Does International Migration Affect Political Participation? Evidence from Multiple Data Sources across Mexican Municipalities,1990

Abstract

Does international migration affect political participation in origin countries? We study how different channels of international migration — migrant absence, return, circularity, and transnational engagement — affected political participation across Mexican municipalities between 1990 and 2013. Using two different data sources including panel and longitudinal survey data, we find that generally, international migration has a positive effect on formal political participation in local elections. Results, however, are mixed across specific migratory channels, and the effects are conditional on levels of civic engagement. The study shows that specifying different channels of international migration is key to understanding the conditions under which emigration enhances political and civic participation in the local democratic process in origin countries.

Introduction

For the past four decades, when investigating the consequences of emigration in origin countries, scholars have focused extensively on how emigration affects economic development (Durand, Parrado, and Massey 1996; Adams and Page 2005; Fajnzylber and López 2007). With migrant remittances reaching $601 billion worldwide in 2015 and tripling over the past 15 years (Ratha, Eigen-Zucchi, and Plaza 2016), it is no surprise that international migration’s economic consequences have been front and center. More recently, though, a growing body of research has begun to examine international migration’s political consequences, revealing important implications for democracy. Some research in this vein maintains that emigration depresses democratic functioning (Bravo 2008; Goodman and Hiskey 2008; Germano 2013). Other research is more sanguine and argues that emigration can be a source of democratic diffusion (Levitt 2001; Pérez-Armendáriz and Crow 2010; Pfutze 2012; Chauvet and Mercier 2014; Córdova and Hiskey 2015; Burgess 2016). In this article, we move past a debate about whether migration is “good” or “bad” for democracy and instead analyze the conditions under which different channels of international migration affect a key dimension of democratic functioning — nonmigrant citizens’ local civic and political participation.

In doing so, we unpack the multiple channels through which international migration shapes the political participation of nonmigrants remaining behind, including how migrant absence, circular migration, return migration, and migrant transnational engagement change citizens’ participation in local elections and community associational activities. We conceptualize migrant transnational engagement as the regular cross-border practices in which migrants engage to express social ties, loyalties, and connections to the people and places they leave behind (Schiller, Basch, and Szanton Blanc 1995; Guarnizo, Portes, and Haller 2003). While there are many forms of transnational engagement, we study the extent to which two kinds, household and collective remittances, produce observable changes in nonmigrants’ political behavior.

Household (family) and collective remittances represent individual and collective transnational practices that differ in terms of scope, use, and entity sending the resources. Household remittances are savings accrued by individual migrants that are sent back to households in the origin country to diversify income for private consumption (Durand, Parrado, and Massey 1996; Kapur 2010; Ratha, Eigen-Zucchi, and Plaza 2016). By contrast, collective remittances are pooled resources sent by migrants through voluntary civic associations based on members’ shared hometown ties and called hometown associations (HTAs). HTAs invest in public goods and services in their places of origin, creating opportunities that can be enjoyed by the entire community, including migrant and nonmigrant residents alike (Levitt 2001; Goldring 2002; Smith 2006; Portes, Escobar, and Radford 2007; Iskander 2010; Bada 2014; Duquette-Rury 2014, 2016; Burgess 2016). Compared to household remittances, the volume of collective remittances is much smaller, although systematic data are scarce (Ratha, Eigen-Zucchi, and Plaza 2016). Despite important research that highlights the ways in which migrants become political change agents by mobilizing both individual and collective resources toward political ends at home (Levitt 2001; Smith 2006; Iskander 2010; Meseguer and Aparicio 2012; Bada 2014; Duquette-Rury 2014, 2016; Burgess 2016), researchers have yet to empirically examine the diverse political consequences of different remittance flows in a single study. Our study takes up this task.

Additionally, a central finding in previous studies is the association between international migration and civic engagement in migrant hometown communities (Bravo 2008; Goodman and Hiskey 2008; Pérez-Armendáriz and Crow 2010; Córdova and Hiskey 2015), but these same studies find either a negative or no relationship between migration and electoral participation in migrant hometowns. We improve on existing research by examining how civic engagement is affected by migratory processes and whether civic engagement has a conditional, intervening effect on electoral participation. Thus, we not only examine the link between absence, return, circularity, and individual and collective forms of transnational engagement but also evaluate the extent to which civic engagement spurred in part by international migration plays a mediating role in citizens’ formal participation in local elections.

While previous work has focused primarily on political participation in national elections (Bravo 2008; Goodman and Hiskey 2008; Pérez-Armendáriz and Crow 2010), our study examines citizen participation in local (municipal) elections.¹ We analyze local political engagement because decentralization reforms that devolve administrative, political, and fiscal authority to local government for public goods provision, protection of rights, and social order have been implemented in many countries since the 1990s (Falleti 2005). In decentralized countries, citizens may be more likely to make political demands on the state since they are in closer proximity to elected officials whose decisions directly affect their daily lives (Bardhan 2002; Treisman 2002). As such, it is more important than ever to systematically assess how transnational forces such as international migration affect citizens’ ability to make demands on the local state apparatus through formal and informal political participation in migrant origin countries.

We use Mexico as a strategic case for our analyses because the country has experienced substantial emigration while simultaneously undergoing important changes in political institutions likely to affect political participation. The Mexico-United States migration corridor remains the largest in the world, with over 13 million migrants regularly crossing the border and sending more than $25 billion remittances home in 2015 (Ratha, Eigen-Zucchi, and Plaza 2016). Moreover, while Mexico experienced a national democratic transition in 2000, authoritarian enclaves have persisted across state and local governments, creating vast variation in citizens’ interest and ability to participate in local political institutions (Giraudy 2010). Taken together, long-standing and varied emigration across the country, decentralization reforms, and uneven democratic transition make municipalities in Mexico ideal settings in which to study emigration’s effects on local political participation.

To examine the political consequences of international migration on local political and civic participation in Mexico, we rely on two different panel data sources and separate statistical analyses. First, we compiled panel data of all Mexican municipalities from 1990 to 2013 from multiple public data sources where disaggregated data on international migration were available for all municipal years.² Second, since Mexican government statistics do not include municipal data on civic engagement, we turn to a second data source, the Mexican Family Life Survey (MxFLS), and conduct additional statistical analyses. The MxFLS is a longitudinal, nationally representative survey sample of Mexican localities conducted by researchers from the Iberoamerican Univeristy (UIA) and Center for Economic Research and Teaching (CIDE) in collaboration with researchers from Duke University. The MxFLS includes a battery of questions about civic associational life such as the types, incidence, and frequency of community-related activities between 2000 and 2013. Capturing change over time, both the panel data set and the MxFLS data allow us to take temporal dynamics seriously as migration patterns, political institutions, and political participation rates change over time and space. To temper concerns about selection bias often at issue when using observational data, we use both the difference-in-difference (DiD) estimator and fixed effects models with a municipal linear time trend to draw inferences with greater confidence (Watson 2006).

Our findings suggest that as migrant absence, circular migration, and the collective involvement of migrant HTAs in public goods provision increase, there is a corresponding increase in nonmigrant citizens’ participation in local elections. By contrast, we find that return migration has a negative effect on political participation, contradicting previous research on Mexico and Mali (Chauvet and Mercier 2014; Waddell and Fontenla 2015; García 2017). Moreover, we find no evidence that family remittances produce any observable changes in political participation. Finally, results show that civic engagement has a positive, conditional effect on migrant HTA collective involvement in their places of origin but depresses political participation in places characterized by high rates of emigration. Civic engagement is unrelated to family remittances and circular and return migration. Our results provide compelling evidence that international migration’s effects are at work simultaneously and that it is imperative to empirically disentangle the competing channels (Kapur 2010) through which emigration changes civic and political participation in local democracy.

To develop these ideas, the article proceeds as follows. In the next section, we review the existing literature on international migration’s political consequences in origin countries. From there, we hypothesize how migrant absence, return, circular migration, and individual and collective forms of migrant transnational engagement are likely to affect political participation. The subsequent section describes our research design and data sources. Following a discussion of the empirical approach, we present the results for each set of statistical analyses before describing the substantive findings in the discussion section. The conclusion summarizes the main findings and implications for future research.

Political Consequences of International Migration in Origin Countries

The burgeoning interdisciplinary literature examining how international migration affects politics in home countries reveals important consequences for democratic functioning (Itzigsohn 2000; Levitt 2001; Goldring 2002; Smith 2006; Goodman and Hiskey 2008; Piper 2009; Rother 2009; Pérez-Armendáriz and Crow 2010; Careja and Emmenegger 2012; Pérez-Armendáriz 2014; Pfutze 2012). However, this research also often examines aggregate levels or a single dimension of international migration in a given study, thus obscuring the multiple and potentially conflicting mechanisms related to different migration channels that drive change over time (Bravo 2008; Goodman and Hiskey 2008; Chauvet and Mercier 2014; Waddell and Fontenla 2015; García 2017). The reliance on aggregate, cross-sectional indicators of international migration has been due to data availability, but such approaches cannot effectively adjudicate which aspects of migration affect political change.

On the heels of Goodman and Hiskey’s (2008) influential article, which showed that high emigration rates depress formal political participation but enhance civic engagement in Mexico, subsequent research began to disaggregate international migration to study how different channels of influence (migrant absence, return migration, or migrant remittances) affected different civic and political outcomes in Mexico and beyond (Germano 2013; Chauvet and Mercier 2014; Dionne, Inman, and Montinola 2014; Waddell and Fontenla 2015; García 2017). This work shows, for example, that the circulation of ideas, behaviors, and norms in migrant transnational social networks, called social remittances, can increase informal modes of civic and political engagement (Levitt 2001; Pérez-Armendáriz and Crow 2010; Córdova and Hiskey 2015). Research on return migration, however, comes to mixed conclusions. Chauvet and Mercier (2014) find that in Mali, places with high percentages of migrant returnees experience more national and local electoral participation, while Pérez-Armendáriz (2014) reports that in Mexico, return migration is related to more political apathy. Research on the role of migrant remittances similarly lacks consensus. Some studies find that remittances are a resource “curse” and weaken nonmigrants’ electoral participation in Mexico (Bravo 2008; Germano 2013), while others identify a positive link between remittances and informal political participation in an African context (Dionne, Inman, and Montinola 2014). While more recent studies push our understanding of how different channels of international migration affect political engagement and attitudes (Rother 2009; Germano 2013; Chauvet and Mercier 2014), examining only one channel in isolation leaves researchers unable to decipher the concurrent and perhaps opposing effects that each channel may have for political engagement.

To date, no single study has disentangled how multiple channels of international migration change political participation across space and time. Our goal here is to do just that — to examine how migrant return, absence, circularity, and individual and collective characteristics of transnational migrations affect political participation in local elections and civic engagement. In doing so, we help move the existing debate away from discussions about whether migration helps or hurts democracy to study more directly the conditions under which positive and negative effects on electoral and civic participation are more likely to occur.

International Migration Channels and Political Participation

Addressing international migration’s consequences for formal political participation provides a window into the procedures and results of democratic functioning (Diamond and Morlino 2005). Political participation here refers to the extent to which citizens exercise formal voting rights as well as organize, assemble, protest, lobby, join political parties and civil society associations, and otherwise influence the decision-making process. In theory, more formal participation and civic associationalism make democratic systems more responsive to a larger share of the population (Putnam 1994; Brady, Verba, and Schlozman 1995). Examining how different migration channels change local electoral participation and civic engagement thus offers insight into how emigration affects the functioning of the local democratic system.

We assess international migration’s political consequences for local (municipal) elections in the spirit of Kapur’s (2010) channels approach. Kapur highlights four channels through which migration affects the origin country: absence, prospective migration, return, and diaspora channels.³ The absence channel refers to the ways in which the departure of certain sending community members shapes political life at home. The return channel identifies how individual migrants returning home with new skills, capital, access to networks, and ideational resources affect politics. Finally, the diaspora channel refers to how emigrants affect political change from the destination country.

The absence channel is likely to produce effects on the electoral participation of citizens remaining behind. On the one hand, individuals who migrate may be more likely to engage in formal politics at home. Thus, the absence of more politically engaged citizenry, what Goodman and Hiskey (2008) refer to as “political brain drain,” may depress political participation in sending communities. On the other hand, the absence of migrant “elite” may create space for new groups to make political inroads by ramping up political participation and influence in electoral politics (Kapur 2010). This leads to our first testable hypothesis:

Hypothesis 1: As more migrants emigrate, citizens remaining behind are less (more) likely to turn out to vote in local elections.

Between 2005 and 2010, more than 1.5 million Mexican immigrants returned to Mexico after long stays in the United States (Passell, Cohn, and Gonzalez-Barrera 2012). Many returned voluntarily, while others were forcibly removed from the United States. After spending years in a richer democracy, Mexican return migrants are likely to play an important role in shaping the political behaviors of members of their social networks and origin communities by transmitting political ideas, values, and attitudes about democratic participation upon return (Rother 2009; Pérez-Armendáriz and Crow 2010; Batista and Vicente 2011; Chauvet and Mercier 2014; Pérez-Armendáriz 2014; Waddell and Fontenla 2015). Return migrants, exposed to political norms such as electoral participation and political accountability while abroad, may transfer these ideas through political conversations with social contacts in the hometown or lead by example by showing up to the polls and casting a ballot. Return migration may thus have a positive effect on nonmigrants’ political behavior by exposing them to norms about political engagement in a consolidated democracy such as the United States (Rother 2009; Chauvet and Mercier 2014). Moreover, negative experiences in the destination country may spur return migrants’ political interest and engagement in local elections in the home country. For example, undocumented immigrants who have political interests but are barred from electoral participation in the United States or those who are deported may harbor resentments toward the US political system, which they channel into Mexican political participation upon return by voting and encouraging other individuals to vote in local elections and attempt to change the political system.

However, migrant return may also have different effects, depending on whether it is more permanent or temporary (García 2017). If returnees are negatively perceived in the hometown, their own interest in and participation in politics may wane the longer they are at home. Additionally, other hometowners may be less receptive to returnees’ opinions, attitudes, and preferences because their return symbolizes failure to succeed abroad (Pérez-Armendáriz 2014). Migrants who participate in circular migration, returning temporarily to the hometown before emigrating once again, may be more influential in shaping participation in politics (García 2017). Their ability to regularly cross the border, live in the United States, and bring information home in situ may be received by hometowners with more authority and influence since the information is coming from more “successful” migrants. Finally, returnees’ ability to affect electoral participation may be enabled or constrained by the local institutional context, which is likely to vary substantially by levels of socioeconomic status and economic development. Migrants returning to poorer locales where patron-client ties are more entrenched may not participate or encourage their peers to participate in the political process if they do not believe that voting is likely to make a difference (Holzner 2010). The local political institutional context may then temper returnees’ interest and ability to transmit positive norms and attitudes concerning electoral participation. This leads to the next two testable hypotheses:

Hypothesis 2: As more migrants return from abroad to their hometowns, local voter turnout is likely to decrease (increase).

Hypothesis 3: As more migrants engage in circular migration, local voter turnout is likely to increase (decrease).

Finally, the transnational channel captures the ways in which migrants affect politics in the hometown from abroad. The two dominant ways in which scholars have operationalized the transnational channel include social and family remittances.⁴ When migrants leave yet remain connected to the people and places they leave behind, they transmit ideational and material resources, including ideas, norms, values, and experiences in the destination country (social remittances), through transnational networks (Levitt 1998, 2001), potentially influencing nonmigrants’ political interest and voting behavior (Pérez-Armendáriz 2014; Córdova and Hiskey 2015). Additionally, migrants send home money, which may increase political participation by increasing household resources — time, money, and civic skills — and therefore the communications and organizational capacity associated with political activity (Brady, Verba, and Schlozman 1995). By contrast, receiving remittances may also allow migrant households to subsist more independently of the state and encourage them to divest from political participation as they rely more on migrants abroad (Bravo 2008; Germano 2013).

Sending money home to households, however, is not the only kind of cross-border practice likely to increase communication and the organizational capacity of nonmigrants at home. We argue that collective remittances, which mobilize and organize migrant transnational collective action in hometown development, are another important channel likely to encourage political participation. Migrant collective engagement through HTAs affects political behavior through two possible mechanisms (Duquette-Rury 2016). If migrant HTAs supporting public goods projects recruit local citizens into collective remittance-funded projects, citizens may scale up this civic engagement and participate more in the formal electoral process. However, if HTAs and government partners exclude local citizens from co-provision activities in the hometown, citizens may respond by either politically mobilizing in reaction to exclusion or losing interest in and disengaging from local electoral politics altogether. This leads to our final hypotheses concerning individual and collective features of the transnational channel:

Hypothesis 4a: As more migrants emigrate, they transmit home ideas, norms, values, and opinions through transnational social networks that increase (decrease) local voter turnout.

Hypothesis 4b: As family remittances increase in Mexican municipalities, local voter turnout is likely to decrease (increase).

Hypothesis 4c: As participation in cross-border collective action in public goods provision with collective remittances increases, local voter turnout increases (decreases).

Our central objective is to observe how migrant absence, return, circularity, and forms of transnational engagement shape the formal political participation of local citizens in Mexican municipal elections before turning to their effects on civic engagement. In our analysis, the best proxy for social remittances at the municipal level is the number of migrants who live abroad (absence). However, migrant absence may also indicate the presence of political brain drain from the hometown (Goodman and Hiskey 2008). We are careful to acknowledge in the results and discussion section alternative ways of interpreting the findings on the absence channel given that it is often a close proxy for migrant contact with relatives abroad and the flow of social remittances across borders that may affect political participation (Pérez-Armendáriz 2014; Córdova and Hiskey 2015). In what follows, we analyze how the intensity of out-migration affects our main indicator of formal political participation between 1990 and 2013 before evaluating how other migratory channels produce political change.

Research Methods, Indicators, and Sources for the Mexican Migration Panel Data Set

In the full panel estimates for the political participation models, our main indicator of political participation is voter turnout, the percentage of the voting-age population casting a vote in municipal elections. The denominator is adjusted to reflect the percentage of the migrant population living abroad at the time of the municipal election. Voter turnout data were collected from the Municipal Elections Database compiled by CIDE, from the Centro de Investigación Para el Desarrollo, A.C. (CIDAC), a prominent Mexican think-tank, over the period 1990 to 2013 and from state-level public databases for missing years. Mexican municipal elections occur every three years in a staggered electoral calendar across Mexican states.

Our main indicator of international migration (absence) is the percentage of the total municipal population living abroad. Our disaggregated indicators of international migration are captured by the percentage of municipality households with a migrant returning (return), percentage of households with a migrant circulating between the Unitd States and Mexico (circular), and percentage of households receiving remittances from abroad (family remittances) in the previous five-year period. These data are from the Conteo de Población y Vivienda (Mexican national census). We use a linear interpolation technique to convert the five-year estimates into annual municipal-year observations for data consistency.⁵

To test the hypothesis that migrant HTAs’ collective participation in public goods provision affects voter turnout, we use data from the Mexican 3x1 Program. The 3x1 Program is a federal social spending program, started in 2002, that matches migrant HTAs’ collective remittances sent home for public goods provision. Each level of the Mexican government — local, state, and federal — contributes a quarter of the total cost of a public goods project. We rely on municipal participation in the 3x1 Program to account for places that formally engage in collective action for public goods provision with migrant HTAs.⁶ While all three levels of Mexican government co-finance total project budgets, the municipal-level government coordinates project selection, planning, implementation, and monitoring.⁷ For all indicators of municipal 3x1 participation, we use data from the Mexican Secretaría de Desarrollo Social (Sedesol) database of all approved 3x1 projects from 2002 to 2013.⁸ Measuring municipal participation is challenging since municipalities can start, stop, and reenter the program annually throughout the observation period. To account for these temporal dynamics, our indicators of 3x1 participation reflect whether a municipality participated continually, whether it participated in the previous three-year electoral cycle, and how many times a municipality participated from 2002 to 2013.

The first measure, 3x1 participation, is a dichotomous variable taking the value of “1” if a municipality participated in the program at least once in any three-year electoral cycle between 2002 and 2013 or “0,” otherwise. We also analyze whether a municipality ever participated in the 3x1 Program but do not report those results.⁹ Since municipalities can start and stop participation in the 3x1 Program at will, we assess the systematic difference between those municipalities that participated frequently and those that participated less regularly during the active program period. We code regular participation as any municipality that participated in the 3x1 Program once every three years. This step is necessary since previous research shows that municipal officials time their participation in the 3x1 Program with the three-year electoral cycle to maximize electoral benefits associated with credit claiming of public goods provision (Simpser et al. 2016).

The second and third measures indicate short-term and cumulative program participation. The short-term indicator, electoral cycle, accounts for the number of times a municipality participated in the previous electoral cycle.¹⁰ On average, municipalities participated about once each three-year cycle. Finally, we assess how “cumulative participation” in the 3x1 Program affects voter turnout. This indicator is a continuous variable measuring the frequency of program participation each year. Half of all 3x1 municipalities participated more than four times in the 12-year observation period. For the cumulative indicator, the relevant reference group (control group) combines all municipalities that participated less frequently to isolate the comparison between habitual and irregular participants.

Between 2002 and 2013, 1,234 municipalities participated in the 3x1 Program at least once, accounting for half of all Mexican municipalities (50.2%). Figure 1 plots total annual new municipalities participating in the 3x1 Program and total municipalities over time. In 2008, for example, while 539 different municipalities participated, only 87 municipalities entered the program for the first time. The number of new municipalities starting the program decreased over the program’s duration, indicating that many municipalities repeatedly engaged in the program rather than indicating diffuse policy adoption across municipalities.

Figure 1.

Number of Municipalities That Participate in 3x1.

Controlling for Confounders

The statistical analysis also includes additional covariates hypothesized to influence both the baseline change common to all units of observation and the amount of change predicted by our different indicators of international migration. We include several socioeconomic, demographic, and political characteristics likely to affect municipal voter turnout. Our first control variable, poverty, is a continuous index of economic marginalization, which includes data on the percentage of the municipal population living without primary education, drainage or toilet, electricity, or piped water; the percentage who are illiterate; and the percentage living with overcrowding, earthen floor, income below the poverty level, and in locations with fewer than 5,000 people. The index captures important levels of income, social infrastructure, and educational attainment in each municipality. A square term for poverty is included because of the curvilinear relationship with political participation. Participation is often lowest in the most affluent cities, slightly higher in the poorest cities, and highest in middle-income cities (Oliver 1999).

Second, we include a measure of the percentage of the population that speaks an indigenous language (“indigenous”) since previous research shows that ethno-linguistic fractionalization affects democratic quality (Alesina, Baqir, and Easterly 1999). Third, we include an additional measure for literacy, the percentage of the total population over the age of 12 that is literate (“literacy”) because studies show that literacy is positively associated with voter turnout (Brady, Verba, and Schlozman 1995; Hiskey 2003; Cleary 2010) and because we want to assess its independent impact apart from its role in municipal poverty. Finally, “total population” (logged) reflects the sociodemographic profile of municipalities in all specifications and is a standard control in voter turnout models (Powell 1986).

Additionally, we include a vector of political variables. The degree of electoral competition has been hypothesized to affect trends in political participation (Hiskey 2003; Cleary 2010). We include a measure of multiparty electoral competition, following Laakso and Taagepera’s (1979) formula that measures the number of parties effectively competing for office. This strategy helps account for the relative size of effective votes according to the vote share garnered for each competing party in the election. The effective number of parties (ENP) is calculated by the equation:

N= \frac{1}{\sum_{i=1}^{n} p_{i}^{2}}

where N is the number of parties with at least one vote and $p_{i}^{2}$ is the square of each party’s proportion of all votes or seats. If all vote shares are equal across parties, N is only slightly larger than 1. A hallmark feature of the ENP measure (“number of parties”) is that it can detect small shifts likely to be missed otherwise.¹¹ Also, since different political parties court voters using different strategies, we include the municipal incumbent’s party label lagged one electoral cycle in all specifications (Meseguer and Aparicio 2012).

Finally, to control for municipal government capacity and public spending that are likely to affect political interest and participation in local elections, we include a lagged measure of total revenues (per capita). Total revenues measure government budget constraint and include all sources of funding for municipal government (i.e., state, federal, and locally sourced revenue). Elections data were collected from publicly available data sources including the CIDAC, Base de Datos Electorales, and state-level electoral institutes for some states for some years. Municipal public finance and budget data were collected from the Instituto Nacional de Estadísticas y Geografía (INEGI) and Sistema Estatal y Municipal de Bases de Datos (SIMBAD), and migration, socioeconomic, and demographic indicators were obtained from the Censo Nacional de Población (CONAPO), Conteo de Población y Vivienda (CONTEO), and INEGI.

Data Sources and Model Specification for the Mexican Family Life Survey

The MxFLS is a longitudinal and multithematic survey taken over three panel waves (2002, 2005–2006, 2009–2012). It asks questions about community activities across a random sample of Mexican municipalities, data that are unavailable for all Mexican municipalities in the panel data set. The MxFLS community questionnaire includes a battery of questions related to community activities, including whether the community organizes activities, meetings, and assemblies and what types of community activities are organized (religious, political, social, or other). It also collects data on whether activities are more recent, occurring over the preceding 12-month period.¹² Over the MxFLS’s three panels, the total number of municipalities for which comprehensive data are available is 272, with some missing data. Analysis reveals that about 30 percent of MxFLS municipalities participate in the 3x1 Program during the study period.

