Abstract

This study estimates the impact of hurricanes on migration from 30 Central American and Caribbean countries to the United States from 1989 to 2005. In contrast to previous studies, hurricane destruction indices are employed to study the relationship of hurricanes and migration. These indices measure geographical destruction, which gives a more comprehensive and accurate view of the damage and impact that hurricanes have on the movement of people to international destinations. Controlling for the host country’s migrant stock and the home country’s income, country fixed-effects estimation shows that hurricanes have a positive impact on the ratio of the number of migrants to the home country’s population. On average, hurricanes increase migration by roughly 6%, but the impact is greater for more damaging storms. Estimating the geographical effects reveals that the size of this impact varies across countries. The most damaging storms are related to an increase up to 540% in the ratio of migrants to the home country’s population.

1. Introduction

Many countries across the world are vulnerable to natural disasters. Hurricanes, which often result in extensive damages, are an example of one such natural disaster that is faced by countries both in the developed and developing worlds. The Central American and Caribbean (CAC) region is particularly vulnerable to hurricanes, with many of the countries from this region experiencing damages that often hamper recovery and growth. Historical hurricane tracks housed by the National Oceanic and Atmospheric Administration between 1989 and 2005 record approximately 269 storms1 that have tracked through the locations of CAC countries. Hurricanes, a high-impact disaster in the region, are predicted to become less frequent (Knutson et al. 2010), though most agree that future storms will be more severe (Webster et al. 2005), which has the potential to impact agriculture and other industries, welfare, conflict, and economies (Mohan 2016; Hsiang 2010; Spencer and Polachek 2015; Spencer 2016). In addition, the International Organization for Migration (IOM 2017) records that there will be at most 200 million climate change–induced migrants by 2050 moving either within or outside of their countries. Further, a strand of literature has shown the occurrence of population changes resulting from hurricanes and tropical storms (including Schultz and Elliott 2013; Logan and Xu 2015; Fussell et al. 2017). Thus, it appears that the ill effects of hurricanes could possibly induce migration (Ouattara and Strobl 2014; McLeman and Hunter 2010), whether internally or externally; however, Findlay (2011) points out that environmentally induced migration is likely short distance, and the possibility of international migration exists only if individuals are already part of the migration system. However, the physical harm, the loss of livelihood and homes, and the strong desire to relocate from areas that are regarded as high-risk disaster areas act as push factors for movement because people want to avoid unfortunate experiences (McLeman and Smit 2006; Afonso 2011; Morris et al. 2002; Logan and Xu 2015). The loss of livelihood, for example, can impact household welfare, inhibiting the ability of families to provide for themselves. In developing countries, insurance schemes to protect crops and livestock are severely limited, unaffordable, or nonexistent. Thus, external migration is an attractive alternative (Afonso 2011) for securing livelihood and welfare.

This paper uses the CAC region as a case study to investigate the movement of people to the United States resulting from hurricane strikes, where the United States is a top immigration choice for these individuals (Afonso 2011). Migration to the United States dates back to the early 1900s for the Caribbean, with an acceleration in the 1960s, which saw a 248% increase in migrants. Considering 1980, the population increased by 1.5 times in 1990 and by almost 2.7 times in 2006, with a threefold increase by 2014 (Zong and Batalova 2016). The 1980s marked the beginning of migration to the United States for Central America. In comparison to 1980, by 1990, the immigrant population grew by more than 3 times, and there was an observed growth of almost 7.5 times in 2006; by 2015, there was a tenfold increase (Lesser and Batalova 2017). Migration was driven by political instability, violence, natural disasters, and economic pressures. These conditions continue to drive migration to the United States (Lesser and Batalova 2017). As it relates to natural disasters, Mahajan and Yang’s (2017) study using census data concludes that new legal, permanent residents in the United States are associated with movements as a result of hurricanes, thus supporting the point that disasters drive migration. Sherburne (2017) also confirmed the increase in migration to the United States when hurricanes take place in other countries, including those from the CAC region.

The literature quantifying the impact of hurricanes on migration is rather limited, with some unable to identify the impact on the movement of people. For example, for the U.S. Gulf Coast, Logan et al. (2016) clearly establish that hurricane-induced population changes do take place, but they could not discern the type of migration taking place. There is, however, evidence to show that disasters can influence external migration. See, for instance, Afonso (2011), who alludes to the out-migration of Bangladeshis following massive flooding in 2007, and Warner (2010), who shows that volcanic eruptions in Montserrat result in external movement of people. More specifically, the literature features two noteworthy hurricane-migration examples. Hurricane Katrina, for example, which struck the United States in 2005, resulted in widespread migration of those displaced immediately after the storm (McIntosh 2008). As it relates to the CAC region, McGirk (2000) documents the slow recovery of Hondurans that were affected by Hurricane Mitch. This probably explains the large out-migration that followed after the hurricane, an increase of 40% (UNFAO 2001; McLeman and Smit 2006). Moreover, U.S. immigration reported granting the Hondurans and Nicaraguans the opportunity to stay and work in the United States by 2003; however, migrants also included those from Guatemala and El Salvador, countries also affected by Hurricane Mitch (Kugler and Yuksel 2008). Afonso (2011) is the only known study that investigates the impact of hurricanes on migration from the CAC region to the United States. The study finds a 16.5% increase in migration through the examination of only severe storms. From the results of Afonso (2011) and the case-based evidence (UNFAO 2001) provided, it is clear that hurricanes impact external migration. They further underscore the fact that the literature is severely lacking in the number of studies quantifying the impact of hurricanes on outward migration.

