Claremont-UC Undergraduate Research Conference on the Claremont-UC Undergraduate Research Conference on the European Union European Union

In the wake of the 2015 migration crisis, immigration policy has become one of the most critical topics of academic scholarship and political debate. Despite this prolific response, very little research has investigated how the gender of policymakers affects immigration policy. This raises an interesting question: is there any difference in immigration policy among countries with high and low numbers of female legislators? To investigate this matter, I use panel fixed-effects regression to systematically compare the immigration policies of the original EU-15 from 2000 to 2010. As a single policy area, I find female representation has no significant impact on immigration policy. However, by breaking immigration policy into five separate sub-dimensions, I find female representation does have a significant impact on three dimensions—family reunification, asylum/refugee, and enforcement—but not on the other two—labor migration and co-ethnics. In this study, I explore several reasons why this inconsistent influence occurs.


Introduction
In the past decade, the migration crisis has been at the center of every major political debate in Europe.Scholars and policymakers alike have grappled with how to respond to the steady flow of immigrants and refugees.Amidst the turmoil, a vital question emerges: how do women fit into this narrative?Scholars have addressed this question hoping to understand and improve government policies on immigration.However, a glaring void still remains.Most of the current research focuses only on the gender of the immigrants themselves, largely ignoring the gender of the policymakers (Mahler & Pessar, 2006;Nastuta & Tompea, 2011;Pessar & Mahler, 2003).
This is an intriguing deficiency; many scholars have shown that when women participate in the policymaking process, the resulting policies are different (Hunt, 2007;Matthews, 2017).The implication is that, in countries where more women participate in the legislature, the resulting immigration policies may be fundamentally different than in countries with fewer female legislators.Accordingly, my research addresses the following question: What is the relationship between the percentage of women in legislatures and the restrictiveness of immigration policy?
To answer this question and to address the gap in the literature, I use panel data for the original EU-15 from 2000 to 2010 to evaluate the relationship between female representation and the restrictiveness of immigration policy.Contrary to my expectations, my initial analysis reveals no statistically significant relationship between the two.This result is rather surprising based on my theoretical framework.However, when I disaggregate immigration policy into five individual sub-dimensions, I find that, though female representation has no impact at the aggregated level, it does have a significant impact on several of the subdimensions of immigration.In my analysis, I investigate why female representation affects some areas of immigration policy, but not others.

Theoretical Framework
This study lives at the intersection of two of the most important fields of research within politics and international relations: gender and immigration.Much of the current literature on gender asserts that feminine values, such as sympathy and nurturing behaviors, have long been undervalued and underrepresented in society (Matthews, 2017).Because most legislators and policymakers are male, most legislation and policies adhere to traditional male values, such as authority and autonomy (Gilligan, 1993;Noddings, 1984).This male perspective is certainly valid and beneficial.However, the equally valid female perspective has been consistently underrepresented in governments throughout the world.This is why many scholars believe achieving higher female representation in legislatures is so crucial.Women comprise half of the world's population, yet few countries even come close to achieving gender parity in their legislatures.
Greater female representation strongly correlates with numerous measures of good governance, including lower corruption, increased economic competitiveness, and greater political stability (Esarey & Schwindt-Bayer, 2019;Hudson et al., 2012;Hunt, 2007).Joni Lovenduski (2001) asserts that, due to their distinct characteristics and experiences, women provide a unique standpoint and have different policy priorities from the traditional male focus.For example, women often have more experience working in NGOs and nonprofits, which makes them more familiar with social problems and marginalized populations (Chattopadhyay & Duflo, 2004;Hunt, 2007;Matthews, 2017).Additionally, even when https://scholarship.claremont.edu/urceu/vol2019/iss1/5women work in prominent government positions, they are more likely to be appointed to departments and responsibilities that deal with sociocultural matters (Crage et al., 2013).
Because of these unique experiences, women are often more likely to focus on care issues, they tend to have a broader definition of security, and they are generally more ethical and trustworthy (Hunt, 2007;Lovenduski, 2001).Combining this distinct female perspective with the traditional male approach provides a more comprehensive understanding in any policy area, but particularly in areas that are traditionally neglected by men (Matthews, 2017).Because women define security more broadly than men, they often pay more attention to these "low politics" issues like healthcare, education, and the environment (Krook & O'Brien, 2012;Paxton & Hughes, 2010;Reynolds, 1999;Studlar & Moncrief, 1999).
Immigration policy is certainly not considered a low politics issue.Most often, it is included with security issues, which are typically shaped by more masculine values (Crage et al., 2013;Faist, 2004).However, it is actually better classified as both a security and a care issue.A care issue is one that "contributes to the well-being or development of other people" (Dwyer, 2013;England, 1992;England 2005).Thus, Crage and her colleagues classify a policy dealing with border control as a security issue because it involves state safety, but a policy about immigrant integration as a care issue because it involves individual well-being (Crage et al., 2013;Heckmann & Schnapper, 2003).
Because of this duality, male and female opinions about immigration policy often differ (Sides & Citrin, 2007).For example, women are more likely to control prejudice, which influences their attitudes and voting patterns on immigration issues (Harteveld & Ivarsflaten, 2018).One very recent study found that asylum policies are significantly more women-friendly in countries with higher female representation (Emmenegger & Stigwall, 2019).This research provides some initial evidence that women in legislatures do have a discernable impact on immigration policy.However, asylum is only one small aspect of immigration policy, which is complex and multifaceted.The female influence is also likely to affect other characteristics of immigration policy beyond women-friendliness.
This combined scholarship indicates that the gender of policymakers plays a significant role in shaping immigration policy.Women's broader definition of security, their focus on marginalized populations, their distinct policy priorities, and their experience in care issues give them a valuable perspective that shapes their views about immigration policy.Based on this evidence, I present my hypothesis:

As female representation in legislatures increases, the restrictiveness of immigration policies will decrease.
I expect that this will occur because as more women participate in legislatures, there will be increased focus on care issues, including the care aspects of immigration.This increased attention and additional perspective will alter how legislatures approached immigration policy.With a greater focus on marginalized populations, immigration policy will be less restrictive in order to accommodate more immigrants and refugees.

Representation and Restrictiveness Defined
Based on this theoretical framework, I investigate female representation as my key independent variable of interest.For the purpose of this research, this term refers to the Uneven Influence percentage of female legislators in a country's national parliament.The female perspective could reasonably affect immigration policy through other forms of representation, including interest groups, elections, or referendums.However, because my resources are limited, exploring every aspect of this important issue is beyond the scope of this study.I instead focus my research on the proportion of female legislators, rather than other measures of female representation.
I am most interested in this aspect of the female effect on immigration policy because it is not only politically relevant, but also requires a concerted public effort to change.One person can choose to join an interest group or vote in a referendum, but if female legislators are central to shaping immigration policy, change requires cooperation.Though there are many other measures I could use to assess the female influence in policymaking, female representation in legislatures provides the most consistent, quantifiable, and accessible measurement available.Female representation is also particularly relevant in the current literature because it represents an important indicator of women's participation in policymaking (Davidson-Schmich, 2016;Emmenegger & Stigwal, 2019).It offers a reasonable approximation of how well women's voices in a country are heard and translated into policy.Approximating the female influence in this way allows me to address my research question objectively by applying a consistent measurement to a complex phenomenon.
I define female representation as the percentage of female legislators elected to the lower or single house of a country's national legislature.I use only the lower or single house in order to standardize the measurement across countries.Many governments have multiple houses with different electoral rules, which produce variation in representation among the houses.Additionally, in some countries, the upper house is significantly less powerful or influential than the lower house.Thus, creating a combined score or selecting only one of the houses based on its relative merit would misrepresent its effect on policy.Additionally, using only the lower or single house is a common practice many datasets use when calculating female representation (Inter-parliamentary Union, 2019; The World Bank, 2019c).I operationalize female representation by using data from the World Bank to calculate the percentage of female legislators for each country in my study (The World Bank, 2019c).
To define my primary dependent variable, policy restrictiveness, I use data from the Immigration Policies in Comparison (IMPIC) index.This dataset represents the results of a comprehensive study designed to objectively evaluate the restrictiveness of immigration policies across 33 Organisation for Economic Co-operation and Development (OECD) countries from 1980 to 2010 (Helbling et al., 2016).The authors define restrictiveness as the degree to which "a regulation limits or liberalises the rights and freedoms of immigrants" (Helbling et al., 2017).In evaluating restrictiveness, the authors design the study to avoid normative evaluations and instead to create a neutral tool that systematically compares different aspects of immigration policy.
The index evaluates each country on five key dimensions experts agreed were most relevant to immigration policy-family reunification, labor migration, asylum and refugees, co-ethnics, and control.Family reunification policy refers to laws that make it easier for separated family members to obtain legal authorization to cross national borders to join their families.Labor migration involves laws about work visas, employment eligibility, etc. Asylum and refugee policies encompass recognized refugees, asylum seekers, and people with humanitarian protection.Co-ethnic policies involve regulations about migrants who are "entitled to easier access to immigration and settlement in a country because of https://scholarship.claremont.edu/urceu/vol2019/iss1/5 a cultural or historical affinity with the native population" (Bjerre et al., 2016).Control policy incorporates laws that dictate the enforcement of immigration laws, both internally and externally.Though control policies include border control, they also involve other laws that dictate implementation of other policies within a country.For the sake of clarity and precision, I therefore refer to control policies as enforcement policies.
The authors of the IMPIC study select several specific measures to assess each of these five dimensions and then interview experts on each country and policy area.They closely follow conventional procedures in the literature in order to verify that their criteria are valid.Using this method, they create an extensive index rating immigration policy restrictiveness across countries and within the five policy dimensions.They use a continuous scale from zero to one to measure restrictiveness, with higher numbers indicating more restrictive policies.

Control Variables
Besides female representation and policy restrictiveness, there are many other factors that could affect immigration policy.Generally, the two main influences on immigration policy are economic and ideological issues (Givens & Luedtke, 2005;Milner & Tingley, 2011).In order to account for these other influences, I incorporate several control variables into my study.To control for the economic factors, I use data from the World Bank about each country's yearly Gross Domestic Product (GDP) per capita, unemployment rate, and growth rate (The World Bank, 2018; The World Bank, 2019a; The World Bank, 2019b).These are important because if a country is struggling economically, its citizens are more likely to oppose immigration out of fear that immigrants will threaten their jobs or consume their resources.
In addition to economic factors, I also add several variables to account for ideological concerns.For example, countries that receive more immigrants could oppose immigration more than others because they have to bear heavier costs.To control for this, I include each country's yearly immigrant flows using data from the OECD's International Migration Database (OECD, 2019).I also expect countries that experