Mexican and Latin American Migrants in the United States

For many decades, researchers have predicted a growing volume of migration in the world (^{Lee, 1966}). At present, these projections continue to be confirmed. Recently, the International Organization for Migration (IOM) noted that there was an 81 percent increase in the total number of international migrants in the world. The number of international migrants in 2020 was 272 million, a figure that represents 3.5 percent of the world population (^{OIM, 2019}).

In the case of the United States, the immigrant population has been increasing (^{Gaspar Olvera, 2012}). Their number has quadrupled since the Immigration and Naturalization Act (Public Law 89- 236 of 1965) came into force. Foreign born residing in the U.S. in 2018 reached 44.8 million, that is, 13.7 percent of its population (^{OIM, 2019}). In the particular case of those born in Mexico, they represent 25%,³ while those born in other Latin American nations represent another 25% (^{Budiman et al., 2020}).

Migration from Latin America, and particularly from Mexico to the United States has been a permanent phenomenon where, according to the United Nations (^{ONU, 2009}), “the multidimensional complexity of the migratory phenomenon can only be understood if it is considering Mexico as a country of origin, transit and destination” (p. 22). The ^{Inter-American Commission on Human Rights [Comisión Interamericana de Derechos Humanos] (2013)} characterizes this phenomenon as complex population movements, which include different profiles of people in search of a dignified life, work, asylum, or refuge, who “in particular come from vulnerable groups” (p. 28).

What Databases can Capture from the Movement of Migrants Passing Through Mexico

Transit to the U.S. is a process that has become intrinsic to the context of life of many families seeking the strength of the U.S. economy (^{Keeley, 2012}; Doctors Without Borders, 2019b). In 2010, the official figures from civil organizations estimate the transit of people passing through Mexico to the US. between 400 000 and 500 000 per year (^{ONU, 2009}; ^{Ponces et al., 2010}). For 2019, the Ministry of the Interior confessed: “That is an amount that I want to tell you that I do not know from where or how 144 thousand people passed through our country” (^{Delgado, 2019}, p. 1).

In January 2019, the report of the National Institute of Migration (INM) stated that the Mexican territory received the first Central American foreigner to request asylum in the United States for humanitarian reasons for the year, and the report from July of that same year indicated that the applications reached 18 503 (^{INM, 2019}). There are thousands of people seeking to enter the

U.S. but not all succeed. For example, the return figures for the first half of 2018 were: 4 798 in Ciudad Juárez, Chihuahua; 4 435 in Tijuana; and 2 689 in Mexicali, Baja California (^{Amnistía Internacional, 2018d}).

It seems that transit through Mexico is no longer temporary. The migratory population begins to benefit from the offers made by the government and the private initiative: school workshops, shelters, and health. “It has occurred both in formal jobs and in the informal sector, depending on the date of their arrival in the entity” (^{Ramírez Meda & Moreno Gutiérrez, 2022}, p. 3), however, according to international experience, these welfare policies may not be permanent (^{Castles, 2004}).

The Time Needed for Migrants to Traverse Mexico

The National Human Rights Commission (CNDH, initials in Spanish for Comisión Nacional de los Derechos Humanos) estimated that the time it takes to travel through Mexico, from the point of entry to the point where it is planned to cross into the United States is between one and four weeks on average (CNDH, 2018). The testimonials indicate that the travel time is variable depending on the route changes that are made along the way. ^{Nájar (2018}) identified that migrants can opt for longer journeys to be safer: “how ever long it takes, as long as it takes us to arrive but alive, the time does not matter” (p. 1). Twenty-three days was the time of the journey documented by ^{Tucker and Tillotson (2018}) from Honduras to the capital of Mexico.

Some Theories on Migration Issues

The books by ^{De Haas et al. (2020}) and ^{Massey et al. (1998}) synthesize definitions and migration theories. Nonetheless, there is no universally accepted definition for the concepts of migration or migrant, since there are several elaborated for different contexts. In the recommendations on international migration of the United Nations Department of Economic and Social Affairs (DESA), an international migrant is defined as “any person who changes their country of habitual residence,” distinguished from short-term migrants who change countries for periods of up to one year (^{OIM, 2019, p. 27}).

In relation to the issue of the paths taken by migrants, the contribution of ^{Ravenstein (1885}) when proposing that the routes are fixed and well defined is relevant. The author states that “the bulk of migrants come from short distances” and that “the number of migrants who arrive at a center decreases as the distance increases” (p. 198), depending on the existence of four factors: a) those associated with the area of origin, b) the destination area, c) intermediate obstacles, and d) personal factors such as perception, intelligence, contacts, or information. In particular, the third point is useful for this research, since intermediate obstacles, as well as violence or authority filters, seem to be the causes of the change of plans for the original route that passes through Mexico (^{Casillas, 2008}; ^{Foote & Small, 2013}; ^{MSF, 2019a}).

