## Dirty Dance: Tourism and Environment — wpiea2022178-print-pdf

## Source details

**Canonical URL:** [Dirty Dance: Tourism and Environment — wpiea2022178-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022178-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022178-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022178-print-pdf.pdf.json)

---

### Abstract and central result
- Tourism was one of the fastest-growing sectors before the COVID-19 pandemic, accounting for about 10 percent of global GDP.
- In the Caribbean, tourism “accounts for as much as 90 percent of GDP and 80 percent of energy-related carbon dioxide (CO2) emissions.”
- Main empirical finding: “An increase of 10 percent in international tourist arrivals is associated with an increase of as much as 8 percent in CO2 emissions,” after controlling for economic, institutional and social factors.
- Policy implication: managing tourism sustainably requires reducing its environmental impact and curbing excessive dependency on fossil fuel-based energy consumption.

### Climate context and policy tension
- Global warming and projections:
  - Global warming already increased by “about 1.1 degrees Celsius (°C) compared with the preindustrial average.”
  - If GHG emissions continue at current rates, global warming is projected to reach “4-6°C by 2100.”
- Small island state vulnerability and disaster costs:
  - Hurricane Ivan for Grenada in 2004: “148 percent of GDP.”
  - Hurricane Maria for Dominica in 2017: “260 percent” of GDP.
- Policy tension and implication:
  - Trade-offs between mitigation and adaptation in small island states — potentially high marginal returns to adaptation and smaller marginal returns to mitigation — but mitigation and adaptation can be complementary.
  - Urgent investment need: “small island states in the Caribbean also urgently need to invest in climate-resilient infrastructure against stronger and more destructive hurricanes and rising sea levels.”

### Channels linking tourism to CO2 emissions
- Tourism increases CO2 emissions through:
  - carbon-intensive energy production and consumption tied to accommodation, transportation and other tourist activities;
  - consumption of material resources in tourism-related sectors;
  - changes in land use associated with tourism-related investments.
- Energy supply and consumption are “determined to a great extent by tourism-related activities” in the Caribbean.

### Data and sample
- Sample: 15 Caribbean countries (Antigua and Barbuda, Aruba, Bahamas, Barbados, Belize, Dominica, Dominican Republic, Grenada, Guyana, Haiti, Jamaica, St Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, and Suriname).
- Period: unbalanced panel “over the period 1960–2019.”
- Dependent variable: territory-based CO2 emissions in metric tons per capita.
  - Territorial CO2 emissions include “the use of coal, oil and gas (combustion and industrial processes), the process of gas flaring, and the manufacture of cement.”
  - Includes CO2 emissions from domestic flights, but “not from bunker fuels associated with international aviation and maritime operations.”
- Main explanatory variable: international tourist arrivals (number of international tourist arrivals).

### Summary statistics (exact values from Table 1)
- CO2 emissions per capita (obs. 880): Mean 4.55; Std. Dev. 6.81; Min. 0.02; Max. 49.26.
- Tourism arrivals (obs. 403): Mean 1,255,450; Std. Dev. 1,533,166; Min. 43,000; Max. 7,600,000.
- Real GDP per capita (obs. 843): Mean 8,494; Std. Dev. 7,627; Min. 666; Max. 32,237.
- Real GDP growth (obs. 827): Mean 1.9; Std. Dev. 4.9; Min. -19.1; Max. 27.3.
- Trade openness (obs. 617): Mean 100.4; Std. Dev. 35.1; Min. 31.1; Max. 275.0.
- Government effectiveness (obs. 492): Mean 2.1; Std. Dev. 1.1; Min. 0.0; Max. 3.0.
- Population (obs. 728): Mean 1,374,275; Std. Dev. 2,717,864; Min. 40,259; Max. 11,000,000.
- Urbanization (obs. 728): Mean 44.8; Std. Dev. 16.6; Min. 15.6; Max. 83.1.
- Historical change: CO2 emissions per capita “increased by over 260 percent from 1.40 tons in 1960 to 5.06 tons in 2019.”
- Tourism growth: foreign visitors increased “from 8.6 million in 1995 to 19.8 million in 2019,” with growth rates of “5.3 percent in the second half of the 1990s,” “2.9 percent in the 2000s,” and “3.6 percent in the 2010s.”

