A. Introduction

1. Large contractions in output could generate a protracted economic recovery. When the economic damages to the factors of production originated by recessions, or more generally by a shock, preclude economies from recovering to their pre-shock trend, we say the shock generates scarring effects, also called hysteresis.

2. The COVID-19 shock, temporary in principle, would likely have long-term economic repercussions. The transmission channels, many still to be understood and discovered, have revealed the interaction of supply and demand factors.² The pandemic was peculiar in the sense that the fate of the virus itself produced a negative feedback loop on the behavior of consumers and producers alike, thus revealing the importance of beliefs and uncertainty, more generally, in affecting the transmission channels.³The economic effects would also differ across countries.

3. This chapter adds to recent studies seeking to understand the potential long-term effects of COVID-19.⁴ Most of the studies report that the potential for medium-term scarring from the pandemic appears related to the interplay of four elements: (1) the future path of the pandemic and associated containment measures; (2) the heavier impact of the pandemic shock on high-contact sectors and its sectoral spillovers; (3) the capability of businesses and workers to adapt to a lower-contact working environment and lower-contact transactions; and (4) the effectiveness of the policy response to limit economic damage.

4. Assessing the extent of the scarring effects is essential for the conduct of future economic policy in the ECCU. A better understanding of the factors affecting the scarring effects and their fiscal implications could help inform the discussions on policies needed to overcome them, especially for economies with limited economic diversification and high vulnerability to frequent shocks and natural disasters such as the ECCU countries. In this context, this chapter seeks to shed some light on answering the following two sets of questions:

• How is the pandemic-related downturn different from previous downturns? How does the ECCU compare to other regions? Which economic features of ECCU countries are related to the output losses observed during the pandemic?

• What are the fiscal implications of the pandemic in terms of fiscal space and debt dynamics in ECCU countries?

B. Output Losses and Economic Structure

5. Output losses in the ECCU countries during the pandemic have been large. To contextualize the magnitude of the pandemic-related downturn, we used forecast revisions across WEO vintages before, during, and after a particular period to measure the extent of the unexpected losses. The comparison period used is 2008–12 because of the downturn synchronization across economies and the global nature of the financial crisis, features that were prominent during 2020–21.⁵ For the ECCU countries, the forecast revisions were sizeable, during both the global financial crisis (GFC) and the ongoing pandemic, but pandemic losses outstripped losses from the GFC in the first year of the pandemic and increased as the pandemic evolved in the subsequent year. In advanced and emerging market economies, the output revisions were more prominent in the GFC than in COVID-19, and the forecast revisions in advanced and emerging market economies in the second year of the pandemic were lower than in the first year, indicating an earlier recovery than in the ECCU region.⁶

Output Losses

(Percentage difference from precrisis WEO)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculations.Note: Bars show the percent difference in real GDP after the crisis and anticipated GDP for the same year prior to the crisis for the indicated group. For the COVID-19 crisis, it compares the forecasted GDP levels from WEO Jan 2022 for 2020 and 2021 versus the ones from the WEO Oct 2019 (prior to the pandemic) For the global financial crisis, it compares the eve from WEO April 2007 for 2012 versus the observed 2012 GDP,

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Output Losses

(Percentage difference from precrisis WEO)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculations.Note: Bars show the percent difference in real GDP after the crisis and anticipated GDP for the same year prior to the crisis for the indicated group. For the COVID-19 crisis, it compares the forecasted GDP levels from WEO Jan 2022 for 2020 and 2021 versus the ones from the WEO Oct 2019 (prior to the pandemic) For the global financial crisis, it compares the eve from WEO April 2007 for 2012 versus the observed 2012 GDP,

• Download Figure
• Download figure as PowerPoint slide

Output Losses

(Percentage difference from precrisis WEO)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculations.Note: Bars show the percent difference in real GDP after the crisis and anticipated GDP for the same year prior to the crisis for the indicated group. For the COVID-19 crisis, it compares the forecasted GDP levels from WEO Jan 2022 for 2020 and 2021 versus the ones from the WEO Oct 2019 (prior to the pandemic) For the global financial crisis, it compares the eve from WEO April 2007 for 2012 versus the observed 2012 GDP,

• Download Figure
• Download figure as PowerPoint slide

6. The significant output contraction would generate scarring effects in the ECCU countries. Using the January 2022 WEO forecast, real GDP for ECCU was estimated to reach the 2019 level only in 2024 and remain below its pre- pandemic projected path throughout the forecasting horizon. In advanced economies, such a pattern was not present.⁷ A way to distinguish the nature of the output contraction during the pandemic across countries is to compute the correlation between the forecast revisions, what we call output losses, and initial conditions – measured by the pre-pandemic levels of GDP per capita. Interestingly, during the first year of the pandemic, such correlation was positive, although small, showing a fast spread of the virus to more affluent countries, which affected beliefs and distorted investment opportunities.

