I. Introduction

Over time a steady influx of migrants from the Northern Triangle countries (NT: El Salvador, Guatemala, and Honduras) have settled in the United States, mostly undocumented. Out of the 3 million migrants of NT origin living in the U.S in 2019, more than half were undocumented, representing the second largest demographic group of undocumented migrants in the U.S. (Center for Migration Studies, 2022). The main mode of arrival is undocumented entry across the U.S. Southwest border, and during 2014–19 migrants from the NT countries represented the largest group arriving undocumented at the U.S. Southwest border.

This steady increase in the population of undocumented NT migrants to the U.S. occurred against the backdrop of tighter U.S. immigration policy and enforcement measures. Between 1990 and 2019 the U.S. border enforcement tightened substantially, including due to: (i) an increase in the number of border agents (which increased by nearly 400 percent), (ii) greater coverage of the border wall barrier (which increased by over 240 percent, to 654 miles), and (iii) expanded coverage of versatile border surveillance equipment. At the same time, undocumented migration from the NT countries increased almost four-fold, from half million in 1990 to 1.8 million by 2019.

Understanding the factors that drive undocumented migration has important policy implications for NT countries, as well as broadly for other countries which are experiencing increased emigration. Understanding migration more broadly has macro-critical implications for the economy, as migrants send remittances home, and emigration reduces the domestic labor supply, and thereby could lower economic growth.¹ Media has documented well the human costs of the perilous journey to reach the U.S. for an undocumented migrant, especially from the NT countries. Most importantly, these articles—based on individual interviews—as well as U.S. Customs and Border Protection (CBP) Border Patrol reports present anecdotal evidence of various factors driving undocumented migration from the NT countries, citing frequently among other causes high crime, especially gang related, loss of livelihood due to low commodity prices, droughts, and frequent natural disaster events. The Migration Policy Institute shows in 2022 based on survey data that economic conditions—wages/incomes, unemployment, and job prospects—represent the vast majority of the motivating factors for the decision to migrate. Amid slowly emerging empirical studies on specific factors, a more rigorous analysis estimating the role of all drivers of migration is needed, filing a gap in literature.²

This paper disentangles the role of various drivers in explaining the increasing undocumented migration from the NT countries to the U.S. To gauge undocumented migration—an unobserved variable— we use data on the U.S. border apprehensions, similar to Hanson and Spilimbergo (1999).³ To assess the role of various factors we employ explanatory variables consistent with an investment decision theory model, with migrants assessing if the quantifiable benefits from undocumented migration are higher than its costs. As such, along with the traditional economic factors (wages in the host compared to home country, and unemployment rates, in the NT and the U.S., as in Hanson and Spilimbergo, 1999) we also control for the role of the probability of being apprehended (measured by the U.S. border enforcement). Moreover, we also control for non-traditional factors of migration, including changes in coffee production, changes in temperature, homicide rates, and natural disaster events. We also account for sudden spells of undocumented migration.

Our findings have important policy implications. We find that undocumented migration of people of NT country of origin are well explained by both traditional factors—such as indicators of income (real wages or GDP per capita) and indirect measure of probability being apprehended—as well as non-traditional factors. We find that, as survey data suggest, economic considerations (real wages and unemployment rates, especially in the U.S). play an important role in driving undocumented migration. Most importantly, our results suggest that border enforcement measures act as a deterrent for undocumented migration. However, we also find that large and sudden increases in border apprehensions are at times linked to favorable changes in the perception of probability of being apprehended. The results are robust to various specifications, estimation methods, and to different measures of income, probability of apprehension, decline in livelihoods due to coffee production and changes in temperature, as well as to additional explanatory variables. Changes at home should be centered around providing conditions for sustained inclusive growth, to create enough jobs, and advancing economic transformation to ensure higher wages as well as livelihood resilience to climate change and natural disasters, along with reducing crime.