### Correlations and multicollinearity diagnostics (exact values)
- Selected correlation coefficients (from Table 2):
  - CO2 emissions per capita — Tourist arrivals: 0.41
  - CO2 emissions per capita — Real GDP per capita: 0.89
  - Tourist arrivals — Real GDP per capita: 0.50
  - Government effectiveness — Real GDP per capita: 0.82
- Variable Inflation Factor (VIF) results (from Table 3):
  - Real GDP per capita: 3.97
  - Government effectiveness: 3.56
  - Tourist arrivals: 2.58
  - Trade openness: 2.38
  - Real GDP growth: 1.02
- Conclusion: “none of the independent variables has a VIF above 10” and pairwise correlations do not exceed 0.7 for key variables, suggesting multicollinearity is not a major concern.

### Empirical model and estimation
- Baseline dynamic specification:
  - ln(CO2_c,t) = β1 + β2 ln(CO2_c,t−1) + β3 ln(tourist_c,t) + β4 X_c,t + η_c + μ_t + ε_c,t
  - X_c,t includes controls: logarithm of real GDP per capita and its quadratic term (log demeaned), real GDP growth, trade openness, population, share of urban population, and government effectiveness.
  - η_c denotes time-invariant country-specific effects; μ_t denotes time effects controlling for common shocks.
  - Robust standard errors are clustered at the country level.
- Estimation approach:
  - Static fixed effects model used as reference; dynamic persistence captured with System GMM (Arellano and Bover (1995); Blundell and Bond (1998)).
  - One-step system GMM applied; instrument proliferation addressed following Roodman (2009).
  - Identification checks: AR(1) and AR(2) p-values reported; Hansen J-test p-values indicate validity of internal instruments.

### Baseline estimation results and interpretation (exact coefficient values)
- Key coefficient estimates from Table 4 (baseline estimations):
  - CO2 emissions per capita t-1: 0.865***, 0.737***, 0.947***, 0.737*** [0.044][0.136][0.010][0.113]
  - Tourist arrivals: 0.077***, 0.066*** [0.030][0.025]
  - Real GDP per capita: 0.363***, 1.047**, 0.272***, 1.030*** [0.045][0.231][0.124][0.227]
  - Real GDP per capita squared: -0.016**, -0.059**, -0.013**, -0.063** [0.026][0.033][0.007][0.013]
  - Real GDP growth: 0.003***, 0.002***, 0.004***, 0.001** [0.002][0.006][0.002][0.001]
  - Trade openness: 0.001***, 0.048 [0.040][0.039]
  - Government effectiveness: -0.076, -0.056*** [0.024][0.017]
  - Number of observations: 80, 71, 72, 80, 71, 72
  - Number of countries: 15
  - Country FE: Yes; Year FE: Yes
  - Adj R2: 0.89, 0.71
  - AR1 p-value: 0.002, 0.002
  - AR2 p-value: 0.647, 0.225
  - Hansen J-test p-value: 0.245, 0.188
- Interpretation of tourism coefficient:
  - Static specification: tourism coefficient = 0.077 (statistically and economically highly significant).
  - Dynamic (system GMM) specification: tourism coefficient = 0.066 (positive and statistically significant).
  - Elasticity interpretation: an increase of 10 percent in international visitors is associated with an increase of about 7 percent in CO2 emissions (0.066–0.077 elasticity range reported). Conclusion reiterates “as much as 8 percent” in CO2 emissions for a 10 percent increase in visitors.