Although Afonso’s (2011) study is unique to the literature in terms of addressing migration from the CAC region, it is likely that the estimated impact is an overestimation of the effect of hurricanes on migration due to the limited cases of the storms observed. As a result, it is difficult to make a conclusive argument as it relates to the impact of hurricanes on migration from the CAC region. Further, Drabo and Mbaye (2015) study different categories of natural disasters, including meteorological disasters, which possibly include hurricanes. The results show that meteorological disasters increase net migration by about 29% and 41%, a large difference compared to Afonso (2011). One plausible reason for this disparity in the results may be that residents in disaster-prone areas, such as the CAC region, may simply adjust their behavior so that they learn to mitigate the impact of hurricane shocks, or governments implement strategies to lessen the effects of the storms (Paul 2005). When changes in the environment are gradual, these residents may be forced to relocate temporarily by way of internal migration, but may decide to return and rebuild (Tacoli 2009; Alexeev et al. 2011); however, this is not always the case. Another reason could be attributed to the measurement of the storm variable; for example, Afonso (2011) uses a dummy variable to indicate severe storms, while Drabo and Mbaye (2015) utilize the number of storms. Another obvious contributor might perhaps be the nonseparation of meteorological disasters, as is the case in Drabo and Mbaye’s (2015) study.

This study seeks to estimate the impact of hurricanes on migration from the CAC region to the United States over the period 1989 to 2005. In undertaking this investigation, this paper contributes to the literature in two ways. First, it adopts a comprehensive measure of hurricane destruction (Strobl 2012), which proxies localized destruction with the use of historical hurricane tracks computed using geographic details for each CAC country. The use of hurricane destruction indices is now part of a growing hurricane literature (Anttila-Hughes and Hsiang 2013; Ishizawa and Miranda 2016) and increases the reliability of estimates. Second, this study explores the immediate or short-term impact of hurricanes on migration decisions. The Panel Data on International Migration 1975–2000, which quantify emigration in narrower temporal bans, were used by Drabo and Mbaye (2015). Any contemporaneous effects that Drabo and Mbaye (2015) find are applied over 5-yr intervals, with no specific information on how events at different points during this interval might impact migration flows on an annual basis. By utilizing immigration statistics from the U.S. Department of Homeland Security, this study can estimate the effects of hurricanes on annual immigration.

In line with the current literature, the analysis reveals a positive impact on migration. However, there is no contemporaneous effect, only a lagged effect, which stands in contrast to Drabo and Mbaye (2015), who found both contemporaneous and lagged effects. The findings also reveal that for more damaging storms, significantly larger increases in migration take place.

The following sections present a discussion of the data and hurricane indices, the empirical strategy, estimated results, future climate and hurricane strikes, and finally, a conclusion.

2. Data

a. Migration and population

The annual U.S. migration data come from the 1989–95 Statistical Yearbook of the United States Immigration and Naturalization Service (U.S. Immigration and Naturalization Service 2017) and the 1996–2006 Office of Immigration Statistics’ Yearbook of Immigration Statistics (U.S. Department of Homeland Security 2017), which are both housed by the U.S. Department of Homeland Security. The migration data accessed from the Statistical Yearbook of the United States Immigration and Naturalization Service and the Yearbook of Immigration Statistics are the “persons obtaining lawful permanent resident status by region and country of birth.” The number of persons obtaining permanent resident status is a count of persons who receive “green cards” in the given year. A green card is legal authorization to reside and work in the United States. Thus, obtaining a green card is an official indication that an individual has migrated to the United States. The number of green card holders from one’s home country can signal the possibility for permanency and employment for hopeful migrants. These data on green card holders are used and explained as one of our control variables, seen below in section 3, Eq. (2).

While the number of new green card holders indicates a change in legal immigration status, it does not directly measure when an immigrant relocated to the United States. As such, this study uses year of immigration data collected from the Integrated Public Use Microdata Series-USA (IPUMS-USA), which uses harmonized U.S. Census microdata to create another measure of migration to the United States. The IPUMS-USA data are collected from the American Community Survey 2001, 2002, 2003, 2004, and 2005 and the U.S. Census 1990 and 2000. The American Community Survey data are designed to collect data on households and population annually, and the U.S. Census is taken every 10 years. The year of immigration data report the specific year an individual entered the United States. Thus, to assess migration flows by year, the authors tabulate the number of individuals who enter the United States in a given year. The process of tabulating these flows required that each individual be identified across years, by the available person number and serial number, to prevent double counting. However, for this reason, we are unable to capture incidences of return migration for the 1990–2000 period. The analysis, then, will focus on the effects related to new migrants. Once all individuals are uniquely identified, the year of immigration frequencies by home country are obtained. This method of extraction allows for annual, rather than decennial or quinquennial, analysis. Narrower spatial periods are useful for studying the short-term impacts of natural disasters such as hurricanes. These annual flows are used to construct the dependent variable (migration rates) in Eq. (2), described in section 3 below.

