Table 4, panel A).
    - For firms not paying dividends or for unrated firms in 2019, the global financial stress shock observed thus far in 2020 is estimated to lower environmental performance by 1 and 2 additional points, respectively, compared with dividend-paying or rated firms.
    - Findings are robust to inclusion of firm fixed effects.
  - Output-gap shock magnitude and effects:
    - A large decline in the output gap (10 percentage points, about 50 percent larger than that observed in the Group of Seven (G7) economies during the global financial crisis) would lead to a 3 point decline in firms’ environmental performance in the medium term.
    - Firms’ carbon intensity (total carbon emissions relative to revenue) could increase by up to 8.5 percent in the medium term after such a decline in the output gap.
- Controls and additional macro variables:
  - Firm-level controls include the logarithm of total assets and earnings before interest and taxes.
  - Macroeconomic controls include the price of oil (logarithm of the WTI), country-specific output gaps, and the VIX.

### D. Firms’ environmental performance and oil-market shocks
- Oil shock identification:
  - Oil supply and oil demand shocks are obtained from the structural VAR model of Baumeister and Hamilton (2019), using the median of the posterior distribution and aggregating monthly values to an annual frequency by averaging monthly values over the calendar year.
  - The oil inventory demand (speculative demand) shock is excluded from this analysis.
- Findings on oil-price-source heterogeneity (Table 4, panel B):
  - When oil prices fall due to demand-side factors, environmental corporate performance has historically been weaker.
  - When oil prices decline due to an oil supply shock, environmental performance of firms has generally improved.
  - The COVID-19-induced oil price shock is largely a demand-driven shock (as evident from the decomposition of the oil price shock in March and April 2020); therefore, firms’ environmental performance is likely to suffer significantly.
- Robustness:
  - Including the world economic activity shocks from Baumeister and Hamilton (2019) does not qualitatively change the results.

### Overall synthesis of empirical evidence
- Tighter financial constraints are associated with weaker corporate environmental performance.
- Adverse global financial and output shocks that increase uncertainty and amplify firms’ financial constraints weigh significantly on firms’ environmental performance.
- A reduction in oil prices amid a decline in global economic activity is unlikely on its own to lift corporate environmental performance.
- Absent strong supportive policy actions, tighter financial constraints and weaker economic activity related to the COVID-19 crisis may act as a drag on firms’ environmental performance in the future.

### Policy implications and recommendations
- Public policies and green recovery packages to support firms’ environmental performance during the COVID-19 crisis are warranted.
- To alleviate firms’ financial constraints and aid green investment, key policy measures include support for the sustainable finance sector, specifically:
  - Better disclosure standards.
  - Development of green taxonomies.
  - Product standardization.

*Source: Annex A.1., wpiea2021089-print-pdf*

### 1. Change in Daily CO

### 1. Change in Daily CO

### Changes in daily CO2 emissions and carbon prices
- Panel 1: Reduction in daily CO2 emissions in 2020 compared with 2019 mean levels (Percent).
- Panel 2: Price of futures contracts on carbon emission allowances traded on the Intercontinental Exchange (Euros per metric ton).
- Note: The European Union Emissions Trading System (EU ETS) was subject to several changes in regulation over the sample period that may have affected the price level.
- Sources cited: Global Carbon Project; Refinitiv Datastream; and IMF staff calculations.

### COVID-19 crisis and green investments (Figure 2)
- Panel 1: Green Corporate Bond to Total Corporate Bond Issuance and Total Green Corporate Bond Issuance (global green corporate bond issues).
- Panel 2: Total Amount of Syndicated Loans (billions of US dollars) to Firms with Environmental Scores Higher than Median and Firms with Environmental Scores Lower than Median.
- Panel 3: Sustainable and Environmental Fund Flows as a Share of Total Fund Size, 2003: Q1–2020: Q1 (Percent). MSCI ACWI = Morgan Stanley Capital International All Country World Index.
- Panel 4: Cumulative Returns of Green and Conventional Equity Market Indices (Percent).
- Sources cited: Bloomberg Finance L.P.; Dealogic; Morningstar; Refinitiv Datastream; and IMF staff calculations.

### Environmental performance and financial-market volatility (Figure 3)
- Response of Environmental Score to a VIX Shock (Index).
- Shock definition: a 16.3-point increase in the VIX, corresponding to the difference in the average value of the VIX in 2020 (using data up to July 31, 2020) relative to the average value in 2019.
- Methodology: Local projection approach from firm-level panel regressions that include firm-level controls, country-specific output gaps, the price of oil, and country and industry fixed effects.
- Presentation: Solid line denotes response to the 16.3-point increase; dashed lines denote 90 percent confidence intervals.
- Sources cited: Refinitiv Datastream; Standard & Poor’s; and IMF staff calculations.

### Economic shocks and environmental performance (Figure 4)
- Panel 1: Response of Environmental Score (y-axis) over Time (x-axis) to a fall in the output gap (Index).
- Panel 2: Response of the Logarithm of Total CO2 Emissions Relative to Revenues (y-axis) over Time (x-axis) to a fall in the output gap (Percent).
- Shock scaling: Real economic activity shock is scaled as a 10 percentage points drop in the output gap.
- Regression controls: Firm-level controls (log of total assets, earnings, and a dividend dummy variable); the price of oil (log WTI); the Chicago Board Options Exchange Volatility Index (VIX); and country and sector fixed effects.
- Presentation: Dashed lines represent 90 percent confidence interval.
- Visual time axis label: 0 1 2 3 Years with y-axis ticks from -4.0 to -0.5 indicating negative responses in the plotted index/percent measures.
- Sources cited: Refinitiv Datastream; and IMF staff calculations.