ECCU Real GDP

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Sources: IMF staff calculations

• Download Figure
• Download figure as PowerPoint slide

View Full Size

ECCU Real GDP

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Sources: IMF staff calculations

• Download Figure
• Download figure as PowerPoint slide

ECCU Real GDP

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Sources: IMF staff calculations

• Download Figure
• Download figure as PowerPoint slide

7. The degree of scarring could vary with countries’ economic structure and policy responses to the pandemic. We conduct a simple regression analysis covering 115 countries of estimates of output losses during 2020–22. We analyze the relationship between output losses and variables relevant for the ECCU countries, such as the share of tourism in GDP, the degree of economic complexity embedded in countries’ exports, and the extent of the lockdowns in response to the pandemic. We also control for the initial level of GDP per capita.

• We use the share of tourism in GDP and the percentage of the rural population to represent the extent of contact-intensive sectors. The tourism share is significant. Countries more reliant on tourism were more affected by the pandemic during 2020–22. The percentage of the rural population is significant but changes sign from negative in 2020 to positive in 2022, perhaps reflecting how the virus propagated initially to countries with large urban areas and then expanded to affect all countries and sectors.

• Similarly, economic complexity captured by a measure of the knowledge embedded in the products a country produces is significant in the regression of output losses. The interesting observation is its negative sign, which indicates that countries making more complex products were more resilient to the pandemic.

• As expected, the lockdown stringency variable is positive and significant.

^{Table 1.}

Annual Real GDP Losses

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP Per Capita, Rural Population and Tourism Share of GDP represent the levels pre-pandemic. Real GDP losses were calculated using differences in several WEO vintages of real GDP projections between January 2022 and October 2019. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, and World Bank staff estimates.

^{Table 1.}

Annual Real GDP Losses

(In percentage points)

	2020	2021	2022
Log (GDP Per Capita)	-1.838 (1.443)	0.311 (0.780)	-0.0131 (0.844)
Rural Population	-0.134** (0.0573)	0.0429 (0.0312)	0.0774** (0.0344)
Lockdown Stringency Index	0.209*** (0.0790)	0.0834* (0.0423)	0.0609 (0.0458)
Complexity Index	-0.874 (1.162)	-1.536** (0.622)	-1.519** (0.667)
Tourism Share of GDP	0.530* (0.299)	0.423** (0.163)	0.389** (0.175)
Constant	18.57 (14.20)	-1.417 (7.665)	1.088 (8.284)
Observations	117	116	114
R-squared	0.156	0.218	0.274

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP Per Capita, Rural Population and Tourism Share of GDP represent the levels pre-pandemic. Real GDP losses were calculated using differences in several WEO vintages of real GDP projections between January 2022 and October 2019. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, and World Bank staff estimates.

View Table

^{Table 1.}

Annual Real GDP Losses

(In percentage points)

	2020	2021	2022
Log (GDP Per Capita)	-1.838 (1.443)	0.311 (0.780)	-0.0131 (0.844)
Rural Population	-0.134** (0.0573)	0.0429 (0.0312)	0.0774** (0.0344)
Lockdown Stringency Index	0.209*** (0.0790)	0.0834* (0.0423)	0.0609 (0.0458)
Complexity Index	-0.874 (1.162)	-1.536** (0.622)	-1.519** (0.667)
Tourism Share of GDP	0.530* (0.299)	0.423** (0.163)	0.389** (0.175)
Constant	18.57 (14.20)	-1.417 (7.665)	1.088 (8.284)
Observations	117	116	114
R-squared	0.156	0.218	0.274

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP Per Capita, Rural Population and Tourism Share of GDP represent the levels pre-pandemic. Real GDP losses were calculated using differences in several WEO vintages of real GDP projections between January 2022 and October 2019. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, and World Bank staff estimates.