This paper is organized as follows. In Section II, we present the model and in section III we describe the data used (and some styled facts about this data). The empirical analysis is presented in Section IV. Section V presents the concluding remarks, outlook, and policy implications.

II. Model

Our empirical analysis uses the investment decision theory and border apprehensions data. According to the investment decision theory, an individual decides to migrate when expected discounted difference in the stream of income between the new and old location exceeds the moving costs (see Sjaastad, 1962). Furthermore, for a migrant who decides to migrate undocumented, the decision also depends on the difference between the perceived probability of being apprehended today, as well as in the future. Empirical application of the investment decision theory into analysis of undocumented migration is challenging, as both the arrival of undocumented migrants and the probability of apprehension are not directly observed variables.⁴ Similar to Hanson and Spilimbergo (1999), we used the number of border apprehensions as an indirect indicator of undocumented migration.⁵ Annex I presents the theoretical underpinnings of the model, and Annex II details why border apprehensions is the best measure of flows of undocumented migration.

We start with the reduced form model of Hanson and Spilimbergo (1999). In this specification, border apprehensions are modeled as a function of factors of the potential stream of income in the U.S., factors of the forsaken stream of income in the country of the origin, and indicators of migrants’ perceptions of contemporaneous and future probabilities of being apprehended. Potential stream of income is associated with gains in real income following a move to the U.S., while forsaken stream of income is associated with the real income the migrant could have earned if (s)he stayed in the country of origin. As gains in real income materialize only if the migrant finds a job, or the real income is forsaken only if the migrant has a job, the decision to emigrate would also depend on the unemployment rate in both U.S. and the country of origin. Contemporaneous and future probabilities of being apprehended, while not directly observed, could be indirectly gauged by border enforcement measures and changes in the U.S. immigration policy, including legal admissions and legalization of undocumented migrants.⁶

We extend this model with additional drivers of undocumented migration from NT countries. First, we added variables affecting the livelihoods in NT countries, which were also cited by apprehended migrants as reasons to migrate, including crime rate, coffee production, natural disaster events, and climate change.⁷ Second, we further expanded the set of variables to capture indirectly the impact of probability of apprehension, by controlling for the recidivism rate of deportations on border apprehensions. Third, we control for perceptions in the change of U.S. migration policy. In the two recent episodes of migration waves, the first wave occurred in FY2012-FY2014, following the announcement of the Deferred Action for Childhood Arrivals (DACA), and the second wave which peaked in FY2019, in anticipation of announced changes in the U.S. immigration policy.^{8 9}

The baseline model explaining the annual border apprehensions of NT-born migrants at the U.S. Southwest border is represented by the following regression specification, with home and host factors:

Potential stream of income in the U.S. (including possibility of earning such an income):

• ■ W_US,t is the median real wage for Hispanics in the U.S.;

• ■ U_US,t is the U.S. Hispanic unemployment rate;

Forsaken stream of income in the NT countries:

• ■ W_NT,t is the average real wage in the NT country;

• ■ U_NT,t is the unemployment rate in the NT country;

Probability of being apprehended:

• ■ BA_US,t is the number of border enforcement agents at the U.S. Southwest border;

• ■ Deport_NT,t is the number of deportations of undocumented migrants of NT country origin from the U.S.;

Perceptions of changes in the U.S. migration policy:

• ■ Wavel_NT,t is an immigration wave 1 in FY2012-FY2014, a dummy variable;

• ■ Wave2_NT,t is an immigration wave 2 in FY2019, a dummy variable;

• ■ LA_NT,t is the number of legal admissions from the NT country to the U.S.;

Other NT-specific factors:

• ■ Coffee_NT,t is coffee production in the NT country;

• ■ Homicide_NT,t is the homicide rate per 100,000 people in the NT country;

• ■ Temp_NT,t is the two-year average deviation of temperature in March-August from historical trend (1951–1980) in the NT country;

• ■ Disaster_NT,t is a dummy variable to control for natural disaster event in the NT country;

• ■ Trend_t is a time trend; and

• ■ v_NT is country fixed effect and ε_NT,t is an error term.