### Tables: descriptive statistics and regression analyses
- Table 1. Descriptive Statistics
  - Environmental score: Refinitiv Asset4 Environmental Pillar Score (weighted average relative rating based on reported environmental information and three environmental categories: Emissions, Resource Use, and Innovation).
  - Environmental Investments Initiatives: binary variable answering “Does the company report on making proactive environmental investments or expenditures to reduce future risks or increase future opportunities?”.
  - Definitions and variable descriptions follow the authors’ estimates and Refinitiv/Standard & Poor’s conventions.
- Table 2. Environmental Performance and Financial Constraints
  - Reports regression estimates of environmental scores (Panel A) or environmental investment decisions (Panel B) on measures of financial constraints and control variables.
  - Data sources: Refinitiv Datastream and Standard and Poor’s.
  - “Size” is the log of total assets, with sign reversed so higher values indicate smaller firms.
  - All right-hand side variables are lagged by one year.
  - Control variables: firm size (log of total assets) and firm profitability (earnings before interest expense and income taxes).
  - Standard errors clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, or 10% level, respectively.
- Table 3. Alternative Environmental Performance Measures and Financial Constraints
  - Reports regression estimates of resource use scores (Panel A) or emissions score (Panel B) on measures of financial constraints and control variables.
  - Data and variable definitions as in Table 2.
  - Same lagging, controls, clustering, and significance notation as Table 2.
- Table 4. Oil Market-shocks and Environmental Performance
  - Reports regression estimates of environmental scores on the VIX (Panel A) and oil-market shocks (Panel B), and control variables.
  - Estimation frequency: annual. Estimation sample extends from 2002 to 2019.
  - Control variables: domestic output gap, log of the price of oil (WTI), firm size (log of total assets), firm profitability (earnings before interest expense and income taxes), and a dummy for firms that do not pay dividends.
  - In Panel B, oil-market shocks correspond to an increase in the real price of oil.
  - Standard errors clustered at the firm level. ***, **, * indicate significance at the 1%, 5%, or 10% level, respectively.
  - See notes to Table 1 for additional data details.

### Robustness checks and supplementary figures (Appendix excerpts)
- Appendix Table A.1. Country-Level Data Sources
  - Macroeconomic and financial variables: Exchange Rate (Refinitiv Datastream); Oil Price (Spot WTI; Haver Analytics); Output Gap (IMF, World Economic Outlook); Real GDP (IMF, WEO); Short-Term Nominal Interest Rate (IMF, WEO); VIX (Refinitiv Datastream); U.S. Consumer Price Index (Federal Reserve Bank of St. Louis); World Oil Production (U.S. Energy Information Administration); Global Economic Activity (Updated series from Baumeister and Hamilton (2019)); Global Oil Inventories (U.S. Energy Information Administration).
  - Firm-level variables: Cash and Short-Term Investments; Cash Dividends; Date of Incorporation; Debt-to-Asset Ratio; Dividends per Share; EBIT; Interest Coverage Ratio; Market Capitalization; Operating Income Before Depreciation and Amortization; Ratings (Standard & Poor’s); Total Assets. All sourced to Refinitiv Datastream unless otherwise noted.
  - Other indicators: Carbon Dioxide (CO2) Emission (The Global Carbon Project); Carbon Price (Refinitiv Datastream); Climate Change Commercial Risk/Opportunities (Refinitiv Datastream); Coverage of National Carbon Pricing Schemes (IMF and the World Bank); Emissions Category Score (Refinitiv Datastream); Enforcement of Environmental Regulations (World Economic Forum); Environmental Investments Initiatives (Refinitiv Datastream); Environmental Score (Refinitiv Datastream); Environmental Policy Stringency Index (Organisation for Economic Co-operation and Development); Stringency of Environmental Regulations (World Economic Forum); Total CO2 Equivalent Emissions to Revenues (Refinitiv Datastream).
- Appendix Table A.2. Environmental Performance and Financial Constraints (Robustness Checks)
  - Reports regression estimates using a balanced panel from 2010 (Panel A) and alternative measures of financial constraints and control variables (Panel B).
  - Alternative measures include continuous and dummy ratings, firm age, size dummies, continuous KZ score, dividend payout dummy as a measure of financial constraints, and replacement of country fixed effects with OECD environmental policy stringency index or World Economic Forum indices.
  - All right-hand side variables lagged by one year; control variables and clustering as in main tables. ***, **, * indicate significance at 1%, 5%, or 10% levels.

### Online annex figure
- Online Annex Figure 1: Historical Decompositions of the Oil Price (Percent).
  - Solid line: actual change in the real price of oil.
  - Bars: median estimate of the historical contribution of structural shocks of the energy market VAR from Baumeister and Hamilton (2019) to the price of oil.
  - “Other factors” include the oil inventory demand shocks and the unexplained component.
  - Sources cited: Federal Reserve Bank of St. Louis; U.S. Energy Information Administration; Updated data from Baumeister and Hamilton (2019); and IMF staff calculations.

*Source: wpiea2021089-print-pdf — IMF staff figures, tables, and appendix material as provided in the content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021089-print-pdf.pdf_