To have a comprehensive data set with all the same controls and explanatory variables in the two data sources, we match the MxFLS sample with municipal data from the full panel data set we compiled. This allows us to assess how international migration changes the incidence of civic engagement and in turn the conditional effects of civic engagement on changes in voter turnout. This is possible because all places included in the MxFLS are matched with the same data from the full panel set. We focus the analysis on the level and recency of community activities to try to isolate the conditional effects of migrant absence and 3x1 participation on voter turnout.¹³

We measure the level and recency of community activities in several ways. First, we create an indicator for recent activity, which is coded “1” if the municipality reports community activities in the previous 12-month period and “0” otherwise. Second, activity level is a continuous variable reflecting the total number of community activities in which residents are most recently involved. Finally, we construct five separate dichotomous indicators for community civic engagement level based on the total number of community activities.¹⁴ The first dichotomous indicator (activity level 1) takes the value of “1” if the community has greater than or equal to the mean level of community activities, which is 10 activities, and “0” otherwise; the second indicator (activity level 2) takes the value of “1” if the community has greater than or equal to 20 activities and “0” otherwise; the third indicator (activity level 3) takes the value of “1” if the community has greater than or equal to 30 activities and “0” otherwise; the fourth indicator (activity level 4) takes the value of “1” if the community has greater than or equal to 50 activities and “0” otherwise; and the fifth activity level indicator (activity level 5) takes the value of “1” if the community has greater than or equal to 100 activities and “0” otherwise. In the full sample, minimum number of activities is 0, and the max is 800. About 46 percent of the sample has at least 10 activities, 25 percent 20 or more, 14 percent 30 or more, 11 percent 50 or more, and 7 percent more than 100. The coding of activity level as different dichotomous variables is for ease of interpretation of the interactions between civic engagement, migration channels, and voter turnout. Observations for all community variables reflect community activities before 2002 (Wave 1), between 2002 and 2004 (Wave 2), and 2005 to 2008 (Wave 3). This yields three separate observations for the civic engagement variables.

In the MxFLS analysis, we also include several indicators of 3x1 Program participation. First, 3x1 participation is equal to “1” if the municipality participates in the 3x1 Program in any year in the period before the survey years (Wave 2, 2002–2004 and Wave 3, 2005–2008) and “0” otherwise. Second, we include a cumulative treatment indicator that reflects the total number of years the municipality participated in the 3x1 Program between panel Waves 2 and 3. Finally, since the 3x1 Program database also contains information regarding the number of projects annually completed in each municipality, we include an additional continuous count variable as an indicator of program participation, called cumulative projects.

Community activity variables, which proxy for local civic engagement, are sourced from the MxFLS, while all other socioeconomic, demographic, public finance, political, and migration indicators are sourced from the panel data set described previously. All the same controls are included in the civic engagement models; however, controls are averaged for the pre-survey period for each panel wave. Since we have comprehensive data on all Mexican municipalities before and after the start of the 3x1 Program period, we also examine whether survey respondents are systematically different across our dependent and explanatory variables prior to the observation period (2002–2013). Results yield no concerns of selection bias.¹⁵

Effects of International Migration Channels on Local Voter Turnout

We report descriptive statistics for key variables included in the analysis for all municipalities in Supplemental Data Appendix S1. We start our analysis by exploring how international migration (absence) generally affected municipal voter turnout across all Mexican municipalities between 1990 and 2013, using the ordinary least squares (OLS) estimator with municipal and year fixed effects. In Table 1, we observe that over the full sample, international migration had a positive and statistically significant effect on voter turnout. For every 1 percent increase in the number of people emigrating, voter turnout increased by less than a percent, but this is hardly a substantive effect. We then split the sample to analyze how migration affected political participation before and after 2002 (the start of the 3x1 Program period). In these models, we find that the positive effect is driven by the post-2002 period since before 2002, migrants’ absence had no systematic effect on participation in local elections.¹⁶ After 2002, every 1 percent increase in international migrants increased voter turnout by 1.5 percent. For municipalities experiencing rapid demographic fluctuations of 10 percent, the corresponding increase in voter turnout would be as much as 15 percent. This positive effect contrasts with findings from previous research on the depressive nature of high migration on formal political participation in national elections (Bravo 2008; Goodman and Hiskey 2008).

Table 1.

Effects of International Migration on Voter Turnout, 1990–2013.

	Full Panel	Before 2002	After 2002
International migrants	0.884***	1.002	1.476***
	(0.246)	(0.770)	(0.224)
Poverty	4.931***	–5.550**	4.676*
	(1.457)	(2.037)	(2.383)
Poverty squared	–1.192*	0.738	–0.998
	(0.498)	(0.954)	(0.885)
Literacy rate	–0.410***	0.192	–0.800***
	(0.099)	(0.120)	(0.205)
Indigenous population	0.183*	0.025	0.979***
	(0.072)	(0.066)	(0.193)
Population (log)	19.392***	–2.716	12.849***
	(1.702)	(2.395)	(3.085)
Number of parties	3.188***	2.305***	4.445***
	(0.176)	(0.310)	(0.233)
Partido Revolucionario Internacional	–0.910**	–1.366***	–1.268*
	(0.305)	(0.288)	(0.545)
Partido de la Revolución Democrática	1.831***	0.306	1.198⁺
	(0.446)	(0.416)	(0.723)
Partido Acción Nacional	0.876*	–0.226	1.999**
	(0.359)	(0.297)	(0.704)
Tax revenue (per cap)	–0.002***	–0.002***	–0.000
	(0.000)	(0.000)	(0.000)
Constant	–92.449***	64.031**	–22.648
	(16.522)	(22.578)	(33.870)
Observations	35,296	17,534	17,762

Note. Robust standard errors in parentheses. All models include state and time fixed effects.

⁺p < .10. *p < .05. **p < .01. ***p < .001.

Initial results could indicate that the absence of specific migrants creates new political space for local citizens to participate in politics. It could also indicate that absence is capturing how social remittance transfers to households through migrant social networks influences the voting behavior of nonmigrant recipients. We are unable to adjudicate between these two mechanisms given the unavailability of a direct measure for social remittances, but the evidence does suggest that the aggregate effect of migrant absence across Mexican municipalities is positively associated with voter turnout. To delve more deeply into how other channels are driving the results, we first analyze the full panel (1990–2013) with the DiD estimation and then dive into the post-2002 period, for which we have complete data for the 3x1 Program and other channels of international migration.

We examine first how migrant return, circular migration, and family and collective remittances drive voter turnout. Since a key indicator of the transnational channel is municipal participation in the 3x1 Program, we must approach the statistical analysis with a different estimator given the likelihood of selection bias often present when using quasi-experimental data. Since our indicator of collective remittances is municipal participation in the 3x1 Program, self-selection into the program may be confounded with the outcomes of interest. The DiD and a fixed effects estimator alleviate many of these concerns.¹⁷ The DiD estimator is a simple and powerful tool for estimating treatment effects with observational data (Ashenfelter 1978; Buckley and Shang 2003; Imbens and Wooldridge 2009).¹⁸ Here, we use it to measure the difference of voter turnout between municipalities that participated in the 3x1 Program regularly (treatment group) (i.e., at least once per three years) and those that did not participate regularly (control group). In other words, all fixed, time-constant characteristics that are correlated with voter turnout and migration are differenced out, and we reduce concern for unobserved bias.¹⁹ This strategy helps ensure that variables remaining constant over time (but unobserved) and correlated with both the decision to participate in the program and the outcomes of interest do not bias the estimated effect (Buckley and Shang 2003). Figure 2 graphs the time trends between 1990 and 2013 by treatment and control group for voter turnout. As it shows, pretreatment trends are similar and stable before 2002 but become different after 2002, providing evidence that the data observe the parallel trend assumption.

Figure 2.

Voter Turnout Trend.

Although our DiD models mitigate migration’s selection effects in the full panel, this approach cannot fully consider the temporal dynamics of 3x1 participation. Thus, we also evaluate the effects of 3x1 participation using fixed effects models with a municipal linear time trend in the post-2002 period. Different from our handling of the DiD models, we focus only on municipalities that participated in the 3x1 Program between 2002 and 2013. In this way, we can fully isolate impact from the selection into the 3x1 Program because the comparison is done with only those municipalities that participated in the 3x1 Program.