All variables have been transformed in natural logarithm.

III. Data and Stylized Facts

Apprehensions of citizens of NT origin at the U.S. Southwest border (the main port of entry in the U.S.) has steadily increased over the past 40 years, accelerating at times. In the 1990s and early 2000s three episodes of natural disasters stand out for the region: the volcano eruption in Guatemala, the Mitch Hurricane in 1998 with severe impact in Honduras, 2001 earthquakes in El Salvador, followed by devastating floods. More recently, a prolonged drought in the region affected agricultural production, including coffee production, adversely impacting the livelihoods of many NT citizens. Following these episodes, one could observe an increase in undocumented migration (as measured by apprehensions at U.S. Southwest border).

There are a few recent periods in which apprehensions of NT citizens increased rapidly which cannot be explained by natural disasters. These episodes were observed during 2003–2005, 2012–2014, and 2017–2019. In FY2005, a total of 112,000 citizens of NT countries were apprehended at the U.S. Southwest border. Since FY2005 border apprehensions of NT citizens plateaued at 40,000 people per year on average until September 2017. After October 2017 U.S. Southwest border apprehensions of NT citizens increased 170 percent until September 2019 (reaching a record high of 608,000 people in FY2019). To put things into perspective, apprehensions reached 2.6 percent of home population for Honduras in 2019. While this relatively recent spike is part of a larger phenomenon, data shows that apprehensions of NT citizens have increased against the background of declining total apprehensions of all citizens at the U.S. border since 2000¹⁰ (see Figure 1).

Evolution of Apprehensions at U.S. Southwest Border ^1/

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: DHS, U.S. CBP, and authors’ calculations.^1/ In U.S. fiscal year runs from October of the previous year to September of the current year, 12 months. Nationwide apprehensions includes apprehensions at all US borders. Includes Title 8 apprehensions and Title 42 apprehensions and immediate expulsions.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Evolution of Apprehensions at U.S. Southwest Border ^1/

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: DHS, U.S. CBP, and authors’ calculations.^1/ In U.S. fiscal year runs from October of the previous year to September of the current year, 12 months. Nationwide apprehensions includes apprehensions at all US borders. Includes Title 8 apprehensions and Title 42 apprehensions and immediate expulsions.

• Download Figure
• Download figure as PowerPoint slide

Evolution of Apprehensions at U.S. Southwest Border ^1/

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: DHS, U.S. CBP, and authors’ calculations.^1/ In U.S. fiscal year runs from October of the previous year to September of the current year, 12 months. Nationwide apprehensions includes apprehensions at all US borders. Includes Title 8 apprehensions and Title 42 apprehensions and immediate expulsions.

• Download Figure
• Download figure as PowerPoint slide

However, since the COVID-19 pandemic, apprehensions of NT citizens moderated. There was a sharp decrease in border apprehensions in FY2020—due to travel restrictions and lockdowns. While apprehensions of NT citizens rebounded in 2021, they remained well below FY2019 levels (left chart).¹¹ Most importantly, while the NT citizens represented the highest number of apprehended people during 2014–19, since FY2021 they represent the third largest group, below the number of apprehensions of citizens of Mexico (which was the case historically) and other countries (including Haiti, Cuba, and Nicaragua).¹²

