## wpiea2025131-print-pdf

## Source details

**Canonical URL:** [wpiea2025131-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf.json)

---

### 2.1 Trade and Economic Growth
- Overview:
  - Trade contributes to economic growth via specialization and comparative advantage.
  - Export expansion is associated with improved economic performance in low and middle income countries and can stimulate growth in African contexts with outward-looking strategies.
  - The relationship between trade openness and economic growth in Sub-Saharan Africa (SSA) is not strictly linear.
- Key empirical citations:
  - Frankel and Romer (1999); Kavoussi (1984); Onafowora and Owoye (1998); Zahonogo (2016).
- Environmental consequences of trade-driven growth:
  - Over-exploitation of natural resources.
  - Increased carbon emissions.
  - Increased pollution.
  - Deforestation, soil erosion, and loss of biodiversity.
- Policy implication for SSA:
  - Balance promoting trade-led growth with measures to limit environmental degradation, including encouraging energy efficiency and enhancing use of clean energy technologies.

### 3.2 Trade and sustainable Trade
- Definitions of "sustainable trade":
  - Broader: 224 environmental goods per UN Comtrade (connected to environmental protection or modified to be more "environmentally friendly").
  - Narrower (policy-actionable): 125 traded low carbon goods per the IMF Climate Change Dashboard (include wind turbines, solar panels, biomass systems, carbon capture equipment).
- Trends (summary):
  - Median UN Comtrade environmental goods as % of GDP: very low relative to general trade; consistent after 2008.
  - Median TLCT as % of GDP: comparable pattern; slowdown after 2008.
  - Mean UN Comtrade value higher than TLCT mean, possibly due to larger goods set.
- Regression design (dependent / main independent variables):
  - 1: Global Footprint Network Ecological Footprint — UN Comtrade Green Trade
  - 2: Global Footprint Network Ecological Footprint — Trade in Low Carbon Technology
  - 3: Global Carbon Footprint Net CO2 Emissions — UN Comtrade Green Trade
  - 4: Global Carbon Footprint Net CO2 Emissions — Trade in Low Carbon Technology
  - 5: Global Burden of Disease PM2.5 — UN Comtrade Green Trade
  - 6: Global Burden of Disease PM2.5 — Trade in Low Carbon Technology
  - Each analysis repeated with World Bank general trade (% of GDP) as the main independent variable for comparison; sustainable trade independent variables sourced from the IMF Climate Change Dashboard.
  - Total regressions: 8 (the 6 listed analyses plus repetitions replacing the sustainable-trade shock with general trade).
- OECD vs SSA patterns:
  - OECD sample shows much higher median values of traded environmental goods than the SSA sample.
  - SSA trades environmental goods at much lower rates than OECD.
- Main control variables (sourced from World Bank Development Indicators):
  - GDP per capita (% annual growth), in constant LCU.
  - Population growth (POP) (% annual growth).
  - Urbanization growth (URB) (% annual growth).
- Country sample and data frequency:
  - Sample years: 2001-2020, yearly frequency.
  - Country sample (17): Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cote d’Ivoire, Eswatini, Madagascar, Mauritius, Namibia, Niger, Rwanda, Senegal, South Africa, Tanzania, Uganda, Zimbabwe.
  - In sustainable trade analyses, control for trade as % of GDP; in general trade analyses, control for sustainable trade by calculating ratio of sustainable goods trade to total trade (IMF database).
- Summary statistics highlights (as reported):
  - Analyses 1 & 5 (UN Comtrade):
    - N = 320 for UN Comtrade and WB Trade rows; GFN EF and GBD PM2.5 N = 320 and 320.
    - UN Comtrade: Mean 1.59, St. Dev. 0.92, Min 0.26, Med 1.38, Max 5.75
    - WB Trade: Mean 62.30, St. Dev. 27.92, Min 20.96, Med 55.05, Max 175.80
    - GFN EF: Mean 0.75, St. Dev. 3.72, Min −14.08, Med 0.73, Max 22.84
    - GBD PM2.5: Mean −0.07, St. Dev. 3.38, Min −15.16, Med 0.01, Max 20.18
    - Trade Ratio: Mean 0.20, St. Dev. 0.19, Min 0.02, Med 0.14, Max 1.26
  - Analysis 3 (UN Comtrade; GCB Net CO2):
    - N = 260 for UN Comtrade and WB Trade; GCB Net CO2 N = 260
    - UN Comtrade: Mean 1.58, St. Dev. 0.92, Min 0.26, Med 1.40, Max 5.75
    - WB Trade: Mean 62.63, St. Dev. 24.67, Min 27.49, Med 55.60, Max 129.78
    - GCB Net CO2: Mean 37.68, St. Dev. 46.30, Min −43.84, Med 31.00, Max 258.22
  - Analyses 2 & 6 (TLCT):
    - N = 320 where listed
    - TLCT: Mean 0.83, St. Dev. 0.51, Min 0.12, Med 0.71, Max 3.69
    - WB Trade: Mean 63.09, St. Dev. 27.20, Min 27.49, Med 55.29, Max 175.80
    - GFN EF: Mean 0.77, St. Dev. 3.62, Min −14.08, Med 0.73, Max 22.84
    - GBD PM2.5: Mean −0.08, St. Dev. 3.52, Min −15.16, Med −0.08, Max 20.18
    - Trade Ratio: Mean 0.11, St. Dev. 0.10, Min 0.01, Med 0.09, Max 0.86
  - Analysis 4 (TLCT; GCB Net CO2):
    - N = 280 for TLCT and WB Trade; GCB Net CO2 N = 280
    - TLCT: Mean 0.84, St. Dev. 0.52, Min 0.12, Med 0.72, Max 3.69
    - WB Trade: Mean 61.43, St. Dev. 24.20, Min 27.49, Med 55.39, Max 129.78
    - GCB Net CO2: Mean 36.09, St. Dev. 45.04, Min −43.84, Med 28.94, Max 258.22
  - Controls (summary across analyses):
    - GDP % growth: Mean 1.55, St. Dev. 4.22, Min −18.32, Med 2.01, Max 19.94
    - POP % growth: Mean 3.52, St. Dev. 1.78, Min −0.15, Med 3.67, Max 7.60
    - URB % growth: Mean 3.54, St. Dev. 1.65, Min −0.15, Med 3.68, Max 7.60
- Descriptive conclusions:
  - Sustainable trade as % of GDP has much lower mean levels than general trade for the SSA sample.
  - Mean values of GFN and GCB environmental outcomes are positive (resource use increasing; CO2 emissions in trade focused on imports).
  - Mean GBD PM2.5 is negative and close to zero, indicating little average change in air pollution in the sample.

