## References (wpiea2020243-print-pdf) ## Source details **Canonical URL:** [References (wpiea2020243-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020243-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020243-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020243-print-pdf.pdf.json) --- ### I. Introduction — climate context and tourism vulnerability - Global average surface temperature has risen by 1.1 degrees Celsius since 1880. - Global average sea level has increased by 21-24 centimeters, with about a third of that occurring in the last two and a half decades. - Projections: global annual mean temperatures could increase by as much as 4 degrees Celsius over the next century and global mean sea level is likely to rise at least 30 centimeters above 2000 levels. - The Caribbean’s climate-sensitive tourism industry accounts for 20 to 90 percent of GDP across countries in the region, and on average 30 percent of employment. - About 95 percent of accommodation facilities and 80 percent of tourist attractions in the Caribbean are located at sea level along the coast. - Examples of disaster costs in small island states: - Cost of Hurricane Ivan for Grenada in 2004 amounted to 148 percent of GDP. - Cost of Hurricane Maria for Dominica in 2017 reached 260 percent of GDP. ### II. Key empirical finding on climate vulnerability and tourism revenues - Main empirical result: a ten percentage-point increase in climate change vulnerability leads to: - a decline of 9 percentage points in tourism earnings per visitor, or - a reduction of 10 percentage points in tourism revenues as a share of GDP. - Cross-country range implication: the difference between the minimum and maximum levels of climate change vulnerability among Caribbean countries is 20 percentage points; a country with high vulnerability already earns about 18 percentage points less in tourism revenues compared to a country with low vulnerability. - The ND-GAIN vulnerability index is used as the main explanatory variable; it captures susceptibility based on 36 variables covering 6 sectors (food, water, health, ecosystem services, human habitat, infrastructure). ### III. Data overview and summary statistics - Sample: 15 Caribbean countries; unbalanced annual panel over the period 1995–2017. - ND-GAIN database coverage: 184 countries over the period 1995–2017. - Dependent variable: international tourism receipts per visitor (ratio of annual tourism earnings to number of international visitors, or as a share of GDP). - Control variables included: real GDP per capita, real effective exchange rate (REER), crime (homicides per 100,000 inhabitants), composite index of government effectiveness. - Key descriptive statistics: - Tourism revenues per visitor: Obs. 365; Mean 1139.3; Std. Dev. 542.8; Min. 219.4; Max. 3240.0. - Tourism revenues as a share of GDP: Obs. 368; Mean 3.1; Std. Dev. 11.6; Min. 0.0; Max. 68.3. - Real GDP per capita: Obs. 368; Mean 8790.2; Std. Dev. 6852.9; Min. 662.3; Max. 32080.4. - Climate vulnerability index: Obs. 368; Mean 0.43; Std. Dev. 0.05; Min. 0.37; Max. 0.57. - REER: Obs. 345; Mean 99.5; Std. Dev. 12.5; Min. 51.6; Max. 132.4. - Crime: Obs. 263; Mean 19.8; Std. Dev. 12.6; Min. 1.4; Max. 65.4. - Government effectiveness: Obs. 263; Mean 0.3; Std. Dev. 0.5; Min. -0.7; Max. 1.6. - Trend in vulnerability: climate change vulnerability improved among Caribbean countries between 1995 and 2017, with the average climate change vulnerability index declining by 1.95 percent; the rate of change varies from a minimum of 0.38 percent to a maximum of 3.40 percent. - Relative vulnerability: Caribbean countries remain almost twice as vulnerable to climate change relative to the U.S. - Time-series properties: Im-Pesaran-Shin (2003) unit root tests indicate variables are stationary after logarithmic transformations. ### IV. Empirical methodology - Baseline dynamic specification includes a lagged dependent variable with country fixed effects and time effects. - Estimation approach: system Generalized Method of Moments (system GMM) following Arellano and Bover (1995) and Blundell and Bond (1998); one-step version reported. - Rationale: system GMM used to address bias from correlation between lagged dependent variable and unobserved country-specific effects; robust standard errors clustered at country level to account for heteroskedasticity. - Instrument strategy: avoid instrument proliferation by following Roodman (2009). - Diagnostic checks reported: - AR(1) and AR(2) tests reported as p-values for first- and second-order autocorrelated disturbances in the first-differenced equation. - Hansen J-test reported to assess validity of internal instruments. - Findings indicate high first-order autocorrelation but no evidence of significant second-order autocorrelation, and Hansen J-test supports instrument validity. ### V. Estimation results and robustness - Determinants of international tourism earnings (baseline): - Real GDP per capita: positive effect. - Government effectiveness: positive effect. - Relative prices (REER): negative effect. - Crime: negative effect. - Climate vulnerability coefficient: negative and statistically significant across system GMM estimations. - Key baseline estimation results (Table 2; Dependent variable: tourism revenue per visitor): - Tourism revenue t-1: 0.695*** [0.052] (Fixed Effects); 0.846*** [0.047] (System GMM). - Climate vulnerability: -1.716 [3.629] (Fixed Effects); -1.117*** [0.577] (System GMM). - Real GDP per capita: 0.209 [0.208] (Fixed Effects); 0.068*** [0.048] (System GMM). - REER: 0.001 [0.002] (Fixed Effects); -0.003* [0.003] (System GMM). - Crime: -0.001 [0.001] (Fixed Effects); -0.001* [0.001] (System GMM). - Government effectiveness: 0.085 [0.056] (Fixed Effects); 0.048 [0.039] (System GMM). - Number of observations: 241; Number of countries: 15. - Country FE: Yes; Year FE: Yes. - Adj R2: 0.79. - AR1 p-value: 0.003. - AR2 p-value: 0.900. - Hansen J-test p-value: 0.220. - Note: Robust standard errors, clustered at the country level, reported in brackets. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. - Robustness checks (Table 3): - Sample: 15 Caribbean countries, period 1995–2017. - Approaches: truncated sample (5th and 95th percentiles), alternative tourism revenue measure (international tourism receipts as a share of GDP), alternative climate vulnerability measure (occurrence of natural disasters from EM-DAT). - Selected estimates: - Tourism revenue t-1: 0.822*** [0.057] (Truncated sample); 0.993*** [0.017] (Alternative measure of tourism revenue); 0.858*** [0.116] (Alternative measure of climate vulnerability). - Climate vulnerability: -0.695*** [0.185] (Truncated sample); -0.690*** [0.031] (Alternative measure of tourism revenue); -0.018*** [0.121] (Alternative measure of climate vulnerability). - Real GDP per capita: 0.065*** [0.135]; 0.054*** [0.120]; 0.026*** [0.112]. - REER: -0.003* [0.002]; -0.002* [0.002]; -0.003* [0.003]. - Crime: -0.001* [0.001]; -0.001* [0.001]; -0.002* [0.001]. - Government effectiveness: 0.042 [0.035]; 0.055 [0.040]; 0.026 [0.074]. - Number of observations: 220; 144; 146. - Number of countries: 15 in all specifications. - AR1 p-values: 0.001; 0.000; 0.000. - AR2 p-values: 0.820; 0.585; 0.803. - Hansen J-test p-values: 0.188; 0.169; 0.225. - Robustness checks confirm the negative and significant association between climate vulnerability and tourism revenue measures. ### VI. Policy implications and recommendations - Empirical findings motivate a comprehensive action plan emphasizing: - Improve structural resilience through comprehensive adaptation policies. - Strengthen financial resilience through fiscal buffers and insurance schemes. - Improve economic diversification and policy management across the Caribbean and other small island states. - Diversification constraints acknowledged: limited scope for diversification due to lack of economies of scale, but opportunities exist to increase the share of domestic input in consumption and production and to move up the value-added chain. - Policy priorities recommended: - Strengthen physical, financial, and institutional resilience through comprehensive adaptation measures. - Strengthen financial resilience via fiscal buffers and insurance schemes. - Improve economic diversification and policy management to reduce extreme dependence on tourism. - Reprioritize public investments to improve resilience against climate change to enhance competitiveness and structural resilience. - Enhance investment in physical infrastructure and technology, strengthen education and labor skills, and create a more nurturing business environment to promote entrepreneurship. - Caribbean-specific actions noted: - Adaptation actions have focused on coastal zones, protecting water resources and agriculture, and improving institutions, governance, and planning to ensure physical resilience and food security. - Vulnerability and impact assessments account for 10 percent