The data was collected from public sources as follows. For U.S. Hispanic wages and unemployment rate we used data from Haver Analytics. For the home country wages, we used real average wages reported by the Economic Commission for Latin America and the Caribbean (ECLAC) and unemployment rate as reported by the UN International Labor Organization of the United Nations (ILO) database. Homicide rates for each of the NT countries are reported by the World Bank. Coffee production by country is reported by the International Coffee Organization. Data on temperature deviations were taken from the UN Statistics Department of the Food and Agriculture Organization of the United Nations (FAOSTAT) Temperature Change database. We used data from the EM-DAT International Disaster Database, to distinguish years with natural disaster events. We constructed a dummy variable, which is set to 1 for earthquakes in El Salvador in 2001, Honduras in 2009, volcano eruptions in El Salvador in 2005, in Guatemala in 1995, 2010, and 2018; and 0 otherwise. We used U.S. CBP’s data on annual border apprehensions of citizens of Northern Triangle countries – El Salvador, Guatemala, and Honduras. We used U.S. Department of Homeland Security (DHS) annual statistical reports to collect data on legal admissions and deportations.

The sample used is 1994–2019, given the availability of data, and the different structural breaks after 2019. As was documented in several official and media reports, the post-pandemic increase of apprehensions is in some part due to Title 42 expediated expulsions, which have a high recidivism rate. Moreover, during this period apprehensions at the U.S. Southwest border were primarily under Title 8. Annex III presents the data and stylized facts in more detail.

IV. Estimation Strategy and Results

The empirical strategy controls for several estimation issues. First, to account for the presence of serial autocorrelation, as shocks to the border apprehensions tend to linger, we estimate the model for the panel of three countries—El Salvador, Guatemala, and Honduras—using the Feasible Generalized Least Squares method with serial autocorrelation in panels. Second, to address potential identification issue—as border enforcement measures could be simultaneously determined with border apprehensions—we estimate the baseline specification in the robustness section also: (i) without border enforcements measures; (ii) using the instrumental variables regression approach; and (iii) using the residual approach.

The baseline specification explains well the evolution of border apprehensions of NT undocumented migrants. Results are shown in Table 1:

• ■ Potential stream of income: U.S. real wage and job opportunities play a very important role in explaining border apprehensions. The elasticity with respect to the U.S. Hispanic real median wage is positive and large, which means that migration attempts are highly sensitive to changes in the median real wages for U.S. Hispanics. Labor demand and job opportunities in the U.S. are significant drivers of undocumented immigration as well. The elasticity with respect to the U.S. Hispanic unemployment rate is negative and significant and large, showing that as employment conditions deteriorate apprehensions also moderate. Those two explanatory coefficients are the second and third highest coefficients of the regression.

• ■ Forsaken stream of income factors also explain well border apprehensions. The elasticity with respect to home country real wage is negative and significant. Its magnitude is smaller than that with respect to the U.S. Hispanic real median wage, which means that home earnings have a smaller effect in deterring undocumented immigration than the pull factors. The elasticity with respect to the home country unemployment rate is positive, significant, and large, with the absolute value greater than that for the U.S. Hispanic unemployment rate (in fact it is the highest coefficient), meaning that home country labor market conditions and job opportunities play the biggest role in driving undocumented migration.¹³ This latter result is consistent with Migration Policy Institute, 2022, and other sources (e.g. local articles), documenting that one of the most important causes of migration is driven by the fact that the number of jobs created in NT countries every year is much lower than the number of entrants in the labor market.

• ■ Indirect measures of probability of being apprehended also contributed to the evolution of border apprehensions. Border enforcement, measured by the number of border agents, is significant and negatively associated with apprehensions. Higher border enforcement measures increase the probability of apprehension, which in turn deters attempts of migration and thus may leads to lower border apprehensions.¹⁴Legal admissions are found to have no effect on border apprehensions, and by extension on attempts of undocumented migration.¹⁵ Deportations have a significant and positive effect on border apprehensions. Once a deportation occurs, the deported migrant perceived probability of being apprehended in the future increases. Hence, the migrant might attempt to migrate soon after deportation, rather than waiting or giving up. The result is consistent with anecdotal evidence as well as calculations by the DHS and other of the recidivism rate, as well as the coyote costs of helping migrants cross the border (with a paid sum including multiple attempts). For more stylized facts of border enforcement and the probability of being apprehended, see Annex II.