This paper specifically utilizes series of migration data for Central America and the Caribbean to the United States from 1989 to 2005 for the 30 countries listed in Table 1. The population data are obtained from the U.S. Census Bureau (2015) International Database and are used to scale the migration flows into a percentage of the home country’s population. These population values are based on estimates and projections of the midyear population in each country.

Table 1.

CAC countries included in this analysis.

CAC countries included in this analysis.
CAC countries included in this analysis.

b. Hurricane

Strobl’s (2012) hurricane wind damage index is used to measure the occurrence and strength of hurricanes. This hurricane index is calculated based on estimates of the wind destruction of all hurricanes that strike each of the 30 studied countries in a given year. The wind damage index is a function of the estimate of the wind velocity relative to the eye of the hurricane and weights for the specific characteristics of a location that make it susceptible to damage. Strobl (2012) calculates destruction estimates using

 
formula

where V is the estimate of wind speed at any point, w is the weight of the potential damage based on the characteristic of the place, j is a particular place, J is the set of j locations in a given country i, and λ is “the parameter that links wind speed to its level of destruction.” More specifically, for the weights, Strobl (2012) makes use of population shares of each j at time t − 1, which in this case follows the argument that locations not densely populated are unlikely to have any major impact on migration resulting from a hurricane, even if they experience severe levels of destruction.3 The index constructed from Eq. (1) enters into our model as H, the hurricane variable seen in Eq. (2), and is measured in local population share weighted averaged kilometers per hour cubed. In addition, the validity of our hurricane variable lies in the assumption that it is exogenous, which is fairly reasonable since it is constructed using noneconomic data. The strength of this index is that it provides a measurement for the potential destruction from a hurricane, which allows for a deeper understanding of the damages of these storms beyond just their occurrences. However, it does not account for rainfall and storm surge, which are usually associated with a hurricane. Most of the Central American and Caribbean countries in this study experienced more than three hurricanes during the period of analysis.

c. GDP and other data

The Institute for Health Metrics and Evaluation real gross domestic product (GDP) series, compiled and estimated by James et al. (2012), is used to overcome the difficulty of accessing the GDP for all the countries in the sample for the period of interest. James et al. (2012) collect GDP data from seven sources and use several regression models to estimate missing values to create a database for 210 countries from 1950 to 2015. The series used in this paper is based on real GDP data collected from the World Bank, which has a base year of 2005 and is measured in U.S. dollars. Descriptive statistics for all the variables are shown in Table 2, and all results obtained are interpreted as percentages.

Table 2.

Descriptive statistics. Using the damage value data (EM-DAT 2009) for a sample of CAC countries, the rough estimation associated with the average value of the hurricane index is $889,000,000.

Descriptive statistics. Using the damage value data (EM-DAT 2009) for a sample of CAC countries, the rough estimation associated with the average value of the hurricane index is $889,000,000.
Descriptive statistics. Using the damage value data (EM-DAT 2009) for a sample of CAC countries, the rough estimation associated with the average value of the hurricane index is $889,000,000.

3. Impact of hurricanes on migration

a. Main estimation

The impact of hurricanes on migration for the countries in the CAC region is estimated using the following panel-country fixed-effects model:4

 
formula

where is the log of migration rates by country i and year t. The migration rate is calculated as the ratio of the number of individuals from country i who entered the United States to the population of country i. The variable H is the measure of hurricane destruction, which is calculated over all hurricanes and all localities in each country at times t and t − 1. The hurricane destruction measure is assumed to be exogenous to all countries since the index is constructed without the use of economic data (Strobl 2012). Income (ln) is the log of GDP per capita, which represents GDP per capita growth for country i at time t − 1. The lagged log of GDP per capita is included since previous studies have shown that income growth is a factor that can influence an individual’s decision to migrate (Boubtane et al. 2013). GDP growth is also an appropriate indication of changes in economic opportunity and the labor market (Naudé 2009). In this paper, current income is not considered to be a factor (Bowles 1970) since an individual’s dissatisfaction with income today is less likely to influence an immediate move. The term captures two additional control variables that are used in separate groups of regressions. With X, we seek to control for the effects of the concentration of migrants from similar territories who already live in the United States on the flow of new migrants. Previous studies have shown that new migrants are drawn to areas with immigrants from their host countries (Bartel 1989; Card 2001; Hatton and Williamson 2005; Card and Lewis 2007). Building off the work of Bartel (1989), Jaeger (2007) finds that this pull effect to migration is strengthened in the case of the relatives of legal permanent residents. This is particularly important, as he points out that this group makes up a significant share of the migrants. In the first set of regressions, we explore the impact of previous migrants by representing the number of persons obtaining a U.S. green card as . Here, individuals obtaining legal permanent status likely serve as a pull factor for family members left behind in their country of birth. In the second set of regressions, we measure the impact of the flow of previous migration by using a one-time-period lag of the dependent variable as X. However, including the lagged dependent variable is likely to result in a biased estimate on this regressor (Nickell 1981) and may also result in other biases in the model. As a result, Bruno’s (2005) bias-correction mechanism is used in estimating the model, which produces bootstrapped standard errors and produces more efficient estimates. The lags are limited to t − 1 so as not to reduce the sample size. Additionally, the time and country fixed effects are included to hold constant any peculiarities that might exist across the years and within each country.