### 3.5 Local Projections Analysis
- Methodology:
  - Linear local projection analysis as in Jorda (2005) and Jorda and Taylor (2024).
  - Shock: 1-standard deviation increase in the independent (shock) variable (trade in environmental goods or TLCT).
  - Shock variable lagged one year and used as an instrumental variable.
  - Two lags estimated (AIC criteria).
  - Horizons h = 0, 1, 2, 3, 4, 5; cumulative impulse responses estimated.
  - Confidence intervals at 67% (primary) and robustness at 90%.
  - Newey-West standard errors used.
- Regression equations (as presented):
  - Equation 1:
    logp_{i,t+h} − logp_{i,t−1} = α^h_i + Σ_{n=1}^2 β_{t−n} p_{i,t−n} + γ^h φ_{i,t} + λ^h_{i,n} X^h_{i,n} + η_i + ν_t + ε^h_{i,t+h}
    - p = ecological footprint or PM2.5; φ = sustainable trade or trade variable; X = controls (GDP, population, urbanization); sustainable trade regressions control for general trade % GDP; general trade regressions control for sustainable trade % of general trade; η and ν are country and time fixed effects.
  - Equation 2:
    p_{i,t+h} = α^h_i + Σ_{n=1}^2 β_{t−n} p_{i,t−n} + γ^h φ_{i,t} + λ^h_{i,n} X^h_{i,n} + η_i + ν_t + ε^h_{i,t+h}
    - Used when GCB Net CO2 emissions is dependent variable.
- Key empirical findings:
  - Ecological footprint (Analyses 1 & 2):
    - Trade in environmental goods improves ecological footprint starting from year 2.
    - Most substantial gains ~3.5% in years 3 and 4.
    - TLCT achieves greatest gains (just under 5%) in year 3.
    - Sustainable trade reduces ecological footprint more rapidly and pronouncedly than general trade.
  - Embedded carbon emissions in trade (Analyses 3 & 4):
    - General trade does not significantly reduce embedded carbon emissions short- or medium-term.
    - Trade in environmental goods leads to cumulative reduction in embedded carbon emissions of approximately 60% over five years.
    - Trade in TLCT achieves cumulative reduction of around 100% over five years.
    - Footnote: Botswana and Namibia outliers; removing them decreases estimates to between [−25%,−35%].
  - Air pollution (PM2.5) (Analyses 5 & 6):
    - Sustainable trade reduces air pollution by approximately 1% after 3-4 years.
    - General trade shows minimal and statistically insignificant reductions.
    - By year 5 IRFs become insignificant; TLCT shows similar peak reductions then return to zero.
  - PM2.5-related deaths (GBD):
    - N = 3200, Mean = 0.02, St. Dev. = 0.01, Min. = 0.01, Med. = 0.02, Max. = 0.05.
    - IRFs: no immediate significant relationship; after 3-5 years percent of deaths associated with air pollution decreases, peaking at 0.003-0.006% after five years.
- OECD sample comparison:
  - OECD: higher sustainable and general trade; lower resource use, net CO2 emissions in trade, and PM2.5; lower GDP, population, urbanization growth.
  - Ecological footprint (OECD): sustainable trade peak decreases at 3 years but smaller than SSA; general trade increases resource use over time (opposite to SSA).
  - Net CO2 in trade (OECD): sustainable trade effect near zero and insignificant; general trade increases net CO2 emissions by 0.3% after five years.
  - PM2.5 (OECD): very small and mostly insignificant effects.
  - Interpretation: smaller marginal benefits in OECD due to prior progress ("catch-up effect").
- Welfare and GDP per capita trade-offs (Section 6 findings):
  - PM2.5-related deaths: sustainable trade peak reductions 0.003-0.006% after five years.
  - Real GDP per capita growth (IRFs non-cumulative; equation 2 with GDP per capita growth dependent):
    - UN Comtrade SSA sample:
      - One-standard deviation increase in sustainable trade as % of GDP corresponds to a $309.67 million increase.
      - Mean GDP per capita (constant international dollars) = $5,874.31.
      - Increasing sustainable trade by one standard deviation decreases real GDP per capita by $100.44 over five years.
    - TLCT SSA sample:
      - One-standard deviation increase corresponds to $183.456 million increase.
      - Mean GDP per capita = $6,086.80.
      - Real GDP per capita decreases by $233.73 over five years.
    - General trade (SSA):
      - One-standard deviation increase corresponds to $9.40 billion increase.
      - Corresponds to increase in real GDP per capita by $99.86 over five years.
    - General trade (OECD):
      - One-standard deviation increase corresponds to $620.42 billion increase.
      - Calculates to $534.92 increase in real GDP per capita.
  - Interpretation:
    - A one-standard deviation increase in sustainable trade may reduce per capita GDP by $100-$233 over five years, while a one-standard deviation increase in general trade increases per capita GDP by less than $100 in SSA (and substantially more in OECD).
    - The dollar increase in general trade is about 30 times that of sustainable trade; many estimates are mostly insignificant.
    - Net implication: sustainable trade may reduce air pollution-related deaths and has negative but statistically insignificant effects on real GDP growth in the samples analyzed.
- Summary conclusions:
  - Significant medium-term relationship between sustainable trade and improved environmental outcomes in SSA.
  - Greatest benefits emerge after 3 to 5 years; infrastructure, labor, human capital, and resources needed to realize benefits.
  - Sustainable trade’s environmental impact is greater and faster than general trade in SSA; OECD shows smaller benefits and some detrimental effects of general trade.

### 7.1 Data and Methodology (Sustainable Practices Index - SPI)
- Purpose:
  - Construct novel SPI for SSA to standardize, rank, and compare sustainable trade practices and provide policy insights.
- Four SPI components:
  - Environmental taxes (Environmental taxes % of GDP) — IMF Climate Change Dashboard.
  - Climate change policies (stock variable: total number of climate change policies in-force) — Climate Policy Database.
  - Renewable energy (sum of wind, solar, bio, hydro use) — IMF Climate Change Dashboard; each value standardized 0–100 before aggregation.
  - Carbon stock (carbon stored in forests, soils, oceans, ecosystems) — included to emphasize natural carbon sinks.
- Index construction:
  - Rescale each component to 0–100 (highest = 100; least favorable = 0) following Hinrich IMD Sustainable Trade Index approach.
  - PCA used for weights. Reported PC proportions of variance (Table 5 as presented):
    - 10.381
    - 20.3216
    - 30.2081
    - Total 0.9107
  - Decision: use first 3 principal components to capture about 90% of variance.
  - Interpretation: first component weighs heavily on renewable energy and carbon stock; second on climate change policies and environmental taxes.
  - Final SPI: sum weighted principal components and rescale to 0–100.
- Sample, frequency, limitations:
  - Countries (14): Cabo Verde, Cameroon, Democratic Republic of the Congo, Cote d’Ivoire, Eswatini, Ghana, Kenya, Mali, Niger, Rwanda, Sierra Leone, South Africa, Togo, Uganda.
  - Time span: 2001-2020, annual frequency.
  - Limited overlap with sustainable trade analysis sample; not enough overlap with GCB Net CO2 emissions sample — alternative CO2 series used.
- Alternative CO2 measure:
  - World Bank’s CO2 emissions from Climate Watch Historical GHG Emissions (metric tons per capita).
  - Transformation: log difference of the WB CO2 series.
  - Note: Figure 20 shows median CO2 plummet in 2020 due to COVID production halts.
- Controls and summary statistics (Table 6 values as presented):
  - GPI: N 280, Mean 46.70, St. Dev. 19.34, Min 0, Med 43.36, Max 100
  - GFN EF: N 280, Mean 0.95, St. Dev. 3.54, Min −22.38, Med 0.88, Max 25.67
  - WB CO2: N 280, Mean 0.73, St. Dev. 4.36, Min −21.60, Med 0.59, Max 24.33
  - GBD PM2.5: N 280, Mean −0.15, St. Dev. 4.89, Min −22.98, Med −0.36, Max 22.87
- Empirical strategy:
  - Linear local projection analysis estimating cumulative IRFs from a 1-standard deviation increase in the SPI (equation 1 with dependent variable replaced by SPI).
- SPI cross-sectional 2020 values (Table 7 as reported):
  - Dem. Rep. of Congo — 99.17
  - Cabo Verde — 68.32
  - Cote d’Ivoire — 64.04
  - Cameroon — 62.66
  - Niger — 61.91
  - Togo — 45.22
  - Uganda — 42.33
  - Mali — 39.43
  - Rwanda — 39.09
  - Ghana — 37.78
  - Sierra Leone — 34.77
  - Kenya — 33.09
  - Eswatini — 29.78
  - South Africa — 2.32
- Notable insight:
  - Renewable energy receives high weight; top SPI performers have high renewable energy. South Africa ranks lowest despite highest number of active climate change policies, reflecting weaknesses in other components and declining rank over the sample period.

