The Economic Growth Impact of Hurricanes in the Caribbean

Transcription

1 The Economic Growth Impact of Hurricanes in the Caribbean Eric Strobl Ecole Polytechnique, CREST, and SALISES Presentation prepared for the 28 th of October workshop on Natural Disasters in the Caribbean: Assessing the Physical and Economic Impacts, Mitigation and Insurance

2 INTRODUCTION Hurricanes are the most frequent and destructive of all types of natural disasters in the Caribbean Media mostly quantifies the effects of hurricanes in terms of cost of physical damages But hurricanes can have much more wide reaching effects (ex: business interruptions etc.) Arguably it is these indirect, longer-term impacts on the economy that policy makers are more interested in Question: What are the quantitative impacts of a hurricane strike on a Caribbean economy?

3 INTRODUCTION What I will discuss today: 1. Estimation Method 2. Results 3. Possible Factors Driving Results 4. Simulations 5. Concluding Remarks

4 -.05 economic growth rate ESTIMATION METHOD Example: Grenada and Hurricane Ivan (2004) Economic Growth Rate: 2003: +6.25% 2004: -6.75% 2005: % Grenada: Economic Growth Rate year

5 ESTIMATION METHOD Assumption: Economies are on some economic growth path and a hurricane is a natural event that causes a perturbation to the functioning of the economic system Statistical Estimation: Hurricane Destruction Economic Growth Rate: Quantitative Impact? Data Needs: [a] Economic Growth Rate: usually cross-country level growth in GDP per capita over time [b] Hurricane Destruction: Two approaches in the literature: 1. Estimate of the Damages (direct losses) 2. Physical aspects of the phenomena (wind speed)

6 RESULTS Results for the Caribbean: Hsiang (2010): no effect Bluedorn (2005): mean strike short term effect of 0.6% Strobl (2011): mean strike short term effect of 0.8% Question: Can we trust these findings?

7 1. Accounting and Underlying Dynamics 2. Aggregation Bias 3. Measurement Error

8 Accounting and Underlying Dynamics

9 Accounting and Underlying Dynamics

10 Accounting and Underlying Dynamics

11 Accounting and Underlying Dynamics

12 Accounting and Underlying Dynamics

13 Accounting and Underlying Dynamics

14 Accounting and Underlying Dynamics

15 Accounting and Underlying Dynamics Estimated economic growth impact over time t-1 to t: [1]Direct losses (replaced) + [2] Creative destruction [3](Gross) Indirect losses [1] Direct losses (replaced) Construction Industry: 1.4% in year t and t+1 (Hsiang,2010) International Financial Aid: average of around 150% of direct losses (damages) Caveat: resources not used elsewhere foregone consumption or alternative investment? [2] Creative destruction Evidence of this historically in plantation economies (Mulcahy, 2006) but little other

16 Accounting and Underlying Dynamics [3] (Gross) Indirect losses Lack of evidence of both the nature and (quantitative) extent of this Will depend on underutilization of resources

17 Aggregation Bias Two types: [a] Sectoral [b] Spatial [a] Sectoral (i) Tourism: Granvorka & Strobl (2011): 2% in arrivals Hsiang (2011): 1.0% in first year and 1.8% thereafter (ii) Agriculture: Strobl (2011): 0.7% Hsiang (2011): 1.8% (iii) Other Sectors: Hsiang (2011): Manufacturing: none; Mining and Utilities: 0.9% ; Services: 0.9%

18 Aggregation Bias [b] Spatial Natural Disasters are a `local phenomena Hurricane Destruction Distribution Ivan (2004): Jamaica

19 Aggregation Bias Lack of local income data Nightlights data as a proxy: Bertinelli & Strobl (2011): 3.4% compared to 0.6%

20 Measurement Error Estimation requires data on [a] Economic growth rate, and [b] Hurricane destruction [a] Economic growth rate: common data quality not perfect Note: Using aggregate nightlights data Bertinelli & Strobl (2011) find 0.6% [b] Hurricane destruction: (1) Ex-post Damages (EMDAT etc.) (2) Wind Field Model Note: measurement error in Hurricane destruction proxy can lead to tendency towards 0 effect

21 SIMULATIONS Downscaling technique of Emanuel et al (2008) for GDFL-CM2.0 climate model: 2,000 tropical storms in Atlantic basin under climate conditions of & : 9 tropical storms annually (max wind speed: avg. of 140 km/hr) : 15 tropical storms annual (max wind speed: avg. of 150 km/hr) Example of effect on economic growth rate: Bahamas using 10,000 random draws For both and : Bahamas affected about every 5 years

22 SIMULATIONS Economic Growth Impact of Hurricane Strikes (0.6% effect) Graphs by climate economic growth rate impact

23 SIMULATIONS Economic Growth Impact of Hurricane Strikes (3.4% effect) Graphs by climate economic growth rate impact

24 CONCLUDING REMARKS Existing studies seem to suggest that the impact of hurricanes on economic growth rates in the Caribbean is small and short-lived Could be many reasons for this: (a) actually true, (b) data masks underlying differences, (c) bad data Better data and further research is required Very Long-term impacts need also to be considered