In the fixed effects and DiD models,²⁰ we assess how different indicators of international migration affect local participation in municipal elections and present findings in Table 2. The signs are consistent across both sets of models, except for family remittances, which is only significant with the DID approach. Since the DiD approach is mainly used to assess the effects of 3x1 participation on voter turnout, we focus on the fixed effects models to interpret the findings related to other migration channels.²¹ First, we find that migrant absence continues to have a positive impact on voter turnout and that the effects are stronger in the restricted period at 6 percent. We also see, however, that another feature of international migration, the return channel, has significant negative effects. Second, circular migration has a positive, statistically significant, and substantive effect. For every 1 percent increase in households with circular migrants, voter turnout increases by 3 percent. Third, and in contrast, as more migrants return to their households, voter turnout decreases between 4 and 5 percent. Fourth, migrant family remittances have a negative effect on voter turnout in the DiD models but do not reach substantive significance. We find no systematic effect from family remittances, one indicator of the transnational channel, in any of the fixed effects models, which we believe are more robust estimates overall.

Table 2.

Difference-in-Difference (DiD) and Fixed Effects Models, Effects of 3x1 Program and International Migration on Voter Turnout.

	Ever Participate 3x1 Program (DiD Model)	Electoral Cycle 3x1 Program Participation	Cumulative 3x1 Program Participation
3x1 Program	0.482	0.368	1.395**
	(0.317)	(0.316)	(0.453)
Post-2002	–9.731***	—	—
	(0.292)	—	—
3x1 Program × post-2002	–0.138	—	—
	(0.448)	—	—
International migrants	2.266***	6.618***	6.125***
	(0.147)	(0.939)	(0.934)
Family remittances	–0.047**	–0.783	–0.827
	(0.016)	(0.811)	(0.809)
Circular migrants	0.243***	3.226*	3.011*
	(0.050)	(1.390)	(1.402)
Return migrants	–0.374***	–5.391***	–4.567***
	(0.068)	(1.108)	(1.102)
Poverty	1.024***	19.464*	20.624*
	(0.173)	(9.308)	(9.245)
Poverty squared	–0.613***	–11.497**	–11.712**
	(0.073)	(4.033)	(4.048)
Literacy rate	0.250***	–1.658*	–1.556*
	(0.017)	(0.689)	(0.687)
Indigenous population	0.091***	1.096⁺	1.011⁺
	(0.004)	(0.576)	(0.545)
Population (log)	–3.582***	5.598	8.058
	(0.078)	(11.894)	(11.927)
Number of parties	2.085***	4.745***	4.712***
	(0.088)	(0.376)	(0.373)
Partido Revolucionario Internacional	–0.090	–1.163⁺	–1.083⁺
	(0.228)	(0.637)	(0.637)
Partido de la Revolución Democrática	2.543***	0.990	1.029
	(0.313)	(0.847)	(0.850)
Partido Acción Nacional	2.908***	0.834	0.855
	(0.263)	(0.857)	(0.853)
Tax revenue (per cap)	0.000*	0.001*	0.000*
	(0.000)	(0.000)	(0.000)
Constant	68.689***	119.938	96.558
	(1.621)	(110.068)	(110.024)
Observations	35,296	10,087	10,087

Note. Fixed effects models with municipal-specific time trend include state fixed effects and municipal-specific time trend. Robust standard errors in parentheses.

⁺p < .10. *p < .05. **p < .01. ***p < .001.

Finally, results in the DiD and fixed effects models show that municipal participation in the 3x1 Program has neither a systematic effect on voter turnout across Mexican municipalities nor a short-term (three-year electoral cycle) effect. However, in Table 2 (Column 3), results show that every additional year of participation in the 3x1 Program is associated with a 1.4 percent increase in voter turnout. Among municipalities that participated in 3x1, the average number of annual years of participation is four, suggesting that the cumulative effects of frequent participation would lead to a 6 percent increase in voter turnout. If a municipality participated in the 3x1 Program 10 times, for example (about 10% of total participants), turnout is associated with a 14 percent increase.

Effects of International Migration and Civic Engagement on Local Voter Turnout

In the MxFLS specifications, when the dependent variable is continuous, OLS is used to estimate the effects. Logistic regression is used when the outcome variable is dichotomous. Results presented in Table 3 suggest a positive but weak relationship between the number of 3x1 public goods projects and the odds of having recent community activities (the reference group is no recent activity). For every additional 3x1 project carried out in a municipality, the odds of having recent community activity increases by 44 percent.

Table 3.

Effects of 3x1 Program Participation on Civic Engagement.

	Incidence of Recent Civic Engagement			Civic Engagement Activity Levels (>10 Activities)
3x1 participation	1.082	—	—	0.885	—	—
	(0.647)	—	—	(0.404)	—	—
Cumulative 3x1	—	1.366	—	—	0.899	—
	—	(0.388)	—	—	(0.173)	—
Cumulative project	—	—	1.440⁺	—	—	1.089
	—	—	(0.313)	—	—	(0.092)
Voter turnout	0.995	0.995	0.994	0.986	0.986	0.985
	(0.017)	(0.017)	(0.017)	(0.014)	(0.014)	(0.014)
Victory margin	1.000	1.000	1.000	1.022*	1.022*	1.022*
	(0.011)	(0.011)	(0.011)	(0.009)	(0.009)	(0.009)
Poverty	0.936	1.015	1.060	0.473	0.461	0.498
	(0.947)	(1.031)	(1.083)	(0.383)	(0.373)	(0.401)
Poverty squared	1.015	1.048	1.065	1.431	1.421	1.508
	(0.388)	(0.401)	(0.409)	(0.442)	(0.435)	(0.468)
International migrants	1.014	0.843	0.742	1.785*	1.862*	1.569⁺
	(0.335)	(0.288)	(0.245)	(0.407)	(0.452)	(0.362)
Literacy rate	1.099	1.121	1.128	0.936	0.931	0.939
	(0.130)	(0.132)	(0.133)	(0.086)	(0.086)	(0.086)
Indigenous population	1.005	1.007	1.007	1.027	1.026	1.028
	(0.022)	(0.022)	(0.022)	(0.018)	(0.018)	(0.018)
Population (log)	0.518*	0.496*	0.487**	1.088	1.101	1.053
	(0.141)	(0.137)	(0.134)	(0.197)	(0.201)	(0.192)
Tax revenue (per cap lagged)	1.002	1.001	1.001	0.999	0.999	0.999
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Constant	1.482	0.457	0.373	23.267	32.061	25.497
	(14.478)	(4.419)	(3.591)	(185.757)	(256.506)	(202.283)
Observations	166	166	166	145	145	145

Note. Robust standard errors in parentheses.

Source. Mexican Family Life Survey and panel data collected by authors from Inegi Simbad, Cidac/Cide, Conapo, and Sedesol.

⁺p < .10. *p < .05. **p < .01.

Results for migrant absence reveal that increasing emigration is not associated with recent changes in levels of civic engagement. In fact, this is consistent across all other indicators of migration (return, family remittances, and circularity). Rather, our findings show that the preexisting level of civic engagement is positively associated with increasing emigration, which supports Goodman and Hiskey’s (2008) conclusion that as migration intensity increases, local citizens may turn their attention to civic associational life in the hometown. Additionally, since migrant absence is also related to nonmigrant social ties to family members in the United States, our results additionally indicate that increasing social contact with migrants abroad positively affects the likelihood of community civic engagement. Taken together, results for the civic engagement models provide compelling evidence that after 2002, HTAs were more likely to drive recent changes in local civic engagement but that high migration and social remittances likely contributed to preexisting “stocks” of civic engagement that often accumulate over time.

The final analyses using the MxFLS examines if 3x1 Program participation and international migration (absence) affect voter turnout conditional on civic engagement. We do not report each migration indicator’s conditional effects due to space constraints as they have no observable impact. Table 4 presents the findings of the interaction models and shows significant positive effects on voter turnout. 3x1 Program participation is associated with an 11.9 percent increase in voter turnout, conditional on 10 or more community activities, and a 12.3 percent increase when community activities reach 100 or more. The positive interaction effect is consistent with cumulative participation in the 3x1 Program and total number of projects. Conditional on having 100 or more activities, cumulative 3x1 participation is associated with a 6 percent increase in local voter turnout. Cumulative 3x1 projects are associated with a 4 percent increase when conditional on 50 or more activities (a quarter of the sample) and a 7 percent increase when conditional on 100 activities or more. As frequency of 3x1 participation increases with civic engagement, voter turnout increases even more.

Table 4.

Conditional Effects of 3x1 Program Participation, Projects, and Civic Engagement on Voter Turnout.