### Robustness checks and sensitivity (exact values from Table 5 and Appendix A1)
- Robustness strategies and outcomes:
  - Excluding lagged dependent variable: results consistent with larger coefficients for explanatory variables.
  - Truncating sample at 5th and 95th percentiles: results remain intact with marginal changes.
  - Adding controls (population, urbanization): expected signs but not statistically significant.
  - Nonlinear tourism effects: tourist arrivals^2 indicates nonlinearity with higher tourism levels causing further emissions increase.
  - Additional sensitivity specifications: (i) without quadratic term of real GDP per capita; (ii) tourists per capita; (iii) excluding real GDP growth; (iv) excluding trade openness and government effectiveness.
- Selected robustness coefficient estimates (Table 5):
  - CO2 emissions per capita t-1: 0.849***, 0.697***, 0.729*** [0.066][0.202][0.182]
  - Tourist arrivals: 0.235***, 0.048***, 0.063***, 0.075*** [0.088][0.031][0.035][0.032]
  - Tourist arrivals^2: 0.005* [0.020]
  - Real GDP per capita: 2.061***, 0.516***, 1.828***, 0.918*** [0.135][0.120][0.238][0.232]
  - Real GDP per capita squared: -0.123***, -0.031*, -0.110**, -0.057** [0.035][0.066][0.043][0.035]
  - Real GDP growth: 0.002***, 0.001*, 0.001**, 0.001** [0.005][0.002][0.002][0.002]
  - Trade openness: -0.001***, -0.001***, -0.001***, -0.001*** [0.002][0.001][0.002][0.003]
  - Government effectiveness: -0.125, -0.062, -0.125, -0.051 [0.083][0.041][0.083][0.083]
  - Population: 0.380 [0.427]
  - Urbanization: -0.002 [0.006]
  - Number of observations: 172, 151, 149, 172
  - Number of countries: 15
  - Adj R2: 0.47, 0.79, 0.70, 0.71
- Appendix Table A1 (selected exact entries):
  - CO2 emissions per capita t-1: (1) 0.739*** [0.133]; (2) 0.698*** [0.191]; (3) 0.738*** [0.134]; (4) 0.648*** [0.086]
  - Tourist arrivals: (1) 0.065*** [0.031]; (2) 0.061*** [0.037]; (3) 0.066*** [0.034]; (4) 0.075*** [0.037]
  - Real GDP per capita: (1) 0.958*** [0.121]; (2) 0.934*** [0.176]; (3) 0.972*** [0.113]; (4) 0.881*** [0.081]
  - Real GDP growth: (1) 0.001*** [0.002]; (2) 0.001*** [0.001]; (3) 0.001*** [0.002]
  - Trade openness: (1) 0.001*** [0.010]; (2) 0.001*** [0.010]; (3) 0.001*** [0.006]
  - Government effectiveness: (1) -0.062 [0.021]; (2) -0.055 [0.018]; (3) -0.060 [0.020]
  - Number of observations: (1) 172; (2) 149; (3) 172; (4) 384
  - Adj R2: (1) 0.70; (2) 0.69; (3) 0.69; (4) 0.62
- Robustness conclusion: positive relationship between tourism and environmental pollution remains unchanged across checks; evidence of diseconomies of scale in tourism due to fossil-fuel dependence and geographic conditions.

### Policy recommendations (exact instruments and targets cited)
- Overarching aim: reconcile tourism-driven growth with environmental quality; promote sustainable and inclusive development and financial resilience.
- Incentivize decarbonization throughout the economy:
  - Introduce a broad-based carbon tax set to gradually increase to US$50 per metric ton of CO2 by 2030.
  - Consider "feebates"—fees on high-emission products combined with rebates for low-emission products in carbon-intensive sectors such as agriculture, tourism, and transportation.
  - Use carbon tax revenue to reduce other taxes, develop insurance instruments and social safety nets, and fund public investment in sustainable and resilient infrastructure.
  - Note: IMF proposes differentiated carbon tax ranges of $75, $50 and $25 per metric ton of CO2 for advanced, high-income emerging markets and low-income emerging markets, respectively (mentioned for context).
- Develop a low-carbon energy sector:
  - Energy sector accounts for more than 70 percent of CO2 emissions in the Caribbean due to dependence on imported fossil fuel for electricity generation.
  - Policy actions: change energy supply composition from fossil fuels to mainly renewables; expand storage capacity and smart transmission grids; decentralize electrification technologies.
- Electrify mobility and transportation:
  - Transportation accounts for more than one-third of oil consumption and energy-related CO2 emissions in the Caribbean.
  - Electrification benefits: diversify fuel portfolio, reduce fossil dependence, lower total cost of ownership, improve price stability, strengthen energy independence, improve environment.
  - Vehicle-to-grid approaches can help integrate more renewable energy.
- Strengthen land-use practices and building regulations:
  - Restore natural capital (forests, coastal ecosystems) to increase resilience to extreme weather and slow-onset changes such as desertification and sea level rise.
  - Smarter urbanization: zoning, building codes, expand green areas, improve energy efficiency in buildings.
- Sector-specific fiscal instruments for tourism:
  - A carbon tax on foreign visitors earmarked for climate change mitigation to decouple tourism growth from CO2 emissions.
  - Use feebates and fees to promote energy savings in commercial and residential buildings and reduce deforestation.
- Climate-resilient infrastructure and mitigation–adaptation linkages:
  - Urgent investment in climate-resilient infrastructure against stronger hurricanes and rising sea levels.
  - Recognize upfront fiscal cost of adaptation but emphasize long-run socioeconomic benefits and higher expected returns to private investment when resilience is built.
  - Mitigation and adaptation should be complementary.