### Section 6 calculations (reported arithmetic)
- Figure 19a:
  - 0.0092∗33.66 billion = 309.67 million dollars (1a.1)
  - First year: 5874.31∗0.0085 = −49.93 dollars (1a.2)
  - Second year: 5874.31∗0.0070 = 41.12 dollars (1a.3)
  - Third year: 5874.31∗0.0029 = 17.04 dollars (1a.4)
  - Fourth year: 5874.31∗0.0171 = −100.45 dollars (1a.5)
  - Fifth year: 5874.31∗0.0014 = −8.22 dollars (1a.6)
- Figure 19b:
  - 0.0052∗35.28 billion = 183.456 million dollars (2a.1)
  - First year: 6086.80∗0.0142 = −86.43 dollars (2a.2)
  - Second year: 6086.80∗0.0123 = 74.87 dollars (2a.3)
  - Third year: 6086.80∗0.003 = 18.26 dollars (2a.4)
  - Fourth year: 6086.80∗0.0289 = −175.91 dollars (2a.5)
  - Fifth year: 6086.80∗0.0106 = −64.52 dollars (2a.6)
- Figure 19c:
  - 0.2792∗33.66 billion = 9.40 billion dollars (3a.1)
  - First year: 5874.31∗0.0041 = 24.08 dollars (3a.2)
  - Second year: 5874.31∗0.0033 = 19.39 dollars (3a.3)
  - Third year: 5874.31∗0.0034 = 19.97 dollars (3a.4)
  - Fourth year: 5874.31∗0.0037 = 21.73 dollars (3a.5)
  - Fifth year: 5874.31∗0.0025 = 14.69 dollars (3a.6)
- Figure 19d:
  - 0.5523∗1187.58 billion = 620.421 billion dollars (4a.1)
  - First year: 46921.68∗0.0035 = 164.23 dollars (4a.2)
  - Second year: 46921.68∗0.0023 = 107.92 dollars (4a.3)
  - Third year: 46921.68∗0.0013 = 61.00 dollars (4a.4)
  - Fourth year: 46921.68∗0.0017 = 79.77 dollars (4a.5)
  - Fifth year: 46921.68∗0.0026 = 122.00 dollars (4a.6)

*Source: wpiea2025131-print-pdf - 2.1 Trade and Economic Growth, https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf*

### 2.1    Trade and Economic Growth   .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### 2.1    Trade and Economic Growth

### Overview
- Trade is widely recognized as a major contributor to economic growth through specialization and comparative advantage.
- Empirical studies provide evidence that export expansion is associated with improved economic performance in low and middle income countries and can stimulate growth in African contexts with outward-looking strategies.
- The relationship between trade openness and economic growth in Sub-Saharan Africa (SSA) is not strictly linear.

### Key empirical findings cited
- Frankel and Romer (1999): trade as an instrument of economic growth.
- Kavoussi (1984): both low and middle income countries see better economic performance associated with greater export expansion.
- Onafowora and Owoye (1998): export expansion can stimulate economic growth in some African countries.
- Zahonogo (2016): relationship between trade openness and economic growth is not exactly linear in Sub-Saharan Africa.

### Environmental consequences of trade-driven growth
- Increased trade and economic growth can lead to higher levels of production and consumption, often resulting in:
  - over-exploitation of natural resources,
  - increased carbon emissions,
  - increased pollution,
  - deforestation, soil erosion, and loss of biodiversity.

### Findings from literature on sustainable trade and growth
- Ahmed et al (2022): using indicators like sustainable innovation, clean energy, and sustainable trade (as defined by the OECD), finds positive contributions to “sustainable growth” (a variable defined by GDP growth, education, fossil fuel consumption, and CO2 emissions) in South Asian countries.
- Tariq et al (2023): theoretical model with sustainable trade openness (GTO) as a component finds GTO can increase sustainable growth and mitigate carbon emissions.
- Xu (2022): in China, sustainable trade moderates the impact of natural resources on sustainable growth.
- Policy implication emphasized: mitigate effects of climate change while continuing to promote growth.

### Implications for Sub-Saharan Africa
- SSA faces the dual challenge of climate vulnerability and the need for sustainable economic growth.
- Policy emphasis should balance promoting trade-led growth with measures to limit environmental degradation, including encouraging energy efficiency and enhancing use of clean energy technologies.

_italic: Source: wpiea2025131-print-pdf - 2.1 Trade and Economic Growth, https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf_

### 3.2    Trade and sustainable Trade

### 3.2    Trade and sustainable Trade

### Definitions of "sustainable trade"
- Two definitions used:
  - Broader definition: 224 environmental goods per UN Comtrade. These goods are defined as having a connection to environmental protection like pollution and resource management and goods that have been modified to be more "environmentally friendly".
  - Narrower, policy-actionable definition: 125 traded low carbon goods per the IMF Climate Change Dashboard. These are defined as products that produce less pollution than their traditional energy counterparts and include wind turbines, solar panels, biomass systems, and carbon capture equipment.

### Trends in trade of sustainable goods (figures referenced)
- Median values of trade in environmental goods (UN Comtrade definition) as a % of GDP:
  - Very low relative to general trade.
  - Stay consistent after 2008, indicating little increase in trade of sustainable goods in recent years.
- Median values of trade in low carbon technology products (TLCT) as a % of GDP:
  - Comparable pattern to UN Comtrade series.
  - Exhibit a slowdown after 2008.
- Mean value:
  - The mean value of the UN Comtrade definition of sustainable trade is higher than trade in low-carbon technology, possibly due to a larger set of environmental goods classified as “sustainable”.

### Regression design and dependent/independent variables (Table 1)
- Dependent variables and main sustainable-trade independent variables included in regressions:
  - 1: Global Footprint Network Ecological Footprint — UN Comtrade Green Trade
  - 2: Global Footprint Network Ecological Footprint — Trade in Low Carbon Technology
  - 3: Global Carbon Footprint Net CO2 Emissions — UN Comtrade Green Trade
  - 4: Global Carbon Footprint Net CO2 Emissions — Trade in Low Carbon Technology
  - 5: Global Burden of Disease PM2.5 — UN Comtrade Green Trade
  - 6: Global Burden of Disease PM2.5 — Trade in Low Carbon Technology
- Note: Each analysis is repeated with World Bank general trade (% of GDP) as the main independent variable for comparison. The sustainable trade independent variables are sourced from the IMF Climate Change Dashboard.
- Total regressions: 8 (each of the 6 listed analyses plus repetitions replacing sustainable-trade shock with general trade as main shock, yielding a total of 8 regressions as described).