	3x1 Program Participation			3x1 Cumulative Participation			3x1 Cumulative Projects
3x1 program participation	–2.307	2.148	2.296	–0.316	1.141	1.133	–0.307	0.143	0.242
	(3.734)	(3.209)	(3.144)	(1.783)	(1.494)	(1.485)	(0.891)	(0.547)	(0.566)
Number of parties	0.200	1.330	1.291	0.487	1.328	1.304	0.367	1.187	1.182
	(2.311)	(2.266)	(2.290)	(2.299)	(2.247)	(2.273)	(2.317)	(2.233)	(2.260)
Poverty	8.932	8.640	8.711	8.490	8.664	8.676	7.377	7.877	8.177
	(6.241)	(6.546)	(6.567)	(6.331)	(6.535)	(6.550)	(6.382)	(6.494)	(6.494)
Poverty squared	2.559	1.554	1.484	2.338	1.477	1.434	1.859	1.403	1.280
	(2.232)	(2.354)	(2.370)	(2.247)	(2.341)	(2.348)	(2.254)	(2.323)	(2.331)
International migrants	3.741*	3.283⁺	3.290⁺	3.594*	3.047	3.070	4.096*	3.537⁺	3.475⁺
	(1.586)	(1.751)	(1.741)	(1.769)	(1.922)	(1.923)	(1.885)	(1.905)	(1.897)
Literacy rate	1.272⁺	1.424*	1.424*	1.274⁺	1.450*	1.445*	1.205⁺	1.309⁺	1.352⁺
	(0.645)	(0.708)	(0.708)	(0.659)	(0.709)	(0.708)	(0.678)	(0.713)	(0.712)
Indigenous population	0.363**	0.366**	0.368**	0.362**	0.369**	0.369**	0.372**	0.353**	0.359**
	(0.118)	(0.128)	(0.128)	(0.120)	(0.125)	(0.126)	(0.127)	(0.130)	(0.130)
Population (log)	1.048	0.900	1.078	1.039	0.843	0.976	1.122	0.906	1.090
	(1.331)	(1.352)	(1.375)	(1.338)	(1.363)	(1.383)	(1.335)	(1.364)	(1.393)
Tax revenue (per cap)	–0.003	–0.002	–0.002	–0.003	–0.002	–0.002	–0.002	–0.002	–0.002
	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
10+ activities	–8.770*	—	—	–6.954*	—	—	–5.879⁺	—	—
	(3.486)	—	—	(3.244)	—	—	(3.013)	—	—
3x1 × 10+ activities	11.940*	—	—	3.246	—	—	0.866	—	—
	(5.143)	—	—	(2.236)	—	—	(0.975)	—	—
50+ activities	—	0.682	—	—	0.699	—	—	–0.468	—
	—	(4.525)	—	—	(4.484)	—	—	(4.321)	—
3x1 × 50+ activities	—	6.167	—	—	5.446	—	—	4.111***	—
	—	(9.429)	—	—	(4.528)	—	—	(1.210)	—
100+ activities	—	—	0.504	—	—	0.554	—	—	0.112
	—	—	(3.775)	—	—	(3.735)	—	—	(3.684)
3x1 × 100+ activities	—	—	12.299*	—	—	6.069*	—	—	7.076**
	—	—	(5.136)	—	—	(2.620)	—	—	(2.250)
Constant	–69.283	–86.151	–87.792	–70.638	–87.561	–88.444	–66.022	–75.760	–81.204
	(53.956)	(58.880)	(58.979)	(55.071)	(58.797)	(58.903)	(56.486)	(59.199)	(59.159)
Observations	145	145	145	145	145	145	145	145	145

Note. Robust standard errors in parentheses.

⁺p < .10. *p < .05. **p < .01. ***p < .001.

Table 5 shows the results for the interaction models with international migration (absence). Results show that migrant absence has the opposite conditional effect on voter turnout in places with high levels of civic engagement. In places with 100 or more community activities, migrant absence is associated with a 16 percent decline in voter turnout. Beyond very high levels of community civic engagement, migrant absence has no conditional effects on voter turnout for any other level or type of community activity. In places where citizens are highly engaged in religious, political, and social associations, nonmigrant citizens are more likely to withdraw from participation in formal electoral politics with increasing out-migration.

Table 5.

Conditional Effects of International Migrants and Civic Engagement on Voter Turnout.

	3x1 Program Participation			3x1 Cumulative Participation			3x1 Cumulative Projects
3x1 program participation	2.543	2.829	2.754	1.171	1.410	1.310	0.351	0.301	0.280
	(3.047)	(3.009)	(3.039)	(1.435)	(1.445)	(1.449)	(0.552)	(0.568)	(0.574)
International migrants	2.972	3.040⁺	3.171⁺	2.799	2.759	2.938	3.182	3.259⁺	3.398⁺
	(1.916)	(1.715)	(1.728)	(2.020)	(1.880)	(1.900)	(1.950)	(1.882)	(1.902)
Number of parties	0.594	1.268	1.554	0.608	1.290	1.562	0.464	1.154	1.437
	(2.339)	(2.233)	(2.289)	(2.321)	(2.214)	(2.272)	(2.307)	(2.205)	(2.262)
Poverty	8.165	8.824	8.729	8.088	8.782	8.662	7.560	8.165	8.082
	(6.475)	(6.479)	(6.584)	(6.454)	(6.471)	(6.570)	(6.406)	(6.408)	(6.516)
Poverty squared	2.041	1.510	1.627	1.963	1.437	1.557	1.849	1.268	1.390
	(2.290)	(2.345)	(2.358)	(2.279)	(2.332)	(2.341)	(2.267)	(2.315)	(2.326)
Literacy rate	1.297⁺	1.443*	1.431*	1.317⁺	1.471*	1.452*	1.228⁺	1.356⁺	1.345⁺
	(0.671)	(0.701)	(0.709)	(0.671)	(0.703)	(0.710)	(0.674)	(0.707)	(0.715)
Indigenous population	0.377**	0.366**	0.367**	0.378**	0.368**	0.368**	0.370**	0.355**	0.356**
	(0.130)	(0.127)	(0.128)	(0.128)	(0.124)	(0.126)	(0.131)	(0.129)	(0.130)
Population (log)	1.186	0.921	0.884	1.083	0.781	0.772	1.134	0.909	0.881
	(1.353)	(1.357)	(1.355)	(1.346)	(1.353)	(1.356)	(1.357)	(1.373)	(1.371)
Tax revenue (per cap)	–0.002	–0.002	–0.002	–0.002	–0.002	–0.002	–0.002	–0.002	–0.002
	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
10+ activities	–6.062	—	—	–5.965	—	—	–6.099	—	—
	(3.998)	—	—	(3.995)	—	—	(3.991)	—	—
% of migrants × 10+ activities	1.417	—	—	1.370	—	—	1.221	—	—
	(2.110)	—	—	(2.069)	—	—	(2.102)	—	—
50+ activities	—	–3.712	—	—	–3.509	—	—	–3.594	—
	—	(8.938)	—	—	(8.903)	—	—	(8.974)	—
% of migrants × 50+ activities	—	9.590	—	—	9.797	—	—	9.338	—
	—	(12.465)	—	—	(12.559)	—	—	(12.580)	—
100+ activities	—	—	11.190⁺	—	—	11.149⁺	—	—	10.975⁺
	—	—	(6.063)	—	—	(5.995)	—	—	(6.138)
% of migrants × 100+ activities	—	—	–15.904*	—	—	–15.798*	—	—	–15.865*
	—	—	(6.820)	—	—	(6.761)	—	—	(7.156)
Constant	–75.080	–87.522	–87.264	–75.588	–88.324	–87.766	–68.072	–79.323	–79.287
	(56.108)	(58.402)	(59.230)	(56.075)	(58.458)	(59.201)	(56.078)	(58.623)	(59.486)
Observations	145	145	145	145	145	145	145	145	145

Note. Robust standard errors in parentheses.

⁺p < .10. *p < .05. **p < .01.

Discussion

These results provide compelling evidence that international migration has important political consequences for noncitizens’ political and civic participation in local democracy. From 1990 to 2013, as international migration intensity increased across Mexican municipalities, we find that more citizens were voting in local elections, especially after 2002. When we look more closely at the post-2002 period, the data show that migrant absence from the hometown continues to explain variation in municipal voter turnout rates but that other migration channels have additional effects, all other things equal. The continued positive and substantive effect of absence suggests that migrants who leave may be creating space for new groups to engage in local politics or that those who leave are not producing political brain drain. It could also be that emigration’s depressing effects on political participation is mitigated by the influx of social remittances — ideas, social capital, norms, and political behaviors transmitted through migrant social networks from abroad. The positive effect of circular migration supports this finding since circular migrants are likely to transmit ideational resources and human and physical capital in situ during their temporary return to the hometown before departing once again for the United States.