of all adaptation actions throughout the Caribbean; financing, technical capacity, and human resources constrain a more comprehensive adaptation strategy, especially in the post-pandemic world. ### Conclusion and main empirical points - Climate change vulnerability is more important than other factors associated with international tourism performance. - In the dynamic model, a ten percentage-point increase in climate change vulnerability is associated with about 9 percentage-point decline in international tourism earnings per visitor; when measured as a share of GDP, the reduction is 10 percentage points. - Given the 20 percentage-point spread in climate vulnerability across Caribbean countries, high-vulnerability countries already earn about 18 percentage points less in tourism revenues relative to low-vulnerability peers. - The negative impact is economically significant even when comparing one year to the next, and cumulative effects over longer horizons are likely larger with intensifying frequency and severity of climate shocks. *IMF Working Paper content (wpiea2020243-print-pdf).* ### References ............................................................................................................. ### References (wpiea2020243-print-pdf) ### I. Introduction — climate context and tourism vulnerability - Global average surface temperature has risen by 1.1 degrees Celsius since 1880. - Global average sea level has increased by 21-24 centimeters, with about a third of that occurring in the last two and a half decades. - Projections: global annual mean temperatures could increase by as much as 4 degrees Celsius over the next century and global mean sea level is likely to rise at least 30 centimeters above 2000 levels. - The Caribbean’s climate-sensitive tourism industry accounts for 20 to 90 percent of GDP across countries in the region, and on average 30 percent of employment. - About 95 percent of accommodation facilities and 80 percent of tourist attractions in the Caribbean are located at sea level along the coast. - Examples of disaster costs in small island states: - Cost of Hurricane Ivan for Grenada in 2004 amounted to 148 percent of GDP. - Cost of Hurricane Maria for Dominica in 2017 reached 260 percent of GDP. ### II. Key empirical finding on climate vulnerability and tourism revenues - Main empirical result: a ten percentage-point increase in climate change vulnerability leads to: - a decline of 9 percentage points in tourism earnings per visitor, or - a reduction of 10 percentage points in tourism revenues as a share of GDP. - Cross-country range implication: the difference between the minimum and maximum levels of climate change vulnerability among Caribbean countries is 20 percentage points; a country with high vulnerability already earns about 18 percentage points less in tourism revenues compared to a country with low vulnerability. - The ND-GAIN vulnerability index is used as the main explanatory variable; it captures susceptibility based on 36 variables covering 6 sectors (food, water, health, ecosystem services, human habitat, infrastructure). ### III. Data overview and summary statistics - Sample: 15 Caribbean countries; unbalanced annual panel over the period 1995–2017. - ND-GAIN database coverage: 184 countries over the period 1995–2017. - Dependent variable: international tourism receipts per visitor (ratio of annual tourism earnings to number of international visitors, or as a share of GDP). - Control variables included: real GDP per capita, real effective exchange rate (REER), crime (homicides per 100,000 inhabitants), composite index of government effectiveness. - Key descriptive statistics (as reported): - Tourism revenues per visitor: Obs. 365; Mean 1139.3; Std. Dev. 542.8; Min. 219.4; Max. 3240.0. - Tourism revenues as a share of GDP: Obs. 368; Mean 3.1; Std. Dev. 11.6; Min. 0.0; Max. 68.3. - Real GDP per capita: Obs. 368; Mean 8790.2; Std. Dev. 6852.9; Min. 662.3; Max. 32080.4. - Climate vulnerability index: Obs. 368; Mean 0.43; Std. Dev. 0.05; Min. 0.37; Max. 0.57. - REER: Obs. 345; Mean 99.5; Std. Dev. 12.5; Min. 51.6; Max. 132.4. - Crime: Obs. 263; Mean 19.8; Std. Dev. 12.6; Min. 1.4; Max. 65.4. - Government effectiveness: Obs. 263; Mean 0.3; Std. Dev. 0.5; Min. -0.7; Max. 1.6. - Trend in vulnerability: climate change vulnerability improved among Caribbean countries between 1995 and 2017, with the average climate change vulnerability index declining by 1.95 percent; the rate of