In line with the literature, this study finds a positive effect of hurricanes on migration. Table 3 shows the results from estimating Eq. (2). Model 3 shows the results from including the number of green card holders as X in Eq. (2), while model 4 uses the dependent variable as X. As the results indicate, hurricanes increase migration. Table 4 shows that the average hurricane increases migration by 6%5 when considering model 3 with all control variables. It is noted that for more damaging storms, migration increases by 34%. Thus, relative to the mean migration, a 6% increase translates into a mean annual movement of around 3834 individuals, while a 34% increase converts into a mean annual migration of 21 723 individuals.6 Model 4, which corrects for Nickell bias, also gives a similar impact. In general, all models indicate a nonpersistent hurricane effect on migration. The results of this paper are in line with the study of Drabo and Mbaye (2015), which finds that meteorological disasters, inclusive of storms, have a contemporaneous effect on migration. Furthermore, the results in this paper are much smaller than those of Afonso (2011), who finds that with the occurrence of a storm, there is a 16.5% increase in migration from the CAC region. However, the result of Afonso (2011) has been noted as being biased due to the consideration of only the top 10 severe storms in the region, while the analysis in this paper considers all severe storms, specifically all storms that are of category 3 and above.7 Finally, the estimations produced expected results as they relate to income and previous migrants living in the United States, where a 1% increase in income decreases migration by at least 0.45%, and a 1% increase in the number of immigrants in the previous year increases migration by approximately 0.09%.

Table 3.

Impact of hurricane on migration. The dependent variable is the log of the ratio of the number of immigrants that entered the United States from a host country to the population of the host country. Models 1–3 are results from panel fixed-effect regressions. In model 3, Log Xt−1 represents the lag of number of persons who received permanent resident status in the United States. In model 4 Log Xt−1, captures data from the lag of the dependent variable. Thus, this model corrects for the known Nickell bias (Nickell 1981) using Bruno’s (2005) bias-correction mechanism.

Impact of hurricane on migration. The dependent variable is the log of the ratio of the number of immigrants that entered the United States from a host country to the population of the host country. Models 1–3 are results from panel fixed-effect regressions. In model 3, Log Xt−1 represents the lag of number of persons who received permanent resident status in the United States. In model 4 Log Xt−1, captures data from the lag of the dependent variable. Thus, this model corrects for the known Nickell bias (Nickell 1981) using Bruno’s (2005) bias-correction mechanism.
Impact of hurricane on migration. The dependent variable is the log of the ratio of the number of immigrants that entered the United States from a host country to the population of the host country. Models 1–3 are results from panel fixed-effect regressions. In model 3, Log Xt−1 represents the lag of number of persons who received permanent resident status in the United States. In model 4 Log Xt−1, captures data from the lag of the dependent variable. Thus, this model corrects for the known Nickell bias (Nickell 1981) using Bruno’s (2005) bias-correction mechanism.
Table 4.

Summary of hurricane migration impact. This table calculates the impact of hurricanes on migration using the average and maximum values of the hurricane index, where the maximum hurricane represents storms that are more destructive. The calculations correspond to models 1–4 in Table 3.

Summary of hurricane migration impact. This table calculates the impact of hurricanes on migration using the average and maximum values of the hurricane index, where the maximum hurricane represents storms that are more destructive. The calculations correspond to models 1–4 in Table 3.
Summary of hurricane migration impact. This table calculates the impact of hurricanes on migration using the average and maximum values of the hurricane index, where the maximum hurricane represents storms that are more destructive. The calculations correspond to models 1–4 in Table 3.

Overall, the results show that controlling for the influence of past migrants is important to the estimation, leading to a doubling of the effect of the hurricane index on the rate of migration as shown in Table 4, model 3. This section is concluded by observing the robustness results presented in Table 5. Two subsamples, 1989–2000 and 1995–2005, are created from the sample in order to rerun Eq. (2). In both subsamples, several years of data are dropped as a robustness check to see if the signs of the estimated coefficients are impacted by changes to the available observations. The robustness checks support the initial results, which indicate that hurricanes have a positive impact on migration.