### Overall empirical conclusion
- International tourism has a statistically and economically significant effect on CO2 emissions in the homogeneous panel of 15 tourism-dependent Caribbean countries over 1960–2019.
- After controls, an increase of 10 percent in international visitors is associated with an increase of as much as 8 percent in CO2 emissions.
- Tourism contributes to environmental degradation via carbon-intensive energy production, material resource consumption in accommodation and transportation, and land-use changes for tourism investments.
- Policy priorities: decarbonize the energy sector, electrify transportation, strengthen land-use and building regulations, implement carbon pricing and sectoral feebates, and invest in climate-resilient infrastructure.

*wpiea2022178-print-pdf - Section 1–3.*

### Section 1

### Dirty Dance: Tourism and Environment — Section 1

### Abstract and central result
- Tourism was one of the fastest-growing sectors before the COVID-19 pandemic, accounting for about 10 percent of global GDP.
- In the Caribbean, tourism “accounts for as much as 90 percent of GDP and 80 percent of energy-related carbon dioxide (CO2) emissions.”
- Main empirical finding: “An increase of 10 percent in international tourist arrivals is associated with an increase of as much as 8 percent in CO2 emissions,” after controlling for economic, institutional and social factors.
- Policy implication highlighted: managing tourism sustainably requires reducing its environmental impact and curbing excessive dependency on fossil fuel-based energy consumption.

### Climate context and policy tension
- Global warming already increased by “about 1.1 degrees Celsius (°C) compared with the preindustrial average.”
- If GHG emissions continue at current rates, global warming is projected to reach “4-6°C by 2100.”
- Small island states face high vulnerability: illustrative disaster costs cited include Hurricane Ivan for Grenada in 2004 amounting to “148 percent of GDP” and Hurricane Maria for Dominica in 2017 reaching “260 percent” of GDP.
- Policy tension: trade-offs between mitigation and adaptation in small island states — potentially high marginal returns to adaptation and smaller marginal returns to mitigation — but mitigation and adaptation can be complementary.
- Urgent investment need: “small island states in the Caribbean also urgently need to invest in climate-resilient infrastructure against stronger and more destructive hurricanes and rising sea levels.”

### Channels linking tourism to CO2 emissions
- Tourism increases CO2 emissions through:
  - carbon-intensive energy production and consumption tied to accommodation, transportation and other tourist activities;
  - consumption of material resources in tourism-related sectors;
  - changes in land use associated with tourism-related investments.
- Energy supply and consumption are “determined to a great extent by tourism-related activities” in the Caribbean.