change varies from a minimum of 0.38 percent to a maximum of 3.40 percent. - Relative vulnerability: Caribbean countries remain almost twice as vulnerable to climate change relative to the U.S. - Time-series properties: Im-Pesaran-Shin (2003) unit root tests indicate variables are stationary after logarithmic transformations. ### IV. Empirical methodology - Baseline dynamic specification (lagged dependent variable included) with country fixed effects and time effects. - Estimation approach: system Generalized Method of Moments (system GMM) following Arellano and Bover (1995) and Blundell and Bond (1998); one-step version reported. - Rationale: system GMM used to address bias from correlation between lagged dependent variable and unobserved country-specific effects; robust standard errors clustered at country level to account for heteroskedasticity. - Instrument strategy: avoid instrument proliferation by following Roodman (2009). - Diagnostic checks: AR(1) and AR(2) tests reported as p-values for first- and second-order autocorrelated disturbances in the first-differenced equation; Hansen J-test reported to assess validity of internal instruments. Findings indicate high first-order autocorrelation but no evidence of significant second-order autocorrelation, and Hansen J-test supports instrument validity. ### V. Estimation results and robustness - Model fit: baseline findings indicate a strong fit of the model to the dataset. - Determinants of international tourism earnings: - Real GDP per capita: positive effect. - Government effectiveness: positive effect. - Relative prices (REER): negative effect. - Crime: negative effect. - Climate vulnerability coefficient: negative and statistically significant across system GMM estimations, indicating greater vulnerability is associated with lower international tourism receipts per visitor (or as a share of GDP) in the 15-country sample over 1995–2017. ### VI. Policy implications and recommendations - The empirical findings motivate a comprehensive action plan emphasizing: - Improve structural resilience through comprehensive adaptation policies. - Strengthen financial resilience through fiscal buffers and insurance schemes. - Improve economic diversification and policy management across the Caribbean and other small island states. - Diversification constraints acknowledged: limited scope for diversification due to lack of economies of scale, but opportunities exist to increase the share of domestic input in consumption and production and to move up the value-added chain. *Source: IMF working paper content (wpiea2020243-print-pdf) — References and excerpts from Introduction through Estimation Results.* ### 2017. The magnitude of estimated coefficients suggests that climate change vulnerability is more ### 2017. The magnitude of estimated coefficients suggests that climate change vulnerability is more ### Main empirical findings - Climate change vulnerability is more important than other factors associated with international tourism performance. - In the dynamic model, a ten percentage-point increase in climate change vulnerability is associated with about 9 percentage-point decline in international tourism earnings per visitor. - When tourism revenues are measured as a share of GDP, a ten percentage-point increase in climate change vulnerability results in a reduction of 10 percentage points (as shown in Table 3). - The effect remains economically and statistically significant after controlling for conventional macroeconomic and social factors (such as the level of income, relative prices, crime, and government effectiveness) and persistence in international tourism demand over time. - The difference between the minimum and maximum levels of climate change vulnerability among Caribbean countries is 20 percentage points, implying a country with high-level of climate change vulnerability earns about 18 percentage points less in tourism revenues compared to a country with a low-level of climate change vulnerability. - The negative impact is economically significant even when comparing one year to the next, and cumulative effects over longer horizons are likely larger with intensifying frequency and severity of climate shocks. ### Key baseline estimation results (Table 2) - Dependent variable: tourism revenue per visitor. - Estimation methods: Fixed Effects and System GMM. - Tourism revenue t-1: 0.695*** [0.052] (Fixed Effects); 0.846*** [0.047] (System GMM). - Climate vulnerability: -1.716 [3.629] (Fixed Effects); -1.117*** [0.577] (System GMM). - Real GDP per capita: 0.209 [0.208] (Fixed Effects); 0.068*** [0.048] (System GMM). - REER: 0.001 [0.002] (Fixed Effects); -0.003* [0.003] (System GMM). - Crime: -0.001 [0.001] (Fixed Effects); -0.001* [0.001] (System GMM). - Government effectiveness: 0.085 [0.056] (Fixed Effects); 0.048 [0.039] (System GMM). - Number of observations: 241; Number of countries: 15. - Country FE: Yes; Year FE: Yes. - Adj R2: 0.79. - AR1 p-value: 0.003. - AR2 p-value: 0.900. - Hansen J-test p-value: 0.220. - Note: Robust standard errors, clustered at the country level, reported in brackets. A constant is included but not shown. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. ### Robustness checks (Table 3) - Sample: 15 Caribbean countries, period 1995–2017. - Approaches: truncated sample (5th and 95th percentiles), alternative tourism revenue measure (international tourism receipts as a share of GDP), alternative climate vulnerability measure (occurrence of natural disasters from EM-DAT). - Tourism revenue t-1: 0.822*** [0.057] (Truncated sample); 0.993*** [0.017] (Alternative measure of tourism revenue); 0.858*** [0.116] (Alternative measure of climate vulnerability). - Climate vulnerability: -0.695*** [0.185] (Truncated sample); -0.690*** [0.031] (Alternative measure of tourism revenue); -0.018*** [0.121] (Alternative measure of climate vulnerability). - Real GDP per capita: 0.065*** [0.135]; 0.054*** [0.120]; 0.026*** [0.112]. - REER: -0.003* [0.002]; -0.002* [0.002]; -0.003* [0.003]. - Crime: -0.001* [0.001]; -0.001* [0.001]; -0.002* [0.001]. - Government effectiveness: 0.042 [0.035]; 0.055 [0.040]; 0.026 [0.074]. - Number of observations: 220; 144; 146. - Number of countries: 15 in all specifications. - Country FE: Yes; Year FE: Yes in all specifications. - AR1 p-values: 0.001; 0.000; 0.000. - AR2 p-values: 0.820; 0.585; 0.803. - Hansen J-test p-values: 0.188; 0.169; 0.225. - Note: Robust standard errors, clustered at the country level, reported in brackets. A constant is included but not shown. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively. ### Conclusion and policy recommendations - The empirical evidence confirms that international tourism revenues across the Caribbean region are adversely affected by greater vulnerability to climate change. - On average, a ten percentage-point increase in climate change vulnerability is associated with a decline of 9 percentage points in tourism earnings per visitor (or a reduction of 10 percentage points in tourism revenues as a share of GDP), after controlling for conventional factors. - Given the 20 percentage-point spread in climate vulnerability across Caribbean countries, high-vulnerability countries already earn about 18 percentage points less in tourism revenues relative to low-vulnerability peers. - As extreme weather outcomes become more frequent and severe, the economic impact of climate-related shocks on tourism is likely to grow. - Policy priorities recommended: - Strengthen physical, financial, and institutional resilience through comprehensive adaptation measures. - Strengthen financial resilience via fiscal buffers and insurance schemes. - Improve economic diversification and policy management to reduce extreme dependence on tourism. - Reprioritize public investments to improve resilience against climate change to enhance competitiveness and structural resilience. - Enhance investment in physical infrastructure and technology, strengthen education and labor skills, and create a more nurturing business environment to promote entrepreneurship. - For Caribbean countries specifically, adaptation actions have focused on coastal zones, protecting water resources and agriculture, and improving institutions, governance, and planning to ensure physical resilience and food security. - Vulnerability and impact assessments account for 10 percent of all adaptation actions throughout the Caribbean; a more comprehensive adaptation strategy is needed but is constrained by financing, technical capacity, and human resources, especially in the post-pandemic world. - Opportunities exist to increase the share of domestic input in consumption and production and to move up the value-added chain in tourism and other sectors such as agriculture and fisheries, despite limited economies of scale. *IMF Working Paper content (wpiea2020243-print-pdf).* --- _Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020243-print-pdf.pdf_