• ■ Perception of changes in the U. S. migration policy are important in explaining evolution of border apprehensions. The migration wave dummies have positive coefficients and significant. The first immigration wave followed the DACA announcement in 2012. The result is consistent with media reports that many apprehended NT migrants perceived these announcements as favorable, resulting in an influx of caravans of families and unaccompanied children attempting to migrate undocumented. The second immigration wave was observed in 2019 and was triggered by the U.S. government tightening border enforcement measures, and anti-immigration political rhetoric, causing a change in the migrants’ perceptions of incoming tightening of the U.S. migration policy, and in turn might have determined some migrants to move forward their decision to migrate, or to decide to migrate, before the costs of migration would be too high/prohibitory.¹⁶

• ■ NT-specific and non-income factors are important drivers of border apprehensions. Elasticity with respect to coffee production is negative and significant, suggesting that greater production of this cash crop provides disincentive for undocumented migration attempts. Elasticity with respect to homicides is positive and significant, suggesting that higher crime pushes out undocumented migration. We also find that elasticity with respect to temperature deviation is positive (and significant at 10 percent), suggesting warmer temperatures played role in explaining evolution of undocumented migration. Natural disaster events coefficient is positive (and significant at 5 percent) highlighting the role such events play in explaining undocumented migration. In the years of such events 14 percent of border apprehensions could be attributed to these events, on average.

^{Table 1:}

Determinants of evolution of border apprehensions of NT citizens (FY1994-FY2019)

Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01 All variables are in natural log form. The regression model was estimated using Feasible Generalized Least Squares method with serial autocorrelation in panels.

^{Table 1:}

Determinants of evolution of border apprehensions of NT citizens (FY1994-FY2019)

Constant	5.383 (8.369)
US: Hispanics real median wage	5.241*** (1.693)
US: Hispanics unemployment rate	-6.101** (2.388)
NT: Real average wage	-1.438*** (0.236)
NT: Unemployment rate	16.160*** (5.230)
US: Southwest border agents	-2.264*** (0.485)
US: Legal admissions of NT citizens to the U.S.	0.013 (0.230)
US: Deportations of NT citizens from the U.S.	0.554*** (0.092)
NT: Immigration wave 1, FY2012-FY2014 (D)	0.831*** (0.095)
NT: Immigration wave 2, FY2019 (D)	0.729*** (0.145)
NT: Coffee production	-0.417*** (0.104)
NT: Homicide rate per 100,000	0.277** (0.117)
NT: Temperature (deviation March-August, 2-year average)	0.197* (0.103)
NT: Natural disaster event (D)	0.144** (0.063)
Time trend	1.569*** (0.418)
Number of observations	78

Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01 All variables are in natural log form. The regression model was estimated using Feasible Generalized Least Squares method with serial autocorrelation in panels.

View Table

^{Table 1:}

Determinants of evolution of border apprehensions of NT citizens (FY1994-FY2019)

Constant	5.383 (8.369)
US: Hispanics real median wage	5.241*** (1.693)
US: Hispanics unemployment rate	-6.101** (2.388)
NT: Real average wage	-1.438*** (0.236)
NT: Unemployment rate	16.160*** (5.230)
US: Southwest border agents	-2.264*** (0.485)
US: Legal admissions of NT citizens to the U.S.	0.013 (0.230)
US: Deportations of NT citizens from the U.S.	0.554*** (0.092)
NT: Immigration wave 1, FY2012-FY2014 (D)	0.831*** (0.095)
NT: Immigration wave 2, FY2019 (D)	0.729*** (0.145)
NT: Coffee production	-0.417*** (0.104)
NT: Homicide rate per 100,000	0.277** (0.117)
NT: Temperature (deviation March-August, 2-year average)	0.197* (0.103)
NT: Natural disaster event (D)	0.144** (0.063)
Time trend	1.569*** (0.418)
Number of observations	78

Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01 All variables are in natural log form. The regression model was estimated using Feasible Generalized Least Squares method with serial autocorrelation in panels.