8. For the ECCU, the long-term effects from the pandemic could be significant. The pandemic impact on the quality of human capital induced by school closures and young death has the potential to bring the biggest negative long-term consequences as those variables have been affected more in ECCU countries than in other regions of the world (see chart).⁸ The regression results presented in Tables 2 and 3 show the response of cumulative output losses up to 2022 to explanatory variables representing human capital, either the share of young people (0 to 40 years old) that died from COVID-19 or the number of weeks of full school closures. Both variables are significant on most of the specifications, after controlling by the population median age, the pre-pandemic share of expenditure in health or education, the lockdown stringency, and others.

^{Table 2.}

Cumulative Real GDP Losses Between 2020–22, Health Component

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, Expenditure in Health and median age of the population represent the levels pre-pandemic.Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are not included in this regression. Sources: IMF staff calculations, World Bank staff estimates, Max Planck Institute for Demographic Research, World Health Organization Global Health Expenditure database, and University of Oxford.

^{Table 2.}

Cumulative Real GDP Losses Between 2020–22, Health Component

(In percentage points)

Log (GDP Per Capita)	-0.323 (0.678)	-0.458 (0.845)	-0.961 (0.852)
Young Deaths	0.345* (0.177)	0.392* (0.211)	0.290 (0.209)
Share of Expenditure in Health		-0.0185 (0.127)	-0.0446 (0.123)
Population Median Age		0.0455 (0.0999)	0.0843 (0.0986)
Lockdown Stringency Index			0.142** (0.0586)
Constant	8.080 (7.189)	7.930 (7.443)	3.940 (7.416)
Observations	67	66	65
R-squared	0.138	0.139	0.218

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, Expenditure in Health and median age of the population represent the levels pre-pandemic.Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are not included in this regression. Sources: IMF staff calculations, World Bank staff estimates, Max Planck Institute for Demographic Research, World Health Organization Global Health Expenditure database, and University of Oxford.

View Table

^{Table 2.}

Cumulative Real GDP Losses Between 2020–22, Health Component

(In percentage points)

Log (GDP Per Capita)	-0.323 (0.678)	-0.458 (0.845)	-0.961 (0.852)
Young Deaths	0.345* (0.177)	0.392* (0.211)	0.290 (0.209)
Share of Expenditure in Health		-0.0185 (0.127)	-0.0446 (0.123)
Population Median Age		0.0455 (0.0999)	0.0843 (0.0986)
Lockdown Stringency Index			0.142** (0.0586)
Constant	8.080 (7.189)	7.930 (7.443)	3.940 (7.416)
Observations	67	66	65
R-squared	0.138	0.139	0.218

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, Expenditure in Health and median age of the population represent the levels pre-pandemic.Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are not included in this regression. Sources: IMF staff calculations, World Bank staff estimates, Max Planck Institute for Demographic Research, World Health Organization Global Health Expenditure database, and University of Oxford.

^{Table 3.}

Regression on Cumulative Real GDP Losses Between 2020–22, Education Component

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita and Expenditure in Education represent the levels pre-pandemic. Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are included in this regression. Sources: IMF staff calculations, World Bank staff estimates, University of Oxford, and UNESCO global dataset.

^{Table 3.}

Regression on Cumulative Real GDP Losses Between 2020–22, Education Component

(In percentage points)

Log (GDP Per Capita)	-0.451 (0.353)	-0.349 (0.388)	-0.709* (0.375)	-0.738** (0.367)
Weeks of full school closures	0.116*** (0.0286)	0.143*** (0.0285)	0.0892*** (0.0324)	0.0799** (0.0319)
Share of Expenditure in Education		-0.0387 (0.0907)	0.0108 (0.0828)	0.0352 (0.0815)
Lockdown Stringency Index			0.105** (0.0430)	0.120*** (0.0425)
Weeks of full school closures_ECCU				0.381** (0.152)
Constant	9.652*** (3.448)	8.348* (4.232)	5.926 (3.930)	5.125 (3.856)
Observations	171	126	120	120
R-squared	0.102	0.183	0.240	0.280

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita and Expenditure in Education represent the levels pre-pandemic. Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are included in this regression. Sources: IMF staff calculations, World Bank staff estimates, University of Oxford, and UNESCO global dataset.

View Table

^{Table 3.}

Regression on Cumulative Real GDP Losses Between 2020–22, Education Component

(In percentage points)

Log (GDP Per Capita)	-0.451 (0.353)	-0.349 (0.388)	-0.709* (0.375)	-0.738** (0.367)
Weeks of full school closures	0.116*** (0.0286)	0.143*** (0.0285)	0.0892*** (0.0324)	0.0799** (0.0319)
Share of Expenditure in Education		-0.0387 (0.0907)	0.0108 (0.0828)	0.0352 (0.0815)
Lockdown Stringency Index			0.105** (0.0430)	0.120*** (0.0425)
Weeks of full school closures_ECCU				0.381** (0.152)
Constant	9.652*** (3.448)	8.348* (4.232)	5.926 (3.930)	5.125 (3.856)
Observations	171	126	120	120
R-squared	0.102	0.183	0.240	0.280

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita and Expenditure in Education represent the levels pre-pandemic. Cumulative real GDP losses were calculated using the difference in two WEO vintages of real GDP projections, January 2022, and October 2019. ECCU countries are included in this regression. Sources: IMF staff calculations, World Bank staff estimates, University of Oxford, and UNESCO global dataset.

COVID-19 Pandemic Impact on Labor Force

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: UNESCO global dataset Max Planck Institute for Demographic Research, and IMF staff calculations.Note: the LHS shows the average number of weeks in the region that schools were closed between Mar 2020 and Oct 2021. The RHS shows the percentage of young COVID-19 deaths registered until Mar 2022.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

COVID-19 Pandemic Impact on Labor Force

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: UNESCO global dataset Max Planck Institute for Demographic Research, and IMF staff calculations.Note: the LHS shows the average number of weeks in the region that schools were closed between Mar 2020 and Oct 2021. The RHS shows the percentage of young COVID-19 deaths registered until Mar 2022.

• Download Figure
• Download figure as PowerPoint slide

COVID-19 Pandemic Impact on Labor Force

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: UNESCO global dataset Max Planck Institute for Demographic Research, and IMF staff calculations.Note: the LHS shows the average number of weeks in the region that schools were closed between Mar 2020 and Oct 2021. The RHS shows the percentage of young COVID-19 deaths registered until Mar 2022.

• Download Figure
• Download figure as PowerPoint slide

C. Fiscal Legacies from COVID

9. Fiscal positions deteriorated significantly in the aftermath of the pandemic. Sharp curtailment of economic activity at the onset of the COVID-19 pandemic eroded public sector tax receipts and weakened the tax base. Meanwhile, spending needs rose to accommodate larger health budgets for pandemic containment and enhanced social assistance programs for household and business support during lockdowns. The economic shock was amplified in tourism-dependent economies as pandemic-related health concerns and mobility restrictions halted cross-border travel. The erosion of fiscal flows accelerated debt accumulation, exacerbating debt-to-GDP ratios that were already elevated from significant output losses.

10. Governments deviated further from their long-run fiscal targets and carried larger debt service burdens. While we do not find a significant difference between post-pandemic and pre-pandemic estimates of the long run debt stabilizing primary balance, fiscal deterioration in the near term has increased the gap between the current primary balance and the debt stabilizing primary balance for many countries. This implies more fiscal effort to return to the debt-stabilizing trajectory and limited fiscal space to cushion the impact of additional shocks. The larger debt burden relative to output also translates into more revenue share being allocated towards interest payments on debt. Moreover, as financial conditions tighten, the cost of rolling over commercial debt is expected to rise further. This is the case for most tourism-dependent countries in the Caribbean, which are seeing a reduction inestimatece available for fiscally sustainable spending in public goods of high societal value such as investment in climate resilient infrastructure, targeted social spending, and the buildup of fiscal buffers against natural disasters.

EMDEs: Fiscal Implications of Scarring Effects

(In Percent of FY GDP)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculation?.Note: The y-axis represents the difference between the primary balance sustainability gap (debt stabilizing primary balance – primary balance) estimated with data from WEO 2022 and WEO 2019; the debt stabilizing primary balance was estimated using each WEO projections for long term growth, implied interest rates and 2022 debt, The x-axis shows the difference between January 2022 WEO Interest Payments Projections end October 2019 WEO Projections for 2022.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

EMDEs: Fiscal Implications of Scarring Effects

(In Percent of FY GDP)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculation?.Note: The y-axis represents the difference between the primary balance sustainability gap (debt stabilizing primary balance – primary balance) estimated with data from WEO 2022 and WEO 2019; the debt stabilizing primary balance was estimated using each WEO projections for long term growth, implied interest rates and 2022 debt, The x-axis shows the difference between January 2022 WEO Interest Payments Projections end October 2019 WEO Projections for 2022.

• Download Figure
• Download figure as PowerPoint slide

EMDEs: Fiscal Implications of Scarring Effects

(In Percent of FY GDP)

Citation: IMF Staff Country Reports 2022, 254; 10.5089/9798400217272.002.A001

Source: IMF staff calculation?.Note: The y-axis represents the difference between the primary balance sustainability gap (debt stabilizing primary balance – primary balance) estimated with data from WEO 2022 and WEO 2019; the debt stabilizing primary balance was estimated using each WEO projections for long term growth, implied interest rates and 2022 debt, The x-axis shows the difference between January 2022 WEO Interest Payments Projections end October 2019 WEO Projections for 2022.

• Download Figure
• Download figure as PowerPoint slide

11. The pandemic brought tourism-dependent economies further away from their pre-pandemic debt trajectories. Cumulative output losses from COVID are a strong predictor of the worsening of debt as a percent of GDP. However, even after accounting for output losses, countries that lie in the top quartile of tourism contribution to GDP have higher debt accumulation than countries that are not as dependent on tourism (Table 4). Consequently, tourism-dependent economies are experiencing a dual shock of larger output losses exacerbating their debt-to-GDP ratios, and more debt accumulation that is not directly attributable to output losses.

^{Table 4.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Sectoral Composition of Output

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

^{Table 4.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Sectoral Composition of Output

(In percentage points)

Log (GDP Per Capita)	-0.285 (1.229)	0.0628 (1.209)	0.0608 (1.197)	0.596 (0.864)
Lockdown Stringency Index	0.0709 (0.0907)	0.0247 (0.0906)	0.0338 (0.0895)
Complexity Index	1.121 (1.322)	1.653 (1.310)	1.594 (1.290)
Tourism Share of GDP	0.796** (0.346)	0.505 (0.358)
Cumulative Output Losses 2020–22		0.598** (0.243)	0.578** (0.236)	0.595*** (0.222)
l(Top Quartile of Tourism GDP = 1)			5.445** (2.628)	4.854* (2.592)
Log (Debt in 2019)				3.033* (1.564)
Log (Average Nominal Interest Rate on Debt)				-1.897* (1.143)
Constant	4.951 (10.26)	0.388 (10.19)	1.490 (10.09)	-8.263 (10.67)
Observations	112	112	112	112
R-squared	0.070	0.121	0.139	0.175

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

View Table

^{Table 4.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Sectoral Composition of Output

(In percentage points)

Log (GDP Per Capita)	-0.285 (1.229)	0.0628 (1.209)	0.0608 (1.197)	0.596 (0.864)
Lockdown Stringency Index	0.0709 (0.0907)	0.0247 (0.0906)	0.0338 (0.0895)
Complexity Index	1.121 (1.322)	1.653 (1.310)	1.594 (1.290)
Tourism Share of GDP	0.796** (0.346)	0.505 (0.358)
Cumulative Output Losses 2020–22		0.598** (0.243)	0.578** (0.236)	0.595*** (0.222)
l(Top Quartile of Tourism GDP = 1)			5.445** (2.628)	4.854* (2.592)
Log (Debt in 2019)				3.033* (1.564)
Log (Average Nominal Interest Rate on Debt)				-1.897* (1.143)
Constant	4.951 (10.26)	0.388 (10.19)	1.490 (10.09)	-8.263 (10.67)
Observations	112	112	112	112
R-squared	0.070	0.121	0.139	0.175

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

12. Debt accumulation during the pandemic is positively correlated with credit access. This is because higher initial debt levels pre-COVID correlate with higher debt accumulation, and higher average borrowing costs pre-COVID (measured as average nominal interest rate on debt) are correlated with lower debt accumulation during the pandemic period (Table 4). Hence, only countries with pre-existing access to financing streams could avail themselves of the favorable global liquidity conditions that existed at the height of the pandemic. In this regard, we find that while the average cost of borrowing is not lower in the most tourism-dependent countries relative to the rest of the sample, they did enter the pandemic with already higher levels of debt, which may have also contributed to their deteriorating debt dynamics during the pandemic.

