Drawing on panel data from six German elections from 1998 to 2017, this study investigates the correlation between far-left positioned parties and immigration per capita using the occurring elections. Our main findings suggest a positive correlation between immigrants per capita and the voting outcomes of the far-left parties.
Given the controversy about Chancellor Merkel's open immigration policies during the peak of the refugee crisis in 2015 and 2016 in Germany, Western nations have more openly than ever responded to existing fears around immigration, favoring nationalist philosophies over more liberal ideologies. A critical determinant of the debate is driven by high xenophobia, where individuals become anxious about ethnic minorities and greater diversity, leading to the intolerance of migrants.
This has a direct impact on political stability and future electoral voting outcomes. One strain of research has focused on voting outcomes of immigrant inflows, mostly focusing on right-wing parties and the associated rise of voter bases halla2017immigration, arzheimer2019alternative. Arzheimer and Berning (2019), for instance, demonstrate that the primary motivation for voting for the Alternative für Deutschland (AfD), a newly established right-wing populist party in Germany, is the voters' negative immigration attitudes. Little research, however, has focused on the influence of voting outcomes for far-left positioned parties in Germany.
An Examination of the Influence of Immigrants on Voting Outcomes of Left-wing Parties in Germany
Introduction
Given the controversy about Chancellor Merkel's open immigration policies during the peak of the refugee crisis in 2015 and 2016 in Germany, Western nations have more openly than ever responded to existing fears around immigration, favoring nationalist philosophies over more liberal ideologies (Dustmann, Vasiljeva, & Piil Damm, 2019). A critical determinant of the debate is driven by high xenophobia, where individuals become anxious about ethnic minorities and greater diversity, leading to the intolerance of migrants (Weber, 2019).
This has a direct impact on political stability and future electoral voting outcomes. One strain of research has focused on voting outcomes of immigrant inflows, mostly focusing on right-wing parties and the associated rise of voter bases halla2017immigration, arzheimer2019alternative. Arzheimer and Berning (2019), for instance, demonstrate that the primary motivation for voting for the Alternative fur
Deutschland (AfD), a newly established right-wing populist party in Germany, is the voters' negative immigration attitudes. Little research, however, has focused on the influence of voting outcomes for far-left positioned parties in Germany (Barone, D'Ignazio, de Blasio, & Naticchioni, 2016).
Drawing on panel data from six German elections from 1998 to 2017, this study investigates the correlation between far-left positioned parties and immigration per capita using the occurring elections. Our main findings suggest a positive correlation between immigrants per capita and the voting outcomes of the far-left parties.
Literature Review
Impact of Immigration
The risen political debate has polarized attitudes toward asylum seekers and migrants throughout the whole country, resulting in divergent considerations of potential impact (Sola, 2018) characterizes this twofold proliferation as the increasing occurrence of hostile acts against refugees and anti-immigrant demonstrations, especially in Eastern Germany. On the other side, however, and in line with the Chancellor's ‘welcome politics,' there are numerous citizens that actively and voluntarily participate in integration processes to welcome migrants and refugees and to facilitate the transition to German society.
Many citizens perceived new migrants to be more culturally distinct and viewed them as a collective group (Dustmann, Fasani, Frattini, Minale, & Schonberg, 2017).
This phenomenon is in agreement with social identity and self-categorization theory, where humans identify themselves with collectives of people and do so based on commonalities (Weber, 2019). Due to the uprising debate of the increasing immigrants' impact and the existing negative view of society on the one side of the discussion, the existing migration inflow may lead to the categorization of groups based on national or ethnic grounds (Weber, 2019). Such categorization can lead to the belief that immigrants can threaten the nation's identity, resulting in growing national concerns about possible harm and xenophobic perceptions (Sola, 2018).
Unemployment and the fear thereof seems to be a common factor that further raises conservative expectations about immigration on a national scope (Jensen & Mouritsen, 2019). Some research already investigated how unemployment and unemployment fears, for instance, through very high migration, are reflected in left-wing voting outcomes. Algan et al. (2018)found that there is a strong positive relationship between unemployment and voting for non-mainstream parties that entail more populist components. Even concerns about future overall employment insecurity lead to similar political orientations, as more emphasis is put on unemployment benefits and unemployment insurance (Guiso, Herrera, Morelli, & Sonno, 2017).
Immigration and Left-wing Parties
Election surveys and polls point out that the party's view on immigration and associated policies is of importance in election campaigns (Weber, 2019). Ideologically, left-wing parties are in alignment with social diversity and should especially attract groups that sometimes are of disadvantage. Moreover, voting for left-wing parties typically implies believing in social egalitarianism and solidarity (Alonso & Fonseca, 2012). Alonso and de Fonesca (2012) describe left-wing voters as either highly educated citizens or inhabitants with liberal socio-cultural values or the working-class. Since the latter group might be more economically vulnerable, unconditional support of the left's values might not be given (Bansak, Hainmueller, & Hangartner, 2016). Nevertheless, even if some fears exist, acceptance of and support for immigrants is still much higher than for other parties (Bansak et al., 2016).
As Weber (2019) points out, the proximity and the exposure to immigrants shape attitudes toward minority groups, such as immigrants. He found that higher exposure to immigrants, e.g. through more immigrants per capita, leads to more inter-group contact possibilities that have a positive net effect on attitudes toward immigrants.
This is reflected in the inclination of voting toward left-wing parties that favor immigration (Weber, 2019).
We contribute to the existing research by examining the relationship between immigrants and voting outcomes for far-left parties in Germany. By looking at aggregated panel data on the country-wide level, we aim at finding support for the above-outlined literature and the tendency of citizens who live in greater proximity to immigrants to have a more positive attitude towards immigration and, therefore, vote for left parties. This reasoning leads to the following hypothesis:
H: Immigrants per capita increase voting outcomes for far left parties.