Danger on the Routes

Until less than a decade ago, Central American migrants could transit through Mexico in relative safety, but as of 2007, organized crime found a stable source of income in the kidnapping of migrants (^{Izcara Palacios, 2015}). Migrants are attacked by mafias and gang groups during their transit through Mexico (^{González Arias & Aikin Araluce, 2015}); in addition, they are forced to participate in illegal activities by joining drug cartels, that is, they are transformed from day laborers to criminals (^{Izcara Palacios, 2016}, ²⁰¹⁷). Migrants lack resources and are not protected by formal organizations (^{Sampson, 2008}), so it seems that violence and migration coexist permanently, both always looking for new routes (^{Ponces et al., 2010}).

A second risk is experienced by migrants in migrant holding centers, due to the possibility of being deported to their countries of origin, which is accentuated by the mistreatment exerted by some agents. In this regard, the INM has a series of complaints filed about cases of abuse (^{Amnistía Internacional, 2017}, ^{Comisión Nacional de Derechos Humanos, 2017}, ²⁰¹⁸).⁴

The Caravans

A recent massive protection strategy implemented by Central American migrants is the caravans (^{Gandini et al., Gutiérrez, 2020}). ^{Amnesty International (2018a}, ^2018c) estimates that the size of each of these caravans can reach up to 10 000 people. It is not clear if these movements are identified as circumstantial to the current migratory dynamics of crossings through Mexico (^{García Aguilar, 2020}) or if it is going to be a permanent phenomenon on a smaller scale. However, these caravans use the migratory routes that pass-through Mexico.

Graph Theory for the Study of Migratory Flows

In a not very rigorous way, we can define graphs as schematic representations of a network of connections, whose points (nodes) are related to each other by paths (edges). These representations make it easier to visualize and interpret these problems and the study of their properties (^{Ortega Reyes, 2020}). A first example of the application of Graph Theory is the solution of the Euler problem of the Königsberg bridges of 1736 (^{Núñez et al., 2004}), represented in Figure 1, where the map of the site is compared with its equivalent graph.

Source: Apple and Haken (2009).

Figure 1 Map and Graph of the Bridges of Königsberg 

Migration analysis has used network tools for a long time. An example is the study of the Uppsala-Näss parish in Sweden, which narrates how that at the end of the 19th century the village “resembled a train station where people come and go” (^{Hanagan, 2008}, p. 59). However, ^{Izquierdo Escribano and Noya (1999}) propose that for the study of migration it is necessary to improve the tools to operationalize their concepts, that is, “the assumptions of the analysis techniques in empirical research must be taken into account: multilevel analysis and mathematical analysis of graphs and networks” (p. 38).

Recent studies already use Graph Theory more frequently to study migration. ^{García Flores (2013}) analyzes the patterns of internal migration in Colombia. ^{Ramirez (2014)}, with a model applicable to different scales, finds that the data from a region of Chile is close to a small world network⁵ with 36 nodes, 977 links, an average clustering coefficient of 0.813, and an average path length of 1 225. ^{Lomelí Carrillo and Ybáñez Zepeda (2017}) show the differences between two regions on the northwest border of Mexico, while ^{Ruiz-Santacruz (2019}) affirms that when analyzing social networks, they “provide a tool to study relationships between countries, the importance of each one within the network, and the influence of country variables on migration patterns” (p. 79).

In summary, the Graph Theory is useful to know if leaving one city you can reach another, to know the similarities between the regions and to determine the types of relationships that exist between each one of them. The contribution of this theory to migratory flows is basically that it allows us to know which cities concentrate more power in terms of traffic and which are more isolated within the network.

Problems with the Registration of Figures on the Movement of Migrants Through Mexico

One of the problems for migration studies, especially in Latin American countries, is their limited ability to collect data and provide reliable figures on the movement of people, this is basically due to geography and complicated border management. For example, in some regions it represents a challenge due to the non-existence of maritime borders, the distance from mountain passes or the control of isolated paths within tropical forests (^{Gallagher & McAuliffe, 2016}). Also, registering the precise number of migrants presents another series of problems due to the fact that not all movements of people are counted, such as entry or exit from Mexico with a tourist, work or educational visa (^{OIM, 2018a}).

Another phenomenon that makes the volume of the movement difficult to calculate are the caravans, since the number of people varies constantly. There are those who return to their homes, others join the journey, and others are divided to form new groups (^{Amnistía Internacional, 2018a}, ^2018c). Therefore, it is a challenge to build a specific database and has the caveats of the method used.

In particular, due to the fact that the statistics of the movements are irregular, the construction of the database of the present study is artificial since it is built based on different sources.