Table 5.

Robustness checks. Models 1 and 2 show the results from using model 3 in Table 3, while models 3 and 4 are results from using model 4 in Table 3.

Robustness checks. Models 1 and 2 show the results from using model 3 in Table 3, while models 3 and 4 are results from using model 4 in Table 3.
Robustness checks. Models 1 and 2 show the results from using model 3 in Table 3, while models 3 and 4 are results from using model 4 in Table 3.

b. Geographical effects

The discussion thus far indicates that hurricanes positively impact migration. It is instructive to see how the effects play out in individual countries. Equation (2) is used to achieve this but allows the coefficients to vary by country. Further, it was estimated using only the number of persons receiving permanent resident status by country of birth as in Eq. (3) below. The estimation follows:

 
formula

The coefficient is a vector that contains a value for each country c. Only the results of the contemporaneous hurricane variable are reported since the authors previously found that in general, there are no lagged effects. The results in Table 6 are the average hurricane effects and are only for the countries whose estimates are positive and significant. Here, it is observed that hurricanes can have different impacts across countries. The estimates reveal that the highest rates of migration for the average hurricane take place in Barbados (92%), Dominica (34%), and Guatemala (32%), with the lowest rates of migration taking place in Belize (4%) and the Bahamas (6%). As the table shows, the impact is greater for more destructive storms. One may wonder why the effect of hurricanes is large in some countries and small in others. Perhaps this large difference may be attributed to GDP per capita, although this might not strictly hold for all countries. The large difference noted between Barbados and the Bahamas, for example, could be a result of the difference in GDP per capita, which is notably larger for the Bahamas than for Barbados.8 Income possibly serves as a coping mechanism and thus controls potential external migration. However, caution must be taken since there are other country-specific characteristics that may be showing up in these estimated impacts. Perhaps those with smaller impacts or no impact might be a result of either the country being more resilient to hurricane strikes or because individuals would like to migrate but are unable to do so.

Table 6.

Geographical effects of hurricanes.

Geographical effects of hurricanes.
Geographical effects of hurricanes.

4. Future climate and hurricanes

The literature is keen on what future changes in climate change can have in terms of hurricanes, where projections reveal that warmer weather will result in hurricanes becoming more intense (IPCC 2001; Webster et al. 2005) though less frequent (Knutson et al. 2010). Research generating future climate change scenarios on the cyclonic activities in the North Atlantic points to an increase in the number of stronger storms (Knippertz et al. 2000; Paciorek et al. 2002; Fischer-Bruns et al. 2005). Further, at high levels of carbon dioxide, stronger storms should be even greater in number, although evidence reveals that storms in the northwest Atlantic may have a marginal tendency to increase in severity (Jiang and Perrie 2007). Troublingly, compared to other ocean basins, the North Atlantic, where the Central American and Caribbean countries reside, shows the largest increase in the strongest hurricanes, which are expected to trend upward with warmer weather (Elsner et al. 2008). Such increases in intensity are likely to generate significant costs for countries faced with Atlantic hurricanes (Bertinelli et al. 2016). These projections of future possibilities are worrying for the Central American and Caribbean region. This could imply large increases in migration from the region. Bertinelli et al. (2016) find that the return period for a damaging hurricane is within 10 years for most Caribbean countries. Given this possibility, arguably, one can expect migration after damaging hurricanes, which are likely to strike, given the future increases in intensity.

5. Conclusions

Natural disasters, including hurricanes, remain a challenge for developing countries. Those in the Central American and Caribbean region are no exception, along with other developing countries that continue to experience significant economic damage from hurricane strikes. This paper investigates the relationship between migration and hurricanes using an innovative measure of local destruction at the country level, which has not been employed in the literature to study the impact on migration. Country fixed-effects estimation shows that hurricanes have a positive impact on migration. This is in line with the findings of developing country studies. As mentioned, Drabo and Mbaye (2015) find that meteorological disasters, which include storms, have an effect on migration; the work of Afonso (2011) finds the same directional impact for 10 severe storms. However, a contrasting difference between Afonso (2011) and this study is that we estimate a smaller impact of 6%, compared to 16.5%. We attribute this difference to a lack of comprehensive consideration by Afonso (2011) of all severe storms that affect the CAC region. Further, this paper uses a more inclusive measure of hurricane destruction, which takes into account the characteristics of the storms that tend to cause severe damages, while Afonso’s (2011) measure is a binary variable indicative of whether or not a storm was severe. This binary variable simply captures the incidence of a severe storm and not the important characteristics of a hurricane that would result in destruction.

Finally, as was previously mentioned, hurricanes can be catastrophic, impacting households’ livelihood and property, which can push those impacted to migrate externally. Since this study was unable to directly take account of the mechanisms through which hurricane destruction causes external migration, we are unable to make any policy recommendations as they relate to investments in infrastructure and development aid, which can assist with the aforementioned hurricane losses.

Acknowledgments

The authors of this paper would like to thank Eric Strobl for generating the hurricane indexes used in the analysis.