Our baseline specification has a good fit and explains well the evolution of border apprehensions, and the results are robust. Fitted values of the baseline specification complies well with the actual data of border apprehensions, suggesting the model performs well in a goodness of fit test (see left chart). Annex IV presents country-specific results. Annex V presents unit root and cointegration tests. The results are robust to various specifications, estimation methods, different measures of variables of interest (of border apprehensions, income, probability of apprehension, decline in livelihoods due to coffee production and changes in temperature), as well as different sample sizes or number of explanatory variables used. Annex VI summarizes the most important sets of additional results (replacing real wages with real GDP per capita, various measure of undocumented migration, use of stock of immigrant population as network effect, instrumental variables, and residuals enforcements estimations). A much larger set of results is available by request from the authors.

To better understand the results, and what role they play overtime, we illustrate the contribution of various group of factors, to the increase in border apprehensions in FY2019 compared to FY2011. We used the model estimates to examine the recent evolution in border apprehensions, in the period FY2012-FY2019. The right chart shows the contribution of each group of factors as a change in the corresponding variable relative to FY2011, times its elasticity. In the chart we aggregated results for overall the NT region and across identified the groups of variables.

Fitness of Estimated Model

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: Staff estimates based on baseline model in Table 1. See data section for sources.1/ Chart shows contribution to detrended series of border apprehensions for an average NT country.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Fitness of Estimated Model

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: Staff estimates based on baseline model in Table 1. See data section for sources.1/ Chart shows contribution to detrended series of border apprehensions for an average NT country.

• Download Figure
• Download figure as PowerPoint slide

Fitness of Estimated Model

Citation: IMF Working Papers 2023, 017; 10.5089/9798400232039.001.A001

Source: Staff estimates based on baseline model in Table 1. See data section for sources.1/ Chart shows contribution to detrended series of border apprehensions for an average NT country.

• Download Figure
• Download figure as PowerPoint slide

The results suggest that the potential stream of income was one of the main drivers of border apprehensions in the past decade, and played an increasing role. The chart shows the increasing contribution of the U.S. labor market conditions related factors (blue bar). Indeed, data shows that since FY2011, the U.S. Hispanic real median wages continuously increased, and the U.S. Hispanic unemployment rate continuously declined.¹⁷ Together these two factors of potential stream of income explained 100 percentage point increase in border apprehensions in FY2019 relative to FY2011 (out of the 230 percent increase).

The perceptions of changing migration policy mattered in driving temporary surges in undocumented migration from the NT countries to the U.S. For example, the first immigration wave in FY2012-FY2014 accounted for an additional 83 percent of apprehensions, compared to FY2011, by far the largest contributing factor of the surge in migration at the time. The second wave accounted for an additional 73 percent of apprehensions in FY2019.

Changes in the probability of being apprehended also explained the recent trends in undocumented migration. The change in the indirect measures of probability being apprehended contributed to 46 percentage points increase in border apprehensions in FY2019 relative to FY2011.¹⁸ Annex VII discusses in more detail border enforcement measures and their role.

The role of NT-specific factors and the forsaken stream of income factors in border apprehensions of NT migrants varied among time and across countries, but overall played a lower role. The chart shows that during FY2014-FY2017, the NT-specific factors (red bar) played on aggregate a limited role in explaining the increased apprehensions. Looking in further detail, at the country level (see Annex IV) we observe that indeed for El Salvador, the high and rising homicide rate at the time and the decline in coffee production played the most important role in explaining apprehensions. In Guatemala and Honduras, the NT-specific factors were more favorable—as coffee production increased and homicides declined—and contributed negatively to the border apprehensions. The forsaken stream of income indicators (yellow bar) contributed positively to apprehensions in 2016–2019, with large heterogeneity. The positive developments in El Salvador during this period—increase in real wages—contributing negatively to border apprehensions, while the opposite is true for the other countries, especially Guatemala and Honduras.