### Data and sample
- Sample: 15 Caribbean countries (Antigua and Barbuda, Aruba, Bahamas, Barbados, Belize, Dominica, Dominican Republic, Grenada, Guyana, Haiti, Jamaica, St Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, and Suriname).
- Period: unbalanced panel “over the period 1960–2019.”
- Dependent variable: territory-based CO2 emissions in metric tons per capita.
  - Territorial CO2 emissions include “the use of coal, oil and gas (combustion and industrial processes), the process of gas flaring, and the manufacture of cement.”
  - Includes CO2 emissions from domestic flights, but “not from bunker fuels associated with international aviation and maritime operations.”
- Main explanatory variable: international tourist arrivals (number of international tourist arrivals).

### Summary statistics (exact values from Table 1)
- CO2 emissions per capita (obs. 880): Mean 4.55; Std. Dev. 6.81; Min. 0.02; Max. 49.26.
- Tourism arrivals (obs. 403): Mean 1,255,450; Std. Dev. 1,533,166; Min. 43,000; Max. 7,600,000.
- Real GDP per capita (obs. 843): Mean 8,494; Std. Dev. 7,627; Min. 666; Max. 32,237.
- Real GDP growth (obs. 827): Mean 1.9; Std. Dev. 4.9; Min. -19.1; Max. 27.3.
- Trade openness (obs. 617): Mean 100.4; Std. Dev. 35.1; Min. 31.1; Max. 275.0.
- Government effectiveness (obs. 492): Mean 2.1; Std. Dev. 1.1; Min. 0.0; Max. 3.0.
- Population (obs. 728): Mean 1,374,275; Std. Dev. 2,717,864; Min. 40,259; Max. 11,000,000.
- Urbanization (obs. 728): Mean 44.8; Std. Dev. 16.6; Min. 15.6; Max. 83.1.
- Historical change: CO2 emissions per capita “increased by over 260 percent from 1.40 tons in 1960 to 5.06 tons in 2019.”
- Tourism growth: foreign visitors increased “from 8.6 million in 1995 to 19.8 million in 2019,” with growth rates of “5.3 percent in the second half of the 1990s,” “2.9 percent in the 2000s,” and “3.6 percent in the 2010s.”

### Correlations and multicollinearity diagnostics (exact values)
- Selected correlation coefficients (from Table 2):
  - CO2 emissions per capita — Tourist arrivals: 0.41
  - CO2 emissions per capita — Real GDP per capita: 0.89
  - Tourist arrivals — Real GDP per capita: 0.50
  - Government effectiveness — Real GDP per capita: 0.82
- Variable Inflation Factor (VIF) results (from Table 3):
  - Real GDP per capita: 3.97
  - Government effectiveness: 3.56
  - Tourist arrivals: 2.58
  - Trade openness: 2.38
  - Real GDP growth: 1.02
- Conclusion: “none of the independent variables has a VIF above 10” and pairwise correlations do not exceed 0.7 for key variables, suggesting multicollinearity is not a major concern.

### Empirical model and estimation
- Baseline dynamic specification (as presented):
  - ln(CO2_c,t) = β1 + β2 ln(CO2_c,t−1) + β3 ln(tourist_c,t) + β4 X_c,t + η_c + μ_t + ε_c,t
  - Where X_c,t includes controls: logarithm of real GDP per capita and its quadratic term (log demeaned to reduce collinearity), real GDP growth, trade openness, population, share of urban population, and government effectiveness.
  - η_c denotes time-invariant country-specific effects; μ_t denotes time effects controlling for common shocks.
  - Robust standard errors are clustered at the country level to account for heteroskedasticity.

### Contribution to literature and robustness
- Contribution: provides empirical analysis of the impact of international tourism on CO2 emissions in a relatively homogeneous panel of small island states in the Caribbean with long-span data (1960–2019) and robustness checks across alternative specifications and methodologies.
- Situates findings within three literature strands: Environmental Kuznets Curve debate; tourism-growth nexus; and tourism-environment links.

### Policy recommendations (from Section I)
- Caribbean countries could reduce CO2 emissions by:
  - (i) decarbonizing the energy sector with very high shares of renewables;
  - (ii) electrifying mobility and transportation;
  - (iii) developing sustainable land-use practices and smarter urbanization.
- Emphasizes complementarity of mitigation and adaptation and urgent need for climate-resilient infrastructure in small island states.