### OECD comparison and geographic patterns
- OECD country sample (37 countries listed in source) shows much higher median values of traded environmental goods than the SSA country sample.
- On a global scale, Sub-Saharan Africa (SSA) trades environmental goods at much lower rates than OECD sample.

### Main control variables (3.3)
- Controls included in analysis (sourced from World Bank Development Indicators):
  - GDP per capita (% annual growth), in constant LCU.
    - Rationale: income and pollution are statistically related (Environmental Kuznets Curve).
  - Population growth (POP) (% annual growth).
    - Rationale: larger populations increase environmental footprint and degradation.
  - Urbanization growth (URB) (% annual growth).
    - Rationale: increased industrialization and urban concentration raise production and emissions.

### Country sample and data frequency (3.4)
- Sample years: 2001-2020, yearly frequency.
- Country sample (17 countries): Benin, Botswana, Burkina Faso, Burundi, Cameroon, Cote d’Ivoire, Eswatini, Madagascar, Mauritius, Namibia, Niger, Rwanda, Senegal, South Africa, Tanzania, Uganda, and Zimbabwe.
- To isolate effects:
  - In sustainable trade analyses, control for trade as % of GDP.
  - In general trade analyses, control for sustainable trade by calculating the ratio of total value of imports and exports of sustainable goods divided by value of total trade in US dollars (total trade sourced from IMF database).
  - This ratio is calculated for each definition of sustainable trade and respective country samples.

### Summary statistics (Table 2 highlights)
- Sample sizes and key descriptive statistics (as reported):
  - Analyses 1 & 5 (UN Comtrade; includes GFN EF and GBD PM2.5)
    - N = 320 for UN Comtrade and WB Trade in those rows; GFN EF and GBD PM2.5 N = 320 and 320 respectively.
    - UN Comtrade: Mean 1.59, St. Dev. 0.92, Min 0.26, Med 1.38, Max 5.75
    - WB Trade: Mean 62.30, St. Dev. 27.92, Min 20.96, Med 55.05, Max 175.80
    - GFN EF: Mean 0.75, St. Dev. 3.72, Min −14.08, Med 0.73, Max 22.84
    - GBD PM2.5: Mean −0.07, St. Dev. 3.38, Min −15.16, Med 0.01, Max 20.18
    - Trade Ratio: Mean 0.20, St. Dev. 0.19, Min 0.02, Med 0.14, Max 1.26
  - Analysis 3 (UN Comtrade; GCB Net CO2)
    - N = 260 for UN Comtrade and WB Trade; GCB Net CO2 N = 260
    - UN Comtrade: Mean 1.58, St. Dev. 0.92, Min 0.26, Med 1.40, Max 5.75
    - WB Trade: Mean 62.63, St. Dev. 24.67, Min 27.49, Med 55.60, Max 129.78
    - GCB Net CO2: Mean 37.68, St. Dev. 46.30, Min −43.84, Med 31.00, Max 258.22
  - Analyses 2 & 6 (TLCT; includes GFN EF and GBD PM2.5)
    - N = 320 where listed
    - TLCT: Mean 0.83, St. Dev. 0.51, Min 0.12, Med 0.71, Max 3.69
    - WB Trade: Mean 63.09, St. Dev. 27.20, Min 27.49, Med 55.29, Max 175.80
    - GFN EF: Mean 0.77, St. Dev. 3.62, Min −14.08, Med 0.73, Max 22.84
    - GBD PM2.5: Mean −0.08, St. Dev. 3.52, Min −15.16, Med −0.08, Max 20.18
    - Trade Ratio: Mean 0.11, St. Dev. 0.10, Min 0.01, Med 0.09, Max 0.86
  - Analysis 4 (TLCT; GCB Net CO2)
    - N = 280 for TLCT and WB Trade; GCB Net CO2 N = 280
    - TLCT: Mean 0.84, St. Dev. 0.52, Min 0.12, Med 0.72, Max 3.69
    - WB Trade: Mean 61.43, St. Dev. 24.20, Min 27.49, Med 55.39, Max 129.78
    - GCB Net CO2: Mean 36.09, St. Dev. 45.04, Min −43.84, Med 28.94, Max 258.22
- Controls (summary across analyses):
  - GDP % growth: Mean 1.55, St. Dev. 4.22, Min −18.32, Med 2.01, Max 19.94
  - POP % growth: Mean 3.52, St. Dev. 1.78, Min −0.15, Med 3.67, Max 7.60
  - URB % growth: Mean 3.54, St. Dev. 1.65, Min −0.15, Med 3.68, Max 7.60

### Descriptive conclusions from the sample
- Both definitions of sustainable trade as % of GDP have much lower mean levels than general trade as % of GDP for the SSA sample, indicating SSA is not trading a lot of “sustainable” goods.
- Mean values of GFN and GCB environmental outcomes are positive, indicating resource use is increasing and CO2 emissions in trade are heavily focused on imports.
- Mean value of GBD PM2.5 is negative and close to zero, indicating that, on average, there is little change in air pollution among the sample.

