Dry growing seasons predicted Central American migration to the US from 2012 to 2018


Existing research has shown that unusual weather variability and unpredictable seasons has contributed to migration globally5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22. We confirmed that this is also true for recent Central American migration to the US. In developing countries with economies heavily reliant on agriculture, declining harvest yields threaten farming livelihoods when investments surpass revenues29,30,31. These deficits force millions of households worldwide to adapt: “Negative shocks to agricultural productivity caused by climate fluctuations significantly increase emigration from developing countries”32.

The economy and the environment are tightly coupled across NTCA. In Guatemala, Honduras, and El Salvador, agriculture, including coffee, corn, and bean cultivation, employed 32%, 37%, and 30% of the working population in 2012, according to the International Labor Organization. These rates are proportionally higher in rural regions of each country. By 2019, the agricultural shares of employment nationally had fallen to 31% (Guatemala), 29% (Honduras), and 16% (El Salvador). Even people who are not farmers but who earn wages harvesting or processing agricultural goods could still lose their livelihoods as yields fall.

Closely comparing our data to NTCA farmers’ experiences reassures us that drier than average growing seasons and emigration were plausibly related. For example, in 2016, the Las Marias community in Olancho department, Honduras, showed journalists and relief workers dry wells that no longer supported their families’ crops33. In the same year, the Famine Early Warning Systems Network expressed grave concerns about hunger in Olancho and other nearby areas34. Our data reflect these conditions, with Olancho experiencing a drier than usual growing season (SPEI03 = −1.18 SD). Furthermore, Olancho’s 2016 emigration rate (379.8 people per 100,000) was above the 90th percentile (287.2 per 100,000).

Similarly, crops in San Miguel, El Salvador, withered without rain during 2016, and residents there reported to relief agencies that conditions were too dry even for alternative varietals35. San Miguel farmers eventually relied exclusively on emergency food supplies. As in Olancho, our data reflect these narratives. The average SPEI03 during the 2016 growing season was  − 1.41 SD that year, and the emigration rate was 624.1 people per 100,000 (three times the regional average). In another example, the Red Cross warned of dire drought conditions in Guatemala during the 2015 El Niño. During El Niño, Guatemala experiences canícula or veranillo, a phenomenon when it stops raining for several weeks or more during the rainy season (also called a “midsummer drought”)36. Among other departments, Huehuetenango is named in their reporting of “3.5 million people in need of assistance”37. Our data effectively captured the drought (SPEI03 = −1.7 SD), and we also observed a correspondingly high emigration rate (529.6 per 100,000 people). These illustrative examples demonstrate how lower than average rainfall impacts livelihoods and, in turn, plausibly raised emigration rates.

Many in NTCA supplement their diets with subsistence crops grown on small plots. Abnormally dry weather can lead to food security challenges through lost farm revenues among smallholders, diminished wages, or failing personal gardens. The World Food Program and International Organization for Migration recently called attention to rising food insecurity among nearly 3.5 million people living in the “dry corridor” that runs through all three countries38. In addition, ethnographic research and household survey data analysis confirm that food insecurity is a dire concern under drying weather conditions in the region39.

The adverse effects of climate change beyond water shortages also contribute to crop failures. Across NTCA, forecasts have shown reductions in the region’s suitability for coffee, a crop that provides the largest share of rural employment40. For example, rising temperatures have contributed to plant diseases like “coffee rust” and aflatoxin41. Coffee rust is caused by a fungus (Hemileia vastatrix) that affects the coffee species most common in El Salvador, Guatemala, and Honduras (Arabica). Higher normally-cool and lower normally-hot monthly temperature averages reduce the latent period of the Hemileia vastatrix fungus. In 2013, severe coffee rust caused over $1 billion in damage42 and reduced yields by 16% compared with 2011–201243. We analyzed potential evapotranspiration indicators (e.g., SPEI03) incorporating temperature trends that account for these ecological impacts.

Independent of agricultural trends in NTCA, global coffee prices fell after 2011 with a temporary spike in 2014–201544 and only a recent increase in 2021. Coffee rust episodes and low average prices combine with short-term weather variability to undermine farmers’ livelihoods. Rising fertilizer, equipment, and transportation costs, unfavorable international currency exchange rates, and a host of other production and supply-chain dynamics13 influence migration decisions in addition to the weather, but we accounted for such temporal trends in our INLA and TWFE OLS models.