*Source: wpiea2022178-print-pdf - Section 1.*

### Section 2

### wpiea2022178-print-pdf - Section 2

### Methodology: dynamic modelling and system GMM
- Static fixed effects model used as a point of reference; dynamic persistence in CO2 emissions captured with a dynamic model.
- System GMM estimator (Arellano and Bover (1995); Blundell and Bond (1998)) employed to account for inertia in CO2 emissions and potential bias from lagged dependent variable with country-specific effects.
- System GMM approach: two sets of equations
  - first differences of endogenous and pre-determined variables instrumented by lags of their own levels;
  - levels of endogenous and pre-determined variables instrumented by lags of their own first differences.
- One-step system GMM applied to avoid downward-biased standard errors that affect two-step variant in small samples.
- Instrument proliferation addressed following Roodman (2009) to balance information extraction and risk of over-identification.
- Identification checks:
  - AR(1) and AR(2) reported as p-values for first- and second-order autocorrelated disturbances in the first-differenced equation; high AR(1) but no significant AR(2).
  - Hansen J-test applied to assess validity of internal instruments; reported p-values indicate validity in the dynamic model.

### Baseline estimation results (Table 4)
- Model fit and general patterns:
  - Baseline empirical findings indicate a strong fit with expected signs across specifications.
  - Elasticity of CO2 emissions per capita w.r.t. real GDP per capita is positive; quadratic term of real GDP per capita is negative — consistent with the EKC hypothesis (inverted U-shaped pattern).
  - Real GDP growth has a strong positive relationship with CO2 emissions per capita.
  - Trade openness has a statistically significant negative effect on CO2 emissions per capita.
  - Government effectiveness coefficient is negative (implying governments can lower CO2 emissions), though not always statistically significant in static models.
- Key coefficient estimates from Table 4 (baseline estimations):
  - CO2 emissions per capita t-1: 0.865***, 0.737***, 0.947***, 0.737*** [0.044][0.136][0.010][0.113]
  - Tourist arrivals: 0.077***, 0.066*** [0.030][0.025]
  - Real GDP per capita: 0.363***, 1.047**, 0.272***, 1.030*** [0.045][0.231][0.124][0.227]
  - Real GDP per capita squared: -0.016**, -0.059**, -0.013**, -0.063** [0.026][0.033][0.007][0.013]
  - Real GDP growth: 0.003***, 0.002***, 0.004***, 0.001** [0.002][0.006][0.002][0.001]
  - Trade openness: 0.001***, 0.048 [0.040][0.039]
  - Government effectiveness: -0.076, -0.056*** [0.024][0.017]
  - Number of observations: 80, 71, 72, 80, 71, 72 (Table shows "80 71 72 80 71 72" formatting across columns)
  - Number of countries: 15 15 15 15
  - Country FE: Yes; Year FE: Yes
  - Adj R2: 0.89, 0.71
  - AR1 p-value: 0.002, 0.002
  - AR2 p-value: 0.647, 0.225
  - Hansen J-test p-value: 0.245, 0.188
  - Estimation methods shown: System GMM and Fixed Effects
- Interpretation of tourism coefficient:
  - Static specification: estimated coefficient on international tourist arrivals is 0.077 (statistically and economically highly significant).
  - Dynamic (system GMM) specification: estimated coefficient on international tourist arrivals is 0.066 (positive and statistically significant).
  - Interpretation: an increase of 10 percent in the number of international visitors is associated with an increase of about 7 percent in CO2 emissions (0.066–0.077 elasticity range reported).