*Source: wpiea2025131-print-pdf - 3.2 Trade and sustainable Trade*

### 3.5    Local Projections Analysis

### 3.5    Local Projections Analysis

### Methodology and model specification
- The paper uses linear local projection analysis as described by Jorda (2005) and Jorda and Taylor (2024).
- Purpose: time series analysis to analyze medium-term effects of sustainable trade given inertia in adoption of environmental goods and low carbon technology products.
- Rationale: local projections are favorable because many series cannot be assumed stationary; impulse response functions (IRFs) are estimated at each period of interest rather than extrapolated far into the horizon.
- Shock definition: impulse response functions from a 1-standard deviation increase in the independent (shock) variable (either trade in environmental goods or trade in low carbon technology products).
- Endogeneity handling: the shock variable is lagged by one year and used as an instrumental variable.
- Lags: two lags are estimated as established by the AIC criteria.
- Horizons: local projections show a one-standard deviation shock on the φ variable over 5 horizons, h = 0, 1, 2, ... , 5. The results estimate cumulative impulse responses over time.
- Confidence intervals: measured at the 67% level (consistent with Stock and Watson (2005)); estimation was repeated at 90% level confidence interval as a robustness check.
- Standard errors: Newey-West standard errors are used.
- Regression equations (as in source):
  - Equation 1:
    logp_{i,t+h} − logp_{i,t−1} = α^h_i + Σ_{n=1}^2 β_{t−n} p_{i,t−n} + γ^h φ_{i,t} + λ^h_{i,n} X^h_{i,n} + η_i + ν_t + ε^h_{i,t+h}
    - where p indicates the ecological footprint variable or the PM2.5 variable and φ indicates the sustainable trade or trade variable for each country i at time t. X indicates a matrix of control variables including GDP, population, and urbanization. Sustainable trade regressions control for general trade as a percentage of GDP; general trade regressions control for sustainable trade as a percentage of general trade. η and ν indicate country and time fixed effects.
  - Equation 2:
    p_{i,t+h} = α^h_i + Σ_{n=1}^2 β_{t−n} p_{i,t−n} + γ^h φ_{i,t} + λ^h_{i,n} X^h_{i,n} + η_i + ν_t + ε^h_{i,t+h}
    - All specifications are the same as equation 1; equation 2 is used when GCB Net CO2 emissions variable is the dependent variable.

### Key empirical findings: sustainable trade effects on environmental outcomes
- Ecological footprint (Analysis 1 and 2):
  - Trade in environmental goods leads to a significant improvement in the ecological footprint, starting from the second year.
  - The most substantial gains are around 3.5% observed in the third and fourth years.
  - Trade in low-carbon technology products achieves the greatest gains (just under 5%) in the third year.
  - Comparison: trade in environmental goods and low-carbon technology products has a more pronounced and more rapid effect on reducing the ecological footprint than general trade.
  - Conclusion: strong medium-term negative relationship between sustainable trade and ecological footprint.

- Embedded carbon emissions in trade (Analysis 3 and 4):
  - General trade does not significantly influence reduction of carbon emissions embedded in trade, either short-term or medium-term.
  - Trade in environmental goods leads to a cumulative reduction in embedded carbon emissions of approximately 60% over five years.
  - Trade in low-carbon technology products achieves a cumulative reduction of around 100% over five years.
  - Footnote: Botswana and Namibia had high levels of CO2 and were outliers. After removing them from the sample, the estimates decreased to between [−25%,−35%]. (IRF results in appendix referenced in source.)
  - Interpretation: targeted trade in environmentally friendly and low-carbon technologies can critically reduce the carbon intensity of trade in the medium term; general trade lacks this focused impact.

- Air pollution (PM2.5) (Analysis 5 and 6):
  - Sustainable trade leads to a reduction in air pollution by approximately 1% after 3-4 years.
  - General trade shows minimal and statistically insignificant reductions in air pollution.
  - By the fifth year the IRFs become insignificant, indicating sustainable trade’s influence on PM2.5 diminishes beyond the medium term.
  - TLCT-based sustainable trade shows similar peak reductions followed by a gradual return to zero (initial benefits do not sustain beyond the medium term).

- PM2.5-related deaths (Global Burden of Disease) (Figure 18; Table 4 summary statistics)
  - GBD PM2.5 related deaths summary statistics: N = 3200, Mean = 0.02, St. Dev. = 0.01, Min. = 0.01, Med. = 0.02, Max. = 0.05.
  - IRFs: no immediate significant relationship, but after 3-5 years the percent of deaths associated with air pollution decreases, peaking at 0.003-0.006% after five years.
  - Interpretation: absolute levels are small, but sustainable trade can ultimately reduce air pollution-related deaths.

### OECD country sample comparison (Section 5)
- Motivation: re-evaluate main analyses using OECD countries to contextualize SSA results.
- Descriptive differences (OECD vs SSA):
  - OECD countries have higher levels of sustainable trade and general trade; lower levels of resource use, net CO2 emissions in trade, and PM2.5; lower rates of GDP, population, and urbanization growth.
- Main comparative findings:
  - Ecological footprint (OECD sample):
    - Sustainable trade shows lag effects with peak decreases at three years, but decreases are smaller than in SSA.
    - General trade shows a consistent increase in resource use over time (increasing trade increases resource use).
    - Opposite effect compared to SSA sample.
  - Net CO2 emissions embedded in trade (OECD sample):
    - Effect of sustainable trade is near zero and insignificant.
    - Increasing general trade increases net CO2 emissions by 0.3% after five years.
  - PM2.5 (OECD sample):
    - Very small and mostly insignificant effects of sustainable trade and general trade on air pollution.
  - Summary: effects of increasing sustainable trade on environmental outcomes are much smaller in OECD countries; increasing trade tends to be detrimental to environment for OECD sample.
  - Interpretation: a "catch-up effect"—OECD countries have already progressed in sustainable trade and climate policy, so further increases yield smaller marginal benefits than in SSA.