REFERENCES

REFERENCES
Afonso
,
O. C.
,
2011
:
Natural disasters and migration: Storms in Central America and the Caribbean and immigration to the U.S
.
UC Davis Undergrad. Res. J.
,
14
,
1
18
, http://explorations.ucdavis.edu/docs/2011/andrade.pdf.
Alexeev
,
A.
,
D. H.
Good
, and
R.
Reuveny
,
2011
: Weather-related disasters and international migration. Indiana University Working Paper, 38 pp.
Anttila-Hughes
,
J.
, and
S.
Hsiang
,
2013
: Destruction, disinvestment, and death: Economic and human losses following environmental disaster. SSRN Working Paper, 86 pp.
Bartel
,
A.
,
1989
:
Where do the new U.S. immigrants live?
J. Labor Econ.
,
7
,
371
391
, https://doi.org/10.1086/298213.
Bertinelli
,
L.
,
P.
Mohan
, and
E.
Strobl
,
2016
:
Hurricane damage risk assessment in the Caribbean: An analysis using synthetic hurricane events and nightlight imagery
.
Ecol. Econ.
,
124
,
135
144
, https://doi.org/10.1016/j.ecolecon.2016.02.004.
Boubtane
,
E.
,
D.
Coulibaly
, and
C.
Rault
,
2013
:
Immigration, growth, and unemployment: Panel VAR evidence from OECD countries
.
Labour
,
27
,
399
420
, https://doi.org/10.1111/labr.12017.
Bowles
,
S.
,
1970
:
Migration as investment: Empirical tests of the human investment approach to geographical mobility
.
Rev. Econ. Stat.
,
52
,
356
362
, https://doi.org/10.2307/1926312.
Bruno
,
G.
,
2005
:
Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models
.
Econ. Lett.
,
87
,
361
366
, https://doi.org/10.1016/j.econlet.2005.01.005.
Card
,
D.
,
2001
:
Immigrant inflows, native outflows and the local labor market impacts of higher immigration
.
J. Labor Econ.
,
19
,
22
64
, https://doi.org/10.1086/209979.
Card
,
D.
, and
E. G.
Lewis
,
2007
: The diffusion of Mexican immigrants during the 1990s: Explanations and impacts. Mexican Immigration to the United States, G. J. Borjas, Ed., University of Chicago Press, 193–228.
Drabo
,
A.
, and
L. M.
Mbaye
,
2015
:
Natural disasters, migration and education: An empirical analysis in developing countries
.
Environ. Dev. Econ.
,
20
,
767
796
, https://doi.org/10.1017/S1355770X14000606.
Elsner
,
J. B.
,
J. P.
Kossin
, and
T. H.
Jagger
,
2008
:
The increasing intensity of the strongest tropical cyclones
.
Nature
,
455
,
92
95
, https://doi.org/10.1038/nature07234.
EM-DAT
,
2009
: The emergency events database. Centre for Research on the Epidemiology of Disasters, accessed 20 October 2017, www.emdat.be.
Findlay
,
A. M.
,
2011
:
Migrant destinations in an era of environmental change
.
Global Environ. Change
,
21
,
S50
S58
, https://doi.org/10.1016/j.gloenvcha.2011.09.004.
Fischer-Bruns
,
I.
,
H.
von Storch
,
J. F.
González-Rouco
, and
E.
Zorita
,
2005
:
Modelling the variability of midlatitude storm activity on decadal to century time scales
.
Climate Dyn.
,
25
,
461
476
, https://doi.org/10.1007/s00382-005-0036-1.
Fussell
,
E.
,
S. R.
Curran
,
M. D.
Dunbar
,
M. A.
Babb
,
L.
Thompson
, and
J.
Meijer-Irons
,
2017
:
Weather-related hazards and population change: A study of hurricanes and tropical storms in the United States, 1980–2012
.
Ann. Amer. Acad. Pol. Soc. Sci.
,
669
,
146
167
, https://doi.org/10.1177/0002716216682942.
Halpern-Manners
,
A.
,
2011
:
The effect of family member migration on education and work among nonmigrant youth in Mexico
.
Demography
,
48
,
73
99
, https://doi.org/10.1007/s13524-010-0010-3.
Hatton
,
T. J.
, and
J. G.
Williamson
,
2005
: What fundamentals drive world migration? Poverty, International Migration and Asylum, G. Borjas and J. Crips, Eds., Palgrave-Macmillan, 15–38.
Hsiang
,
S. M.
,
2010
:
Temperatures and cyclones strongly associated with economic production in the Caribbean and Central America
.
Proc. Natl. Acad. Sci. USA
,
107
,
15 367
15 372
, https://doi.org/10.1073/pnas.1009510107.
IOM
,
2017
: Migration, climate change and the environment: A complex nexus. International Organization for Migration, https://www.iom.int/complex-nexus#estimates.
IPCC
,
2001
: Climate Change 2001: The Scientific Basis. Cambridge University Press, 881 pp.
Ishizawa
,
O.
, and
J.
Miranda
,
2016
: Weathering storms: Understanding the impact of natural disasters on the poor in Central America. World Bank Policy Research Working Paper WPS7692, 25 pp.