Climate change and natural disaster events explained significant share of border apprehensions of the NT undocumented migrants. The average May-August temperature deviation from the historical values, accelerated since 2015, and averaged around 30 percent, annually. Our model results suggest that warmer temperatures resulted in additional 6 percent of border apprehensions since FY2015. Natural disaster events occurred in Honduras in 2016 and 2018, and in Guatemala in 2019, which our model suggest contributed to additional 14 percent of border apprehensions.

V. Concluding Remarks, Outlook, and Policy Implications

This paper disentangles the role of various drivers of undocumented migration from the NT countries to the U.S., which has increased steadily over time. To gauge undocumented migration—an unobserved variable—we use data on the U.S. border apprehensions, similar to Hanson and Spilimbergo (1999). To assess the role of various factors we employ explanatory variables consistent with an investment decision theory model, with migrants assessing if benefits from undocumented migration are higher than its costs. As such, along with the traditional economic factors, we also control for the role of the probability of being apprehended, sudden spells of undocumented migration, and non-traditional factors of migration. We find that undocumented migration of people of NT country of origin to the U.S. is well explained by traditional factors—such as indicators of income (real wages or GDP per capita, especially in the U.S.). The results are consistent with anecdotal evidence, and MPI survey data that suggest that the economic considerations (real wages and unemployment rates) play the largest role in driving undocumented migration. Most importantly, our results suggest that border enforcement measures act as a deterrent, but large and sudden increases in border apprehensions are at times linked to favorable changes in the perception of probability of being apprehended. We also find that nontraditional factors (increase in temperature, decline in coffee production, increase in homicides rates and natural disasters) also play an important role in driving up undocumented migration, sometimes the most important role in explaining an increase in a given country, and year. The results are robust to various specifications, estimation methods, and different measures and variables used.

Migration factors are expected to persist, given rising challenges from climate change, but mostly depending on the relative economic developments in the U.S. economy versus the NT countries. Climate change experts expect temperatures to continue rising, which would continue to adversely affect the traditional livelihoods in the NT countries. The medium-term outlook for the U.S. economy remains favorable, supporting positive developments in the labor market conditions. The medium-term outlook for the NT countries remains subject to downside risks, as tighter global financial conditions will curb private sector investments. Addressing structural challenges at home to ensure competitive wages and equitable and inclusive growth, as well as improving security conditions, remain key priorities. These factors—uncertainty in the forsaken stream of income and NT-specific factors—would continue motivating undocumented migration. For example, a 10 percent increase in home country real wages would decrease undocumented migration by 14 percentage points. While a one standard deviation increase in temperature equivalent of 1.7 degrees, would increase undocumented migration by 10 percentage points.

Against this backdrop, the immigration wave events are expected to be frequent, especially when there is a change in border enforcement or immigration policy. Deportations are expected to continue, and border enforcement measures continue to tighten, which coupled with speculations of further tightening in the migration process and changes in migration policy, may fuel episodic immigration waves.

Policy response should aim at lifting up economic conditions in NT countries, help traditional producers adapt to climate change, and enhance security measures. Improving economic conditions at home that ensure higher real wages and low unemployment and creating more and better paid jobs would disincentivize undocumented migration from the region. Diversifying income opportunities for the coffee producers—especially in El Salvador—, while investing in climate-change resilient agricultural production would benefit migration-prone population and discourage attempts at undocumented migration. Preserving and further enhancing security measure to tackle crime in the NT countries would also be instrumental in stoving off attempts for undocumented migration. Furthermore, broadening ways of legal migration for NT migration, including in seasonal and temporary jobs, may also contribute at regulating flows, especially if conditions at home improve. Therefore, continued dialogue and efforts to help address the root causes of migration in NT countries should continue with the U.S., as already done with the USAID and other development partners.