### Implications on economic welfare and GDP per capita trade-offs (Section 6)
- Main logic: environmental benefits from sustainable trade should translate into welfare improvements related to pollution; also assess possible trade-offs with economic growth.
- PM2.5-related deaths: sustainable trade peak reductions in deaths 0.003-0.006% after five years (referenced above).
- Real GDP per capita growth (IRFs non-cumulative; regressions use equation 2 specification with GDP per capita growth as dependent variable; controls same except GDP):
  - For the UN Comtrade SSA sample:
    - A one-standard deviation increase in sustainable trade as a % of GDP corresponds to a $309.67 million increase.
    - Using mean GDP per capita (constant international dollars) of $5,874.31 for the sample, increasing sustainable trade by one standard deviation decreases real GDP per capita by $100.44 over the five year horizon.
  - For the TLCT variable (SSA sample):
    - A one-standard deviation increase in sustainable trade as a % of GDP corresponds to a $183.456 million increase.
    - Mean GDP per capita for this country sample is $6,086.80.
    - Real GDP per capita decreases by $233.73 over the five year horizon.
  - For general trade (SSA sample):
    - A one-standard deviation increase in trade as a % of GDP corresponds to a $9.40 billion increase.
    - This corresponds to an increase in real GDP per capita by $99.86 over the five year horizon.
  - For general trade (OECD sample):
    - A one-standard deviation increase in trade as a % of GDP corresponds to a $620.42 billion increase.
    - This calculates to a $534.92 increase in real GDP per capita.
- Interpretation of cost-benefit comparison:
  - A one-standard deviation increase in sustainable trade may result in a $100-$233 decrease in per capita GDP over five years, while increasing general trade by one standard deviation increases per capita GDP by less than $100 in SSA (and substantially more in OECD).
  - Since the increase in trade is about 30 times more than the increase in sustainable trade (in dollar terms), the estimated effect shows sustainable trade increases would have more negative effect on GDP growth than an equivalent-standard-deviation increase in general trade would have a positive effect—however, many estimates are mostly insignificant.
  - Net implication: sustainable trade may reduce air pollution-related deaths and has negative but statistically insignificant effects on real GDP growth in the samples analyzed.

### Summary conclusions (from the analysis)
- There is a significant medium-term relationship between sustainable trade and improved environmental outcomes in the SSA sample.
- Greatest environmental benefits emerge after 3 to 5 years, consistent with time needed to build infrastructure and integrate imported sustainable goods into practical use.
- For benefits to materialize countries need adequate labor, human capital, and resources to effectively utilize imported sustainable technologies, potentially extending implementation timelines.
- The positive impact of sustainable trade on the environment is considerably greater and faster than that of general trade in the SSA sample.
- OECD comparison shows smaller environmental benefits and some detrimental effects of general trade, suggesting region-specific dynamics and a possible "catch-up effect" in SSA.

*Source: wpiea2025131-print-pdf - 3.5    Local Projections Analysis*

### 7.1    Data and Methodology

### 7.1    Data and Methodology

### Purpose and overview
- Construct a novel Sustainable Practices Index (SPI) for Sub-Saharan Africa (SSA) to standardize, rank, and compare sustainable trade practices across countries in SSA and provide actionable policy insights.
- SPI intended as a region-specific benchmark to map country performance across key components and identify leaders/laggards under a common framework.
- Four key variables chosen for the SPI due to relevance to environmental outcomes and data completeness for a sub-sample of SSA:
  - environmental taxes
  - climate change policies
  - renewable energy adoption
  - carbon stock

### Variable definitions and data sources
- Environmental taxes:
  - Measured as Environmental taxes (% of GDP).
  - Source: IMF Climate Change Dashboard.
  - Defined as a tax imposed on a physical unit (or a proxy of it) of something that has a proven, specific, negative impact on the environment.
- Climate change policies:
  - Measured as a stock variable capturing the total number of climate change policies in-force.
  - Source: Climate Policy Database.
- Renewable energy:
  - Measured as the sum of renewable energy use from wind, solar, bio, and hydro energy.
  - Source: IMF Climate Change Dashboard.
  - Each renewable energy value is standardized between 0 and 100 before aggregation.
- Carbon stock:
  - Amount of carbon stored in forests, soils, oceans, and other ecosystems.
  - Included to emphasize preserving and enhancing natural carbon sinks.

### Index construction methodology
- Rescaling:
  - Each component value is rescaled to lie between 0 and 100.
  - Highest value receives a score of 100; least favorable value scored as 0.
  - Methodology follows the Hinrich IMD Sustainable Trade Index approach to enable comparability across differently measured indicators.
- Weighting:
  - Principal component analysis (PCA) used to obtain weights for each variable.
  - Table 5 reports the proportion of variance for each principal component as presented in source:
    - PCProportion of Variance
    - 10.381
    - 20.3216
    - 30.2081
    - Total0.9107
  - Decision: use the first 3 principal components to capture about 90% of total variance.
  - Interpretation of components:
    - First principal component weighs heavily on renewable energy and carbon stock.
    - Second principal component weighs more heavily on climate change policies and environmental taxes.
- Final index aggregation:
  - SPI constructed by adding each weighted principal component and rescaling final values between 0 and 100.