Our findings also reveal that not all migration channels have the same effect on voter turnout. While family remittances are negatively correlated with voter turnout in the DiD approach, the effect is not substantive and the indicator fails to reach statistical significance in the fixed effects models, which we believe are more robust. Additionally, we find that more permanent return of migrants from the United States has a negative relationship with voter turnout, which contrasts with previous studies (Chauvet and Mercier 2014; Waddell and Fontenla 2015; García 2017). One interpretation is that returnees are more likely to be perceived in a negative light upon reintegration into the hometown, and thus, their political messaging about norms of democratic engagement fall on deaf ears. It could also be that on average, after reintegration, returnees become politically withdrawn over time as a function of the political institutional environment, which may have additional negative spillover effects on their social contacts.

Finally, migrant HTA involvement in public goods provision in the hometown through the 3x1 Program is associated with a significant increase in local participation in municipal elections and recent civic engagement but only with frequent program participation. This result suggests a social and political learning process that happens over time with repeated interactions between migrant groups, resident citizens, and political officials. As municipalities engage in public goods activities with migrant HTAs more frequently and citizens become aware of and involved in the process, they may use the formal political process to reward or punish political officials for their performance during 3x1 project provision. Repeated social interactions between migrants, local political officials, and residents during project coordination bring citizens more into project governance, including decision making, information sharing, and monitoring of political officials in office, suggesting a strong positive spillover effect on civic and political engagement from continuous migrant transnational collective action in the hometown (Duquette-Rury 2016).

Taken together, these results show that international migration has important effects on citizens’ formal participation in local elections but also that certain channels depress voter turnout (return, family remittances) while others encourage more formal political activity (circular migration, migrant cross-border collective action). Moreover, in results not reported, we find additional evidence that migration intensity (all the migration indicators in one scalar index) has a positive relationship with voter turnout in the aggregate. We conducted this additional analysis in the likely case that different channels had spillover effects on nonmigrant households. While showing a great deal about how different channels of international migration affect voter turnout, our study evaluates only formal electoral behavior and civic engagement and thus does not speak to partisan identity, vote choice, or campaign contributions. These areas are ripe for future research.

Conclusion

Generally, findings from this study suggest that international migration produces different channels that stimulate nonmigrant citizens’ political participation in local elections. While previous research has identified a positive link between migrant absence and nonelectoral forms of political engagement (Goodman and Hiskey 2008; Pérez-Armendáriz and Crow 2010), research has also found either no empirical effect or negative associations between emigration, remittances, and formal voting behaviors (Bravo 2008; Goodman and Hiskey 2008; Pérez-Armendáriz and Crow 2010; Córdova and Hiskey 2015). In this article, we examined multiple channels through which international migration likely affects nonmigrant citizens’ formal participation in local elections. In particular, we assessed how circular and return migration, migrant absence, and family and collective remittances change voter turnout across municipalities over time.

We find that migrant absence and return channels have consequences for political behavior across and within Mexican municipalities with significant emigration but in opposing ways. As migrants return voluntarily or are forced to return home, voter turnout is more likely to decline. By contrast, in places with higher rates of emigration and more circular, temporary migration, more nonmigrants are formally participating in local elections. However, circular migration’s positive effect on political participation needs to tempered since increased securitization of the US border hampers regular border crossings. Moreover, as the United States continues to escalate interior immigration enforcement and more Mexicans are deported, the number of Mexican returnees is likely to continue. As of 2015, more Mexicans were returning to Mexico than crossing the border into the United States (Gonzalez-Barrera 2015). US border and interior immigration policy thus has important implications for understanding Mexican political engagement.

This research also reveals that migrant HTAs’ involvement in local public goods provision through the federal 3x1 Program stimulates voting but that family remittances have no systematic effects on voter turnout. To our knowledge, this is the first systematic evidence that migrant cross-border collective action has observable consequences on local civic and political participation in migrant hometowns (see also Duquette-Rury 2016). Future research assessing local political participation in Mexico and beyond will need to take seriously migrant HTAs’ collective involvement in local public goods provision as this kind of cross-border practice has unintended but dynamic consequences for local democratic participation.

Since existing research identifies an important link between increasing emigration and nonelectoral engagement (Goodman and Hiskey 2008; Pérez-Armendáriz and Crow 2010; Córdova and Hiskey 2015), we also looked closely at the ways in which different channels of migration shaped civic engagement. Moreover, we assessed how civic engagement played an intervening role in citizens’ participation in local elections. The analysis using a secondary data source, the Mexican Family Life Survey, proved instructive as we learned that only migrant collective involvement in the hometown via the 3x1 Program was associated with recent forms of civic engagement. We also observed that high levels of civic engagement amplified the positive effects of 3x1 participation on local electoral participation but depressed formal voting in high-migration municipalities without 3x1 participation. In 3x1 municipalities, collective engagement in local public goods provision mobilized by migrant HTAs was one steppingstone to engaging in formal politics. Through the co-provision of public goods process with migrants and the local government, noncitizens access information and develop interest in political activity, motivating their participation in the formal local democratic process.

While our research has revealed an important association between different channels of international migration and formal political participation in Mexico, there is more research to be done. First, migrant absence, return, and diaspora channels are likely to affect multiparty elections across places with varying rates of out-migration. As more nonmigrant citizens become involved in electoral politics, political parties will use different strategies to court their votes or lose out to the opposition. This presents an opportunity to study how migratory pathways affect local electoral competition and which political parties win and lose from a more politically active citizenry. Second, our findings suggest that future research should examine whether different migration channels affect local democratic quality more generally. Does the increase in electoral participation induced by migration lead to more accountable, responsive local government, or does more citizen involvement simply indicate new clients for local political party patrons interested in corraling more voters to the polls in what has been termed turnout buying (Nichter 2008)? Future research should further investigate the political repercussions of different migratory pathways on other dimensions of local democratic functioning, including partisan identity and mobilization, party competition, and local clientelism.

Supplemental Material

Supplemental Material, MRX774499_Supplemental_Material - Does International Migration Affect Political Participation? Evidence from Multiple Data Sources across Mexican Municipalities, 1990–2013

Supplemental Material, MRX774499_Supplemental_Material for Does International Migration Affect Political Participation? Evidence from Multiple Data Sources across Mexican Municipalities, 1990–2013 by Lauren Duquette-Rury and Zhenxiang Chen in International Migration Review

Footnotes

Acknowledgments

We would like to thank three anonymous reviewers for helpful comments that improved the paper and Jesse Acevedo (UCLA) for recommending we use the difference-in-difference approach in our analysis. All errors are our own.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

The Supplemental Data Appendix is available with the online version of this article at .

Notes

References

Adams

R. H.

Jr Page

. 2005. “Do International Migration and Remittances Reduce Poverty in Developing Countries?” World Development 33(10):1645–69.

Alesina

Baqir

Easterly

. 1999. “Public Goods and Ethnic Divisions.” The Quarterly Journal of Economics 114(4):1243–84.

Ashenfelter

1978. “Estimating the Effect of Training Programs on Earnings.” The Review of Economics and Statistics 60(1):47–57.

Bada

2014. Mexican Hometown Associations in Chicagoacán: From Local to Transnational Civic Engagement. New Brunswick, NJ: Rutgers University Press.

Bardhan

2002. “Decentralization of Governance and Development.”; Journal of Economic Perspectives 16(4):185–205.

Batista

Vicente

P. C.

. 2011. “Do Migrants Improve Governance at Home? Evidence from a Voting Experiment.” The World Bank Economic Review 25(1):77–104.

Brady

H. E.

Verba

Schlozman

K. L.

. 1995. “Beyond SES: A Resource Model of Political Participation.” American Political Science Review 89(2):271–94.

Bravo

2008. “The Political Economy of Mexico-US Migration.” PhD diss., Duke University.

Buckley

Shang

. 2003. “Estimating Policy and Program Effects with Observational Data: The ‘Differences-in-Differences’ Estimator.” Practical Assessment, Research & Evaluation 8(24):1–10.

10.

Burgess

2016. “Organized Migrants and Accountability from Afar.” Latin American Research Review 51(2):150–73.

11.

Card

Krueger

A. B.

. 2000. “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Reply.” American Economic Review 90(5):1397–420.

12.

Careja

Emmenegger

. 2012. “Making Democratic Citizens: The Effects of Migration Experience on Political Attitudes in Central and Eastern Europe.” Comparative Political Studies 45(7):875–902.

13.

Chauvet

Mercier

. 2014. “Do Return Migrants Transfer Political Norms to Their Origin Country? Evidence from Mali.” Journal of Comparative Economics 42(3):630–51.

14.