Sources of Information

The first methodological phase was based on an exploration whose objective was the construction of a database of the cities of Mexico through which migrants’ transit. In this phase, the recommendation of the organization ^{Arquitectos con la Gente (2016)} [Architects with People] was followed, which proposes to trace the transit of migration based on the location of aid zones and places where people can obtain better living conditions, food, and accommodation, among other services. Along these lines, directories of shelters or support centers for migrants were obtained from government organizations, the CNDH or different civil organizations, such as Doctors Without Borders. As a complement, an exhaustive review of academic articles published in indexed journals on the subject in Mexico was made, following the suggestion of ^{Hernández Sampieri et al. (2006}).

To build the database, an Internet search was carried out based on the constructions: “migrant routes,” “migrants’ shelters,” “migrant housing,” and “migrants soup kitchens/food services.” The list of public and private organizations identified that handle this information between 2010 and 2018 is presented in Table 1.

The points located in the works of ^{Casillas (2008}) and ^{González Arias and Aikin Araluce (2015}) were also included in the database. Additionally, the points identified in 26 complementary journalistic sources that responded to the search criteria indicating some of the routes taken by the migrants were integrated (Table 2).

The second phase consisted of a concentration of data under the following process: 1) each city was assigned a node number; 2) cities were classified into three types of nodes: origin (located in Central America), transit (Mexico cities) and destination (located in the United States); 3) nodes were grouped by state for the purposes of their geographical location.

Construction of the Road Map

The third phase of the methodology consisted in the construction of the node interconnection map. For the construction of roads, the following criteria were followed: 1) if the connection between cities was declared in any of the sources consulted, the indicated path was respected; 2) if the source mentioned a city without indicating its entry or exit routes, it was connected to the closest neighboring nodes (in linear meters, following the recommendation of ^{Kuz et al., 2016}); 3) For this article, connections related to air transportation flows were excluded since they present a non- significant volume (^{Casillas, 2008}).

It was decided on a mixed network model, directed at the entry, and exit nodes of the country, and not directed at the internal nodes, since there are cases in which migrants return to previous points to change their path (^{Amnistía Internacional, 2018b}).

Points of Danger for Migrants

To identify the nodes that represent danger for migrants, two indicators were adopted:

The state homicide rate, calculated from the records of the National Institute of Statistics and Geography (Inegi by its initials in Spanish), in relation to the state population estimated by the National Population Council (Conapo by its initials in Spanish). This indicator was recommended as a reliable measurement of violence because it presents the lowest black figure (^{Álvarez & Manzotti, 2008}). Those states that had a homicide rate above the average plus one standard deviation were the selection criteria (^{Méndez, 2017}).

The location of any of the INM migration stations, not only because of the risk of being deported, but also because of the violence detected (^{Amnistía Internacional, 2017}).

Hypothesis Testing Procedure

For the analysis and testing of the hypotheses, the following definitions were used (^{Ortega Reyes, 2020}):

Node: city or border. For example, Guadalajara and Comitán are city nodes, while the United States and Central America would be border nodes.

Edge: is the transit connection between two nodes. For example, the train route that connects Los Mochis in Sinaloa with Cuauhtemoc City in Chihuahua; the road that connects Santa Rosalía with Guerrero Negro in Baja California Sur; the sea route that connects Cancún with Progreso in the Yucatán Peninsula, or the river that connects Peñitas with Belisario Domínguez in Chiapas.

Path: is the set of edges that are required to connect any two nodes (cities); For example, a path of order 6, which connects Central America with the United States, would include: 1. Central America, 2. Tenosique, 3. Guadalajara, 4. Chihuahua, 5. Ojinaga, 6. U.S.

Graph: is the set of identified nodes and paths. The study graph includes the 183 nodes and the 720 edges of possibilities located.

Elementary path: it is that path that never passes more than once through the same node. For example, route 1. Central America, 2. Tenosique, 3. Guadalajara, 4. Chihuahua, 5. Guadalajara, 6. Hermosillo, 7. Nogales, 8. Naco, 9. United States could not be an elementary path because Guadalajara is repeated.

Shortest path: it is the one that joins two nodes with the minimum number of possible edges. For example, among the options: Celaya-Aguascalientes (by road), and Celaya-Salamanca- Irapuato-León-Aguascalientes (by train), the first path is the shortest since the second has three intermediate nodes.

To test H1 (it can connect the northern border with the southern border by more than one path), an iterative process of reduction of the total number of shortest elementary paths between both borders was used: a) the entry nodes to the United States were identified; b) all the nodes that connect by an edge to the previous nodes were identified; c) the nodes that connect to the previous nodes were iteratively identified, through an inverse tree process, until the exit nodes of Central America were located, and d) the elementary paths that were not the shortest were discarded.