13. Higher pandemic spending is correlated with higher debt accumulation, for all countries in the sample (Table 5). However, the most tourism-dependent economies had higher average levels of fiscal expenditures to combat the pandemic, as a percent of GDP, resulting in greater overall debt accumulation.⁹ Using alternate indices of economic support,¹⁰ which combine domestic government measures and foreign aid measures, on an annual basis, we find that the disparity between debt accumulation because of fiscal support between tourism-dependent economies and other economies is largest in 2020, at the height of the pandemic, persisting into 2021 as tourism-dependent economies continue to suffer from repeated pandemic waves and re-emergence of travel restrictions, and weakening in 2022 as tourism revival strengthened (Table 6). Hence, the link between economic support and debt accumulation appears stronger in tourism-dependent economies, not only in terms of the direct correlation between the provision of economic support and debt accumulation but also because the level of economic support is more significant. To the extent that tourism-dependent countries are small, open economies with greater import dependence, this larger debt accumulation may also be partially attributable to smaller fiscal multipliers and greater leakages to imports and international goods.

^{Table 5.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Fiscal Variables

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

^{Table 5.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Fiscal Variables

(In percentage points)

Log (GDP Per Capita)	0.384 (1297)	-0.311 (0.930)	-0.555 (0.876)	-0.531 (0.856)
If Top Quartile of Tourism GDP=1)	-27.11 (24.55)	1.476 (3.633)	1.772 (3.877)	2232 (3.418)
Log (Covid Cases as percent of population)	0.357 (0.758)
Top Quartile of Tourism * Log (Covid Cases)	2.753 (2.113)
Cumulative Output Losses 2020-22	0.580** (0.223)	0.567** (0.217)	0.759*** (0214)	0.795*** (0.213)
Log (Debt in 2019)	2.575 (1.597)	1.569 (1.606)	1.564 (1.526)	1255 (1.516)
Fiscal Spending on Covid – above/below the line (percent of GDP)		0.249** (0.111)
Top Quartile of Tourism * Fiscal Spending above/below		0.270 (0.221)
Fiscal Spending on Covid – above the line			0.770*** (0.221)
Top Quartile of Tourism * Fiscal Spending above			0.225 (0.417)
Fiscal Spending on Covid-Non-health				0.909*** (0.249)
Top Quartile of Tourism * Fiscal Spending Non-health				0266 (0.441)
Fiscal Spending on Covid- Health
Top Quartile of Tourism’Fiscal Spending Health
Constant	-10.56 (10.46)	1.508 (11.19)	0.922 (10.29)	1.950 (10.18)
Observations	112	112	112	112
R-squared	0.171	D.233	0.274	0.289

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

View Table

^{Table 5.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Fiscal Variables

(In percentage points)

Log (GDP Per Capita)	0.384 (1297)	-0.311 (0.930)	-0.555 (0.876)	-0.531 (0.856)
If Top Quartile of Tourism GDP=1)	-27.11 (24.55)	1.476 (3.633)	1.772 (3.877)	2232 (3.418)
Log (Covid Cases as percent of population)	0.357 (0.758)
Top Quartile of Tourism * Log (Covid Cases)	2.753 (2.113)
Cumulative Output Losses 2020-22	0.580** (0.223)	0.567** (0.217)	0.759*** (0214)	0.795*** (0.213)
Log (Debt in 2019)	2.575 (1.597)	1.569 (1.606)	1.564 (1.526)	1255 (1.516)
Fiscal Spending on Covid – above/below the line (percent of GDP)		0.249** (0.111)
Top Quartile of Tourism * Fiscal Spending above/below		0.270 (0.221)
Fiscal Spending on Covid – above the line			0.770*** (0.221)
Top Quartile of Tourism * Fiscal Spending above			0.225 (0.417)
Fiscal Spending on Covid-Non-health				0.909*** (0.249)
Top Quartile of Tourism * Fiscal Spending Non-health				0266 (0.441)
Fiscal Spending on Covid- Health
Top Quartile of Tourism’Fiscal Spending Health
Constant	-10.56 (10.46)	1.508 (11.19)	0.922 (10.29)	1.950 (10.18)
Observations	112	112	112	112
R-squared	0.171	D.233	0.274	0.289