Jaeger
,
D. A.
,
2007
: Green cards and the location choices of immigrants in the United States, 1971–2000. Immigration, B. Chiswick, Ed., Emerald Group Publishing Ltd, 131–183.
James
,
S. L.
,
P.
Gubbins
,
C. J. L.
Murray
, and
E.
Gakidou
,
2012
:
Developing a comprehensive time series of GDP per capita for 210 countries from 1950 to 2015
.
Pop. Health Metrics
,
10
, https://doi.org/10.1186/1478-7954-10-12.
Jiang
,
J.
, and
W.
Perrie
,
2007
:
The impacts of climate change on autumn North Atlantic midlatitude cyclones
.
J. Climate
,
20
,
1174
1187
, https://doi.org/10.1175/JCLI4058.1.
Knippertz
,
P.
,
U.
Ulbrich
, and
P.
Speth
,
2000
:
Changing cyclones and surface wind speeds over the North Atlantic and Europe in a transient GHG experiment
.
Climate Res.
,
15
,
109
122
, https://doi.org/10.3354/cr015109.
Knutson
,
T. R.
, and Coauthors
,
2010
:
Tropical cyclones and climate change
.
Nat. Geosci.
,
3
,
157
163
, https://doi.org/10.1038/ngeo779.
Kugler
,
A.
, and
M.
Yuksel
,
2008
: Effects of low-skilled immigration on U.S. natives: Evidence from Hurricane Mitch. NBER Working Paper 14293, 40 pp., https://doi.org/10.3386/w14293.
Lesser
,
G.
, and
J.
Batalova
,
2017
: Central American immigrants in the United States. Migration Policy Institute, 5 April 2017, https://www.migrationpolicy.org/article/central-american-immigrants-united-states.
Logan
,
J. R.
, and
Z.
Xu
,
2015
:
Vulnerability to hurricane damage on the U.S. Gulf Coast since 1950
.
Geogr. Rev.
,
105
,
133
155
, https://doi.org/10.1111/j.1931-0846.2014.12064.x.
Logan
,
J. R.
,
S.
Issar
, and
Z.
Xu
,
2016
:
Trapped in place? Segmented resilience to hurricanes in the Gulf Coast, 1970–2005
.
Demography
,
53
,
1511
1534
, https://doi.org/10.1007/s13524-016-0496-4.
Mahajan
,
P.
, and
D.
Yang
,
2017
: Taken by storm: Hurricanes, migrant networks, and U.S. immigration. NBER Working Paper 23756, 44 pp., https://doi.org/10.3386/w23756.
McGirk
,
J.
,
2000
: Forgotten million still reeling from Hurricane Mitch. The Independent, 16 March 2000, https://www.independent.co.uk/news/world/americas/forgotten-million-still-reeling-from-hurricane-mitch-282390.html.
McIntosh
,
M. F.
,
2008
:
Measuring the labor market impacts of Hurricane Katrina migration: Evidence from Houston, Texas
.
Amer. Econ. Rev.
,
98
,
54
57
, https://doi.org/10.1257/aer.98.2.54.
McLeman
,
R. A.
, and
B.
Smit
,
2006
:
Migration as an adaptation to climate change
.
Climatic Change
,
76
,
31
53
, https://doi.org/10.1007/s10584-005-9000-7.
McLeman
,
R. A.
, and
L. M.
Hunter
,
2010
:
Migration in the context of vulnerability and adaptation to climate change: Insights from analogues
.
Wiley Interdiscip. Rev: Climate Change
,
1
,
450
461
, https://doi.org/10.1002/wcc.51.
Mohan
,
P.
,
2016
:
Impact of hurricanes on agriculture: Evidence from the Caribbean
.
Nat. Hazards Rev.
,
18
, https://doi.org/10.1061/(ASCE)NH.1527-6996.0000235.
Morris
,
S. S.
,
O.
Neidecker-Gonzales
,
C.
Carletto
,
M.
Munguía
,
J. M.
Medina
, and
Q.
Wodon
,
2002
:
Hurricane Mitch and the livelihoods of the rural poor in Honduras
.
World Dev.
,
30
,
49
60
, https://doi.org/10.1016/S0305-750X(01)00091-2.
Naudé
,
W.
,
2009
:
Natural disasters and international migration from sub-Saharan Africa
.
Migr. Lett.
,
6
,
165
176
.
Nickell
,
S.
,
1981
:
Biases in dynamic models with fixed effects
.
Econometrica
,
49
,
1417
1426
, https://doi.org/10.2307/1911408.
Ouattara
,
B.
, and
E.
Strobl
,
2014
:
Hurricane strikes and local migration in US coastal counties
.
Econ. Lett.
,
124
,
17
20
, https://doi.org/10.1016/j.econlet.2014.03.025.
Paciorek
,
C. J.
,
J. S.
Risbey
,
V.
Ventura
, and
R. D.
Rosen
,
2002
:
Multiple indices of Northern Hemisphere cyclone activity, winters 1949–99
.
J. Climate
,
15
,
1573
1590
, https://doi.org/10.1175/1520-0442(2002)015<1573:MIONHC>2.0.CO;2.
Paul
,
B. K.
,
2005
:
Evidence against disaster‐induced migration: The 2004 tornado in north‐central Bangladesh
.
Disasters
,
29
,
370
385
, https://doi.org/10.1111/j.0361-3666.2005.00298.x.
Schultz
,
J.
, and
J. R.
Elliott
,
2013
:
Natural disasters and local demographic change in the United States
.