To test H2 (there are cities in Mexico that can be considered as controls) the comparative analysis of nodes of the Graph Theory was used, based on two variables that measure structural centrality (^{Kuz et al., 2015}): a) the degree of centrality, and b) the degree of intermediation (betweenness centrality). Both metrics are defined in Table 3. For the comparative analysis, the Gephi system version 0.9.2 (2017) was used for the analysis of social networks.

The H2 is tested if more than one city with a statistically significant DB (p-val < 0.001) is identified.

To test H3 (the shortest paths between the two borders have at least one node with a dangerous situation), the database of the shortest elementary paths identified between both borders⁶ was cross checked against the data base of points in danger.⁷

The Diagram of Cities and Paths

The diagram of cities and paths (nodes and edges) built with the Gephi software, from the Force Atlas distribution (Figure 4), was created from an algorithm that arranges the nodes on a plane based on two iterative rules: the nodes attract each other based on their geometric distance¹¹ and repel each other based on the number of connected neighbors¹² (^{Jacomy et al., 2014}). As a result, the software calculates an optimal partition of clusters, which concluded in 8 regions, which are presented in the diagram of Figure 2.

Source: Own elaboration with the support of Gephi software, version 0.9.2 (Force Atlas distribution).

Figure 2 Diagram of Cities and Roads of Migrants¹³  

The eight resulting regions (clustering process) are identified with the name of the city with the highest degree of centrality: 1) Mazatlán Region; 2) Northern border region; 3) Saltillo Region; 4) Guadalajara Region; 5) Veracruz Region; 6) Tenosique Region; 7) Southern border region; and 8) Chicoasén Region.

Structural Holes

Six articulation points are identified in the graph, that is, nodes that when removed divide the map into two disconnected maps. Such are the cases of the cities of Campeche¹⁴ and Progreso in the Yucatan peninsula, as well as Reynosa, Chihuahua, El Paso, and Altar in the north. If these nodes are removed, an isolated set of cities from the general map tree type treer¹⁵ occurs. These types of nodes are also known as structural holes, since, if a check point is established, it can break the transit of a sector of the network (^{Borgatti, 2005}; ^{Jimeng & Jie, 2011}).

Control Cities Due to Their Degree of Intermediation

By calculating the degree of intermediation-the number of times the city in question appears within all the shortest paths-of each of the cities in the migrant network in Mexico,¹⁶ its degree of influence on transit to the interior of the graph is identified. On the other hand, if BD is divided by the total possibilities of edges, the probability of control of the transit of the graph is calculated. Table 6 shows the most influential cities.

In other words, Guadalajara, Tenosique, Ixtepec, and Mazatlán are the cities through which the greatest number of minimal interconnections can pass within the network.¹⁷

With the location of control cities, whose probability of traffic control > 0 (p-val < 0.001) H2 is approved, therefore, there are cities in Mexico that can be considered control due to their degree of intermediation.

However, the average degree of intermediation of the graph is DB = 709.91 (DSM¹⁸ =120.92), which rejects the idea that the network of migrants in Mexico is a perfectly connected network (p-val

< 0.001), which would happen if DB were equal to zero (as it happens in a complete graph).

Network of “Negative” Factors that Affect the Migrant’s Decision

To test the hypothesis, a database of cities with a dangerous situation was built.

The homicide rate was calculated for each of the states in Mexico with information from the ^{Inegi (2019)} and ^{Conapo (2019)} on the number of homicides committed per state and the number of people per state during 2017, 2018 and 2019. The national average was 30.9 homicides per 100 000 inhabitants (with a standard deviation of 23.7). Table 7 presents the homicide rate in descending order.

The location of the migrant holding centers and the Beta Groups were obtained from the INM directory and are detailed in Table 8.

If it is considered that the dangerous cities are those that belong to the states whose homicide rate is higher than the national average, plus one standard deviation (>54.6), such as Colima, Baja California, Chihuahua, Guerrero, and those that have some INM office, it can be verified that all the shortest paths identified in Table 5 pass through at least two such cities with a dangerous situation. This confirms the H3 hypothesis since the shortest routes between both borders include at least one node with a dangerous situation. For example, a route was identified that passes through four danger points: Central America, Tenosique, Guadalajara, Chihuahua, Ojinaga, United States.

Pragmatic Distribution

A finding was presented regarding the possibility of subdividing the map of cities and routes of migrants into a hypergraph of order 3 from a construction of three mutually exclusive and collectively exhaustive blocks. The Venn diagram is presented in Figure 3.

Source: Own elaboration.

Figure 3 Diagram of the Three Identified Blocks 

A particularity of this scheme of partition into three blocks or clusters is that a series of tangent borders can be distinguished between them. However, within each block there is an extensive interconnection network. The northern block has three regions, the central south two and the southeast three regions. There are 14 edges between the north block and the south center, 14 between the south center and the southeast, and only two between the north and the southeast.