Methodology
Data
The dataset used consists of longitudinal data where the same cross-sectional units are followed over time. Since it has a cross-sectional and a time series dimension, panel analysis permits to study the dynamics of change with short time-series and enhances the quality and quantity of data in ways that would be impossible using only one of these two dimensions (Yaffee, 2003).
This data set is an balanced panel because it has no missing values except for real income in 1998 (Yaffee, 2003). It consists of 2430 observations describing the outcomes of the German federal elections of 1998 to 2017 in all German counties. Hence, the years are the temporal reference while the spatial dimension pertains cross-sectional units, including some dichotomous variables as identifiers to assign the observations of counties to federal states. However, most variables are continuous, inter alia, voting outcomes for the different parties, voting participation, and the unemployment rate.
Year dummies were created since it helps to incorporate these nominal variables into the regression analysis and because the distinction between individual counties is not relevant for this context (Daly, Dekker, & Hess, 2016).
We used ‘immig_percap' as our primary independent variable, describing how many asylum seekers there are per capita for a specific county in Germany. The variable has a mean of 0.20145, with a minimum of 0 and a maximum of 2.23783.
Our dependent variable is called ‘hard left' and describes the voting outcomes for the far left, including the greens, for a specific county in Germany. The variable has a mean of 0.16783, with a minimum of 0.03051 and a maximum of 0.45644. We further created an interaction variable, ‘ImUn,' to examine whether there is a significant interaction between unemployment and immigrants per capita.
Additional control variables include the crude unemployment rate, foreigners per capita, population density, share of Abitur, the real income, the share of elderly, and the created dummy variables, that all can influence the voting outcomes for the far left and are thus taken into consideration.
Method
A pooled ordinary least squares model was computed since it provides a first simple interpretation on a modular level (Molnar, 2019). However, according to Wooldridge (2010), pooled OLS is employed when different samples for each period of the panel data are selected, which is not the case in this analysis. Therefore, this analysis should be followed up by either a fixed or a random-effects model.
By including fixed effects (group dummies), one can control for the average differences across counties in any observable or unobservable predictors, which greatly reduced the threat of omitted variable bias (Clark & Linzer, 2015). An important assumption of the fixed effects model is that those characteristics are time-invariant, unique to the individual unit, and not correlated with other individual characteristics (Torres-Reyna, 2007). This model can get expressed by the general equation:
Abbildung in dieser Leseprobe nicht enthalten
In contrast to the fixed-effects model, the random-effects model assumes that the variation across entities is assumed to be random and uncorrelated with the predictor or independent variables included in the model (Torres-Reyna, 2007). Therefore, it enables the researcher to include time-invariant variables. However, its most serious drawback is the problem of bias through partial pooling (Clark & Linzer, 2015).
To assess what test is most appropriate, a Hausmann test was performed, which is designed to detect violation of the random effects modeling assumption that the explanatory variables are orthogonal to the unit effects (Clark & Linzer, 2015). If there is no correlation, estimates of the fixed effects model should be similar to estimates in the random-effects model (Clark & Linzer, 2015). The Hausman test statistic measures the difference between the two estimates, where the null hypothesis is that the preferred model is random-effects (Torres-Reyna, 2007).
Another important aspect is homoscedasticity, which is desirable since it shows homogeneity of variance and, therefore, a well-fitted model. The Breusch-Pagan test was applied to test for heteroskedasticity. It fits a linear regression model to the residuals of a linear regression model and rejects if too much of the variance is explained by the additional explanatory variables (Hothorn et al., 2019). To control for heteroskedasticity, a robust covariance matrix (Sandwich estimator) got estimation using the ‘arellano' estimator, which can be used for both heteroskedasticity and serial correlation and is recommended for fixed effects (Torres-Reyna, 2007).
Lastly, the interaction between immigrants per capita and unemployment was further explored by calculating the tipping of the influence of immigrants per capita from positive to negative based on unemployment.
Biases
Plumper Troger (2019) describe possible biases in fixed-effects models in depth. In general, omitted variable bias effects, the issue of dynamic misspecification, and between unit effects are mentioned. As described by Wooldridge (2010), a fixed-effects model is unbiased only under strict exogeneity (Plumper & Troeger, 2019). Given our data, there may be some bias due to possible exogeneity. Considering our independent variable, the settlement of immigrants as such is controlled for, as their settlement is governed by the national government. However, inter alia, economic policies, attitudes, and conflicts change throughout the years in our data, and these exogenous factors are difficult to control for. Hence, some omitted variable bias exists. This corresponds to the assessment of Plumper and Troger (2019), which highlights the importance of dynamic misspecification and the possible resulting bias. Besides, there could be potential of cross-sectional dependence, which was accounted for through the usage of the Breusch-Pagan test. Lastly, it is important to consider the limitation of this study regarding the contact hypothesis due to the high level of aggregation.
Results
The executed tests illustrate that there is a positive correlation between immigrants per capita and far-left voting outcomes (see Appendix A for R Code). The Hausmann test shows a p-value of < 0.05, which leads to the rejection of the null hypothesis, i.e., it cannot be assumed that the fixed-effects model and the random-effects model yields similar results. Consequently, the alternative hypothesis is accepted, stating that the models are inconsistent. Therefore, a fixed-effects model should be used to get more accurate results.
The results of the first fixed effects model (Appendix B) show highly significant correlations for all variables except for popdens with a significant yet bigger p-value of 0.0024798 and realinc with an insignificant correlation since the p-value > 0.05.
[...]
-
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X. -
Upload your own papers! Earn money and win an iPhone X.