### Robustness checks (Table 5 and other checks)
- Robustness strategies:
  - Exclude lagged dependent variable to avoid potential bias — results consistent with larger coefficients for explanatory variables.
  - Truncate sample at the 5th and 95th percentiles to exclude outliers — results remain intact with marginal coefficient changes.
  - Introduce additional control variables: population and share of urban population — expected signs but not statistically significant at conventional levels.
  - Explore nonlinear effects of international tourism by adding a quadratic term of tourist arrivals — evidence of nonlinearity with higher levels of tourism causing further emissions increase.
  - Additional sensitivity checks (referred to) include specifications: (i) without quadratic term of real GDP per capita; (ii) number of international tourists per capita; (iii) excluding real GDP growth; (iv) excluding trade openness and government effectiveness.
- Key coefficient estimates from Table 5 (robustness):
  - CO2 emissions per capita t-1: 0.849***, 0.697***, 0.729*** [0.066][0.202][0.182]
  - Tourist arrivals: 0.235***, 0.048***, 0.063***, 0.075*** [0.088][0.031][0.035][0.032]
  - Tourist arrivals^2: 0.005* [0.020]
  - Real GDP per capita: 2.061***, 0.516***, 1.828***, 0.918*** [0.135][0.120][0.238][0.232]
  - Real GDP per capita squared: -0.123***, -0.031*, -0.110**, -0.057** [0.035][0.066][0.043][0.035]
  - Real GDP growth: 0.002***, 0.001*, 0.001**, 0.001** [0.005][0.002][0.002][0.002]
  - Trade openness: -0.001***, -0.001***, -0.001***, -0.001*** [0.002][0.001][0.002][0.003]
  - Government effectiveness: -0.125, -0.062, -0.125, -0.051 [0.083][0.041][0.083][0.083]
  - Population: 0.380 [0.427]
  - Urbanization: -0.002 [0.006]
  - Number of observations: 172, 151, 149, 172
  - Number of countries: 15 15 15 15
  - Country FE: Yes; Year FE: Yes
  - Adj R2: 0.47, 0.79, 0.70, 0.71
- Conclusion from robustness checks: positive relationship between tourism and environmental pollution remains unchanged across checks; some evidence of diseconomies of scale in tourism due to fossil-fuel dependence and geographic conditions.

### Conclusion and policy recommendations
- Empirical conclusion:
  - International tourism has a statistically and economically significant effect on CO2 emissions in a homogenous panel of 15 tourism-dependent Caribbean countries over 1960–2019.
  - After controls, an increase of 10 percent in international visitors is associated with an increase of as much as 8 percent in CO2 emissions (paper reports "as much as 8 percent" in the conclusion, consistent with earlier elasticity estimates).
  - Tourism contributes to environmental degradation via carbon-intensive energy production, material resource consumption in accommodation and transportation, and land-use changes for tourism investments.
- Policy priorities to reconcile growth and environmental quality:
  - Incentivize decarbonization throughout the economy:
    - Introduce a broad-based carbon tax set to gradually increase to US$50 per metric ton of CO2 by 2030.
    - Consider "feebates"—fees on high-emission products combined with rebates for low-emission products in carbon-intensive sectors such as agriculture, tourism, and transportation.
    - Use carbon tax revenue to reduce other taxes, develop insurance instruments and social safety nets, and fund public investment in sustainable and resilient infrastructure.
    - Note: IMF proposes differentiated carbon tax ranges of $75, $50 and $25 per metric ton of CO2 for advanced, high-income emerging markets and low-income emerging markets, respectively (mentioned for context in the text).
  - Develop a low-carbon energy sector:
    - Energy sector accounts for more than 70 percent of CO2 emissions in the Caribbean due to dependence on imported fossil fuel for electricity generation.
    - Change energy supply composition from fossil fuels to mainly renewables, expand storage capacity and smart transmission grids, decentralize electrification technologies.
  - Electrify mobility and transportation:
    - Transportation accounts for more than one-third of oil consumption and energy-related CO2 emissions in the Caribbean.
    - Electrification offers benefits: diversify fuel portfolio, reduce fossil dependence, lower total cost of ownership, improve price stability, strengthen energy independence, improve environment.
    - Vehicle-to-grid approaches can help integrate more renewable energy by managing intermittency.
  - Strengthen land-use practices and building regulations:
    - Restore natural capital (forests, coastal ecosystems) to increase resilience to extreme weather and slow-onset changes such as desertification and sea level rise.
  - Decarbonize the tourism industry and curb excessive dependence on fossil fuel-based energy consumption to reduce tourism's carbon footprint (noting tourism's carbon footprint may constitute as much as 80 percent of CO2 emissions in some small island countries).
- Overall policy aim: optimize welfare-enhancing economic growth while accounting for climate change and environmental quality, promoting sustainable and inclusive economic development and financial resilience.

*Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022178-print-pdf.pdf*

### Section 3

### wpiea2022178-print-pdf - Section 3

### Transition to a low-carbon urban and tourism economy
- Transition requires smarter urbanization with zoning practices and building codes in tourism-dependent areas designed to:
  - reduce vulnerability to climate change,
  - expand green areas,
  - strengthen CO2 emission management in new projects,
  - improve energy efficiency in buildings.
- Policymakers can implement fees and feebates to:
  - promote energy savings in the stock of commercial and residential buildings,
  - reduce deforestation in the forestry and land-use sectors.

### Decarbonizing the tourism industry — findings and policy instruments
- Shift to more sustainable forms of tourism combined with stronger protection and restoration of ecosystems can:
  - act as a powerful climate solution,
  - create millions of jobs,
  - diversify the tourism sector.
- Sector-specific fiscal instruments suggested:
  - a carbon tax on foreign visitors earmarked for climate change mitigation to decouple tourism growth from CO2 emissions,
  - such instruments would help manage the carbon footprint of international tourism and support greater access to international green finance for climate change adaptation and mitigation.

### Climate-resilient infrastructure and mitigation–adaptation linkages
- Small island states in the Caribbean urgently need investment in climate-resilient infrastructure against:
  - stronger and more destructive hurricanes,
  - rising sea levels (existential threats).
- Fiscal trade-offs and long-run returns:
  - Upfront fiscal cost of climate change adaptation is acknowledged, but inaction would have even greater macro-fiscal cost for generations.
  - Investing in structural resilience would yield long-run socioeconomic benefits, reduce damages from natural disasters, and increase expected returns to private investment in tourism and other sectors.
- Mitigation and adaptation should be complementary rather than strict trade-offs.
- Example interlinkage: energy supply and consumption in the Caribbean is determined to a great extent by tourism-related activities; balancing mitigation and adaptation strategies provides diverse ecological, economic, and social benefits.

### Appendix Table A1 — Additional Sensitivity Checks (dependent variable: CO2 emissions per capita)
Columns correspond to alternative specifications: (1) without the quadratic term of real GDP per capita; (2) with international tourists per capita; (3) excluding real GDP growth; (4) excluding trade and governance that have limited number of observations.

- CO2 emissions per capita t-1:
  - (1) 0.739*** [0.133]
  - (2) 0.698*** [0.191]
  - (3) 0.738*** [0.134]
  - (4) 0.648*** [0.086]
- Tourist arrivals:
  - (1) 0.065*** [0.031]
  - (2) 0.061*** [0.037]
  - (3) 0.066*** [0.034]
  - (4) 0.075*** [0.037]
- Real GDP per capita:
  - (1) 0.958*** [0.121]
  - (2) 0.934*** [0.176]
  - (3) 0.972*** [0.
113]
  - (4) 0.881*** [0.081]
- Real GDP per capita squared:
  - (reported but not shown in table excerpt)
- Real GDP growth:
  - (1) 0.001*** [0.002]
  - (2) 0.001*** [0.001]
  - (3) 0.001*** [0.002]
- Trade openness:
  - (1) 0.001*** [0.010]
  - (2) 0.001*** [0.010]
  - (3) 0.001*** [0.006]
- Government effectiveness:
  - (1) -0.062 [0.021]
  - (2) -0.055 [0.018]
  - (3) -0.060 [0.020]
- Number of observations:
  - (1) 172
  - (2) 149
  - (3) 172
  - (4) 384
- Number of countries: 15, 15, 15, 15
- Country FE: Yes (all columns)
- Year FE: Yes (all columns)
- Adj R2:
  - (1) 0.70
  - (2) 0.69
  - (3) 0.69
  - (4) 0.62

Note: Robust standard errors, clustered at the country level, are reported in brackets. A constant is included in each regression, but not shown in the table. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively.

*wpiea2022178-print-pdf - Section 3*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022178-print-pdf.pdf_