### Sample, frequency, and data limitations
- Countries included in the SPI (14):
  - Cabo Verde, Cameroon, Democratic Republic of the Congo, Cote d’Ivoire, Eswatini, Ghana, Kenya, Mali, Niger, Rwanda, Sierra Leone, South Africa, Togo, and Uganda.
- Time span and frequency:
  - Data runs from 2001-2020 at an annual frequency.
- Data limitations and coverage notes:
  - Country sample limits data availability; limited overlap between the SPI country sample and the sample used for sustainable trade analysis due to indicator coverage in SSA.
  - Not enough overlap with the GCB Net CO2 emissions country sample; an alternative CO2 series was used (see next subsection).

### Alternative CO2 measure and computation
- Replacement CO2 series:
  - Use World Bank’s CO2 emissions from Climate Watch Historical GHG Emissions, defined as amount of CO2 in metric tons per capita.
  - Rationale: provides a more general look at pollution values rather than CO2 production in trade.
- Transformation used in analysis:
  - Apply the log difference of the WB CO2 series.
- Empirical note:
  - Figure 20 (as referenced) shows the median value of CO2 emissions among the country sample each year and highlights a plummet in 2020 percentage change due to production halts caused by COVID.

### Controls, summary statistics, and empirical specification
- Controls used (same as previous analysis):
  - GDP per capita growth
  - Population growth
  - Urbanization growth
- Summary statistics:
  - The source reports summary statistics for the SPI, dependent variables, and controls in an internal table (Table 6 as presented). The values provided in the source appear as:
    - Table 6:  SPI Summary Statistics
    - StatisticNMeanSt.  Dev.MinMedMax
    - GPI28046.7019.34043.36100
    - GFN EF2800.953.54−22.380.8825.67
    - WB CO22800.734.36−21.600.5924.33
    - GBD PM2.5280−0.154.89−22.98−0.3622.87
- Empirical strategy:
  - Use linear local projection analysis to estimate cumulative impulse response functions from a 1-standard deviation increase in the SPI.
  - The estimation models equation 1 with the dependent variable replaced by the SPI.
- SPI cross-sectional 2020 values (Table 7 as reported):
  - Dem.  Rep.  of Congo — 99.17
  - Cabo Verde — 68.32
  - Cote d’Ivoire — 64.04
  - Cameroon — 62.66
  - Niger — 61.91
  - Togo — 45.22
  - Uganda — 42.33
  - Mali — 39.43
  - Rwanda — 39.09
  - Ghana — 37.78
  - Sierra Leone — 34.77
  - Kenya — 33.09
  - Eswatini — 29.78
  - South Africa — 2.32
- Notable descriptive insight:
  - Renewable energy receives high weight in the SPI; top contributors (Democratic Republic of the Congo, Cabo Verde, Cote d’Ivoire, Cameroon, Niger) have high renewable energy amounts. South Africa ranks lowest in the SPI despite having the highest number of active climate change policies, reflecting weaknesses in other components and a declining rank over the data sample period.

*Source: IMF Working Paper content — 7.1 Data and Methodology (content unit supplied).*

### Section 6 calculations

### Section 6 calculations

### Figure 19a
- 0.0092∗33.66 billion = 309.67 million dollars (1a.1)
- First year: 5874.31∗0.0085 = −49.93 dollars (1a.2)
- Second year: 5874.31∗0.0070 = 41.12 dollars (1a.3)
- Third year: 5874.31∗0.0029 = 17.04 dollars (1a.4)
- Fourth year: 5874.31∗0.0171 = −100.45 dollars (1a.5)
- Fifth year: 5874.31∗0.0014 = −8.22 dollars (1a.6)

### Figure 19b
- 0.0052∗35.28 billion = 183.456 million dollars (2a.1)
- First year: 6086.80∗0.0142 = −86.43 dollars (2a.2)
- Second year: 6086.80∗0.0123 = 74.87 dollars (2a.3)
- Third year: 6086.80∗0.003 = 18.26 dollars (2a.4)
- Fourth year: 6086.80∗0.0289 = −175.91 dollars (2a.5)
- Fifth year: 6086.80∗0.0106 = −64.52 dollars (2a.6)

### Figure 19c
- 0.2792∗33.66 billion = 9.40 billion dollars (3a.1)
- First year: 5874.31∗0.0041 = 24.08 dollars (3a.2)
- Second year: 5874.31∗0.0033 = 19.39 dollars (3a.3)
- Third year: 5874.31∗0.0034 = 19.97 dollars (3a.4)
- Fourth year: 5874.31∗0.0037 = 21.73 dollars (3a.5)
- Fifth year: 5874.31∗0.0025 = 14.69 dollars (3a.6)

### Figure 19d
- 0.5523∗1187.58 billion = 620.421 billion dollars (4a.1)
- First year: 46921.68∗0.0035 = 164.23 dollars (4a.2)
- Second year: 46921.68∗0.0023 = 107.92 dollars (4a.3)
- Third year: 46921.68∗0.0013 = 61.00 dollars (4a.4)
- Fourth year: 46921.68∗0.0017 = 79.77 dollars (4a.5)
- Fifth year: 46921.68∗0.0026 = 122.00 dollars (4a.6)

*Source: wpiea2025131-print-pdf - Section 6 calculations*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025131-print-pdf.pdf_