Cleary

M. R.

2010. The Sources of Democratic Responsiveness in Mexico. Notre Dame, IN: University of Notre Dame Press.

15.

Córdova

Hiskey

. 2015. “Shaping Politics at Home: Cross-Border Social Ties and Local-Level Political Engagement.” Comparative Political Studies 48(11):1454–87.

16.

Diamond

Morlino

, eds. 2005. Assessing the Quality of Democracy. Baltimore, MD: Johns Hopkins University Press.

17.

Dionne

K. Y.

Inman

K. L.

Montinola

G. R.

. 2014. Another Resource Curse? The Impact of Remittances on Political Participation. Ghana: Afrobarometer.

18.

Duquette-Rury

2014. “Collective Remittances and Transnational Coproduction: The 3x1 Program for Migrants and Household Access to Public Goods in Mexico.” Studies in Comparative International Development 49(1):112–39.

19.

Duquette-Rury

. 2016. “Migrant Transnational Participation: How Citizen Inclusion and Government Engagement Matter for Local Democratic Development in Mexico.” American Sociological Review 81(4):771–99.

20.

Durand

Parrado

E. A.

Massey

D. S.

. 1996. “Migradollars and Development: A Reconsideration of the Mexican Case.” International Migration Review 30(2):423–44.

21.

Fajnzylber

Humberto

L. J.

. 2007. Close to Home: The Development Impact of Remittances in Latin America. Washington, DC: World Bank.

22.

Falleti

T. G.

2005. “A Sequential Theory of Decentralization: Latin American Cases in Comparative Perspective.” American Political Science Review 99(3):327–46.

23.

García

A. I. L.

2017. “Economic Remittances, Temporary Migration and Voter Turnout in Mexico.” Migration Studies 6(1):20–52.

24.

Germano

2013. “Migrants’ Remittances and Economic Voting in the Mexican Countryside.” Electoral Studies 32(4):875–85.

25.

Giraudy

2010. “The Politics of Subnational Undemocratic Regime Reproduction in Argentina and Mexico.” Journal of Politics in Latin America 2(2):53–84.

26.

Goldring

2002. “The Mexican State and Transmigrant Organizations: Negotiating the Boundaries of Membership and Participation.” Latin American Research Review 37(3):55–99.

27.

Goodman

G. L.

Hiskey

J. T.

. 2008. “Exit without Leaving: Political Disengagement in High Migration Municipalities in Mexico.” Comparative Politics 40(2):169–88.

28.

Gonzalez-Barrera

2015. More Mexicans Leaving Than Coming to the US. Washington, DC: Pew Research Center.

29.

Guarnizo

L. E.

Portes

Haller

. 2003. “Assimilation and Transnationalism: Determinants of Transnational Political Action among Contemporary Migrants.” American Journal of Sociology 108(6):1211–48.

30.

Hastings

J. S.

2004. “Vertical Relationships and Competition in Retail Gasoline Markets: Empirical Evidence from Contract Changes in Southern California.” American Economic Review 94(1):317–28.

31.

Hiskey

J. T.

2003. “Demand-Based Development and Local Electoral Environments in Mexico.” Comparative Politics 36(1):41–59.

32.

Holzner

C. A.

2010. Poverty of Democracy: The Institutional Roots of Political Participation in Mexico. Pittsburgh: University of Pittsburgh Press.

33.

Imbens

G. W.

Wooldridge

J. M.

. 2009. “Recent Developments in the Econometrics of Program Evaluation.” Journal of Economic Literature 47(1):5–86.

34.

Iskander

2010. Creative State: Forty Years of Migration and Development Policy in Morocco and Mexico. Ithaca, NY: Cornell University Press.

35.

Itzigsohn

2000. “Immigration and the Boundaries of Citizenship: The Institutions of Immigrants’ Political Transnationalism.” International Migration Review 34(4):1126–54.

36.

Kapur

2010. Diaspora, Development, and Democracy: The Domestic Impact of International Migration from India. Princeton, NJ: Princeton University Press.

37.

Laakso

Taagepera

. 1979. “‘Effective’ Number of Parties: A Measure with Application to West Europe.” Comparative Political Studies 12(1):3–27.

38.

Lafleur

J.-M.

2013. Transnational Politics and the State: The External Voting Rights of Diasporas. London: Routledge Press.

39.

Levitt

1998. “Social Remittances: Migration Driven Local-Level Forms of Cultural Diffusion.” International Migration Review 32:(4):926–48.

40.

Levitt

. 2001. The Transnational Villagers. Los Angeles: University of California Press.

41.

Meseguer

Aparicio

F. J.

. 2012. “Migration and Distributive Politics: The Political Economy of Mexico’s 3x1 Program.” Latin American Politics and Society 54(4):147–78.

42.

Nichter

2008. “Vote Buying or Turnout Buying? Machine Politics and the Secret Ballot.” American Political Science Review 102(1):19–31.

43.

Nyblade

O’Mahony

. 2014. “Migrants’ Remittances and Home Country Elections: Cross-National and Subnational Evidence.” Studies in Comparative International Development 49(1):44–66.

44.

Oliver

J. E.

1999. “The Effects of Metropolitan Economic Segregation on Local Civic Participation.” American Journal of Political Science 43(1):186–212.

45.

Passel

J. S.

Cohn

Gonzalez-Barrera

. 2012. Net Migration from Mexico Falls to Zero—and Perhaps Less. Washington, DC: Pew Hispanic Center.

46.

Pérez-Armendáriz

2014. “Cross-Border Discussions and Political Behavior in Migrant-Sending Countries.” Studies in Comparative International Development 49(1):67–88.

47.

Pérez-Armendáriz

Crow

. 2010.“Do Migrants Remit Democracy? International Migration, Political Beliefs, and Behavior in Mexico.” Comparative Political Studies 43(1):119–48.

48.

Pfutze

2012. “Does Migration Promote Democratization? Evidence from the Mexican Transition.” Journal of Comparative Economics 40:159–75.

49.

Piper

2009. “The Complex Interconnections of the Migration-Development Nexus: A Social Perspective.” Population, Space and Place 15(2):93–101.

50.

Portes

Escobar

Radford

A. W.

. 2007. “Immigrant Transnational Organizations and Development: A Comparative Study.” International Migration Review 41(1):242–81.

51.

Powell

G. B.

1986. “American Voter Turnout in Comparative Perspective.” American Political Science Review 80(1):17–43.

52.

Putnam

1994. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press.

53.

Ratha

Eigen-Zucchi

Plaza

. 2016. Migration and Remittances Factbook 2016. Washington, DC: World Bank Publications.

54.

Rother

2009. “Changed in Migration? Philippine Return Migrants and (Un) Democratic Remittances.” European Journal of East Asian Studies 8(2):245–74.

55.

Schiller

N. G.

Basch

Blanc

C. S.

. 1995. “From Immigrant to Transmigrant: Theorizing Transnational Migration.” Anthropological Quarterly 68(1):48–63.

56.

Simpser

Duquette-Rury

Hernandez Company

Ibarra

J. F.

. 2016. “The Political Economy of Social Spending by Local Government: A Study of the 3x1 Program in Mexico.” Latin American Research Review 51(1):62–83.

57.

Smith

2006. Mexican New York: Transnational Lives of New Immigrants. Los Angeles: University of California Press.

58.

Treisman

2002. “Decentralization and the Quality of Government.” Unpublished paper, Department of Political Science, UCLA.

59.

Waddell

B. J.

Fontenla

. 2015. “The Mexican Dream? The Effect of Return Migrants on Hometown Development.” The Social Science Journal 52(3):386–96.

60.

Watson

. “Public Health Investments and the Infant Mortality Gap: Evidence from Federal Sanitation Interventions on US Indian Reservations.” Journal of Public Economics 90(8–9):1537–60.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

Does International Migration Affect Political Participation? Evidence from Multiple Data Sources across Mexican Municipalities,1990–2013

Abstract

Introduction

Political Consequences of International Migration in Origin Countries

International Migration Channels and Political Participation

Research Methods, Indicators, and Sources for the Mexican Migration Panel Data Set

Controlling for Confounders

Data Sources and Model Specification for the Mexican Family Life Survey

Effects of International Migration Channels on Local Voter Turnout

Effects of International Migration and Civic Engagement on Local Voter Turnout

Discussion

Conclusion

Supplemental Material

Supplemental Material, MRX774499_Supplemental_Material - Does International Migration Affect Political Participation? Evidence from Multiple Data Sources across Mexican Municipalities, 1990–2013

Footnotes

Acknowledgments

Declaration of Conflicting Interests

Funding

Supplemental Material

Notes

References

Supplementary Material