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

^{Table 6.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Other Economic Support Variables

(In percentage points)

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

^{Table 6.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Other Economic Support Variables

(In percentage points)

Log (GDP Per Capita)	0.859 (0.918)	0.735 (0.904)	0.834 (0.871)
l(Top Quartile of Tourism GDP=1)	-72.93** (33.62)	-30.66 (20.27)	2.002 (3.179)
Log (Economic Support Index 2020)	-0.239 (0.772)
Top Quartile of Tourism * Log(Economic Support 2020)	18.13** (7.819)
Cumulative Output Losses 2020–22	0.593*** (0.224)	0.586** (0.228)	0.586** (0.226)
Log (Debt in 2019)	2.609 (1.576)	2.559 (1.615)	2.734* (1.591)
Log (Economic Support Index 2021)		0.108 (0.640)
Top Quartile of Tourism * Log (Economic Support 2021)		8.496* (4.839)
Log (Economic Support Index 2022:			0.0310 (0.333)
Top Quartile of Tourism * Log(Economic Support 2022)			1.406 (0.852)
Constant	-10.53 (10.31)	-10.44 (10.52)	-11.72 (10.47)
Observations	109	109	109
R-squared	0.193	0.176	0.176

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

View Table

^{Table 6.}

Difference in Debt-to-GDP Projections Between January 2022 WEO and October 2019, Other Economic Support Variables

(In percentage points)

Log (GDP Per Capita)	0.859 (0.918)	0.735 (0.904)	0.834 (0.871)
l(Top Quartile of Tourism GDP=1)	-72.93** (33.62)	-30.66 (20.27)	2.002 (3.179)
Log (Economic Support Index 2020)	-0.239 (0.772)
Top Quartile of Tourism * Log(Economic Support 2020)	18.13** (7.819)
Cumulative Output Losses 2020–22	0.593*** (0.224)	0.586** (0.228)	0.586** (0.226)
Log (Debt in 2019)	2.609 (1.576)	2.559 (1.615)	2.734* (1.591)
Log (Economic Support Index 2021)		0.108 (0.640)
Top Quartile of Tourism * Log (Economic Support 2021)		8.496* (4.839)
Log (Economic Support Index 2022:			0.0310 (0.333)
Top Quartile of Tourism * Log(Economic Support 2022)			1.406 (0.852)
Constant	-10.53 (10.31)	-10.44 (10.52)	-11.72 (10.47)
Observations	109	109	109
R-squared	0.193	0.176	0.176

Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; the variables GDP per capita, median age of the population and tourism share of GDP represent the levels pre-pandemic. Sources: IMF staff calculations, World Travel & Tourism Council, Harvard’s Growth Lab, University of Oxford, and World Bank staff estimates.

D. Conclusions

14. The pandemic has inflicted large output losses in the ECCU that would generate scarring effects. Countries with significant importance of high-contact sectors, such as tourism, or limited fiscal space to mitigate the pandemic, like the ECCU, are expected to suffer the most. After controlling for restrictions on mobility and other variables, our empirical results show that high tourism exposure and low economic diversification are correlated with the output losses during 2020–22 in the ECCU, measured as downward revisions to growth forecasts. Most ECCU countries now have a smaller fiscal space, and some need to allocate a larger share of revenues into interest payments. This suggests that the governments’ ability to invest in projects with a higher fiscal multiplier, such as infrastructure, will be reduced in the coming years. In other words, the impact of the pandemic on the fiscal accounts will also affect long-term growth and can be another potential source of hysteresis in the region. The impact of the pandemic on the quality of human capital, induced by school closures and young deaths, has the potential to have the largest negative long-term consequences.

15. ECCU countries need to balance difficult tradeoffs to mitigate scaring effects of the pandemic, other recent shocks, and limited fiscal policy space. In the short term, the priorities are to continue health spending to cope with the pandemic and use effective social transfers to cope with rising living costs. In the medium term, moving from income support and job retention measures to adopting active labor market policies would facilitate the reallocation of workers and resources to their most productive uses and help foster productivity growth.