Popul. Environ.
,
34
,
293
312
, https://doi.org/10.1007/s11111-012-0171-7.
Sherburne
,
M.
,
2017
: Hurricanes drive immigration to the U.S.—Helped by green cards. Phys.org, 12 September 2017, https://phys.org/news/2017-09-hurricanes-immigration-ushelped-green-cards.html#jCp.
Spencer
,
N.
,
2016
:
Look what the hurricanes just blew in: Analyzing the impact of the storm on criminal activities
.
J. Crime Justice
,
40
,
417
429
, https://doi.org/10.1080/0735648X.2016.1168749.
Spencer
,
N.
, and
S.
Polachek
,
2015
:
Hurricane watch: Battening down the effects of the storm on local crop production
.
Ecol. Econ.
,
120
,
234
240
, https://doi.org/10.1016/j.ecolecon.2015.10.006.
Strobl
,
E.
,
2012
:
The economic growth impact of natural disasters in developing countries: Evidence from hurricane strikes in the Central American and Caribbean regions
.
J. Dev. Econ.
,
97
,
130
141
, https://doi.org/10.1016/j.jdeveco.2010.12.002.
Tacoli
,
C.
,
2009
:
Crisis or adaptation? Migration and climate change in a context of high mobility
.
Environ. Urban.
,
21
,
513
525
, https://doi.org/10.1177/0956247809342182.
UNFAO
,
2001
: Analysis of the medium-term effects of Hurricane Mitch on food security in Central America. FAO Rep., https://reliefweb.int/report/belize/analysis-medium-term-effects-hurricane-mitch-food-security-central-america.
U.S. Census Bureau
,
2015
: International database. U.S. Census Bureau, https://www.census.gov/programs-surveys/international-programs/about/idb.html.
U.S. Department of Homeland Security
,
2017
: Yearbook of Immigration Statistics (years cited: 1996−2006). U.S. Department of Homeland Security, https://www.dhs.gov/immigration-statistics/yearbook.
U.S. Immigration and Naturalization Service
,
2017
: Statistical Yearbook of the Immigration and Naturalization Service (years cited: 1989−1995). U.S. Government Printing Office, https://www.dhs.gov/immigration-statistics/yearbook.
Warner
,
K.
,
2010
: Assessing institutional and governance needs related to environmental change and human migration. Study Team on Climate-Induced Migration Rep., 19 pp., http://archive.unu.edu/africa/activities/files/Warner_K_2010_Assessing_Institutional_and_Governance_Needs_Related_to_Environmental_Change.pdf.
Webster
,
P. J.
,
G. J.
Holland
,
J. A.
Curry
, and
H.-R.
Chang
,
2005
:
Changes in tropical cyclone number, duration, and intensity in a warming environment
.
Science
,
309
,
1844
1846
, https://doi.org/10.1126/science.1116448.
Zong
,
J.
, and
J.
Batalova
,
2016
: Caribbean immigrants in the United States. Migration Policy Institute, 14 September 2016, https://www.migrationpolicy.org/article/caribbean-immigrants-united-states.

Footnotes

© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

2

Strobl (2012) shows that Vjt > 177 has an impact on economic growth, hence the choice for this value for hurricane wind speed.

3

Population density data are used directly by Strobl (2012) and are obtained from the Latin America and Population Database. The population density data are measured at 1-km grid cells, and the wind speed is measured at the same scale; said differently, the wind speeds are measured at their centroids.

4

There are factors that determine international migration that are not observable (e.g., individual aspiration and expectations; Halpern-Manners 2011). Thus, we use a fixed-effects model to control for unobserved location-specific characteristics that can influence migratory decisions.

5

It is possible that the result takes account of those from the CAC region who were granted temporary protected status (TPS) visas following Hurricane Mitch in 1998 since TPS recipients can adjust their statuses to permanent residents through an employer or relative.

6

To obtain these numbers, we calculated out the migrant numbers for the countries using 6% and 34%, each being relative to the mean migration.

7

The Saffir–Simpson scale defines categories 3–5 as major storms, which we interpret to be severe based on the levels of destruction that are possible.

8

Based on the sample period, the dataset shows that the average GDP per capita for Bahamas is $19,091 and $9953 for Barbados.