Studies have found that environmental stress proxied by temperature and precipitation extremes is more likely to lead to international emigration from rural areas than internal migration to other domestic cities45. Such conclusions about the scope of migration suggest that planning to travel to the US from NTCA, rather than remaining in-country, is plausible for struggling farmers. Strong global community networks and anticipated climate-related wage changes may account for these differences.

Migration responses to climate change are complex. Under certain circumstances, relocating can be a highly productive adaptation strategy31,46,47,48. On the one hand, an intuitive scenario links drought to migration, which can occur through agriculture shocks: “municipal-level rainfall deficits relative to historical averages are an important predictor for both international and internal migration, especially in municipalities with predominantly rainfed agriculture”19. On the other hand, scholars have cautioned against reductive, deterministic, and overly-simplistic interpretations of this chain of events because migration responses to climate change are not inevitable. It is beyond the scope of this study, for example, to study so-called “trapped” households, who may lack the financial, institutional, or social capital that facilitates relocation49,50,51. Of course, securitized borders also make international migration dangerous and expensive52. Still, this fact only suggests that our family apprehensions data are a conservative estimate of how severely droughts have affected NTCA communities.

Beyond accounting for secular temporal trends in our analysis, our study accounted for influential national-level political circumstances that might correlate with migration. El Salvador, Guatemala, and Honduras have long histories of autocratic rule, for example, and the legacies of US interventions have contributed to chronic political instability and weak governance. The three countries currently have limited state capacity, and investments in the provision of public goods are low. In 2018, the NTCA countries ranked last in Latin America for collecting tax revenue on GDP, according to the World Bank, which undermines effective governance.

In addition to the impacts of climate change, criminal violence and insecurity were among the most convincing alternative explanations for emigration from Central America to the US. In 2018, El Salvador, Honduras, and Guatemala had murder rates of 51.0, 40.0, and 22.4 per 100,000 people, respectively, compared to 5.0 in the US38. According to a 2015 US Senate Homeland Security and Government Affairs report, violence and insecurity topped the list of causes for NTCA migration53. The Internal Displacement Monitoring Centre reported that 242,000 and 247,000 people in Guatemala and Honduras, respectively, were displaced by violence as of 2020 (statistics are not available for El Salvador). Population survey research with over 49,000 respondents across Latin America found that the probability of “seriously considering family migration to the US” was 30% higher among families who were violence victims than households who were not54. Yet, there is evidence in some Honduran research that out-migration is primarily a function of material conditions (access to service and human capital accumulation) and less a result of high crime rates55. In our analysis isolating the impact of dry weather, we use official national police agency homicide data to control for these effects of violence.

Tables S5S10 show that the INLA model posterior estimate for homicide rate is always reliably positive and distinct from zero. The effect on emigration is relatively small, however. The relationship also cannot be interpreted as plausibly exogenous from social context (as is the case for SPEI03). For example, in preferred model 4 (Table S5), every additional homicide per 100,000 people increases emigration by less than 1.0% (e0.005). Operationalizing the homicide rate using a binary indicator for departments that lie above the average (instead of the population normalized rate) confirms this relatively low magnitude effect (see Table S11). Holding covariates at their mean, a department with more violence than average will have 20.5% (e0.187) more emigration to the US. In these models, the credible effect of dry weather remains (e0.662) and has a stronger influence on emigration rates than violence.

Our results are robust, yet data limitations present opportunities for further research on this important topic. We use FOIA Border Patrol data because comprehensive, cross-national, and longitudinal household survey data are not available in the NTCA countries. Such data would allow scholars to evaluate our findings at a more granular level. One could definitively establish, for example, whether farming households suffering financial losses during droughts were more likely to be the same people arriving at the US border. More generally, the data generating process in our analysis begins with US Border Patrol apprehensions. Selection into the database requires successfully completing a long and arduous journey. Future research about NTCA migration during droughts could focus on communities relocating domestically, though such data may be difficult to obtain.


Source link

Be the first to comment

Leave a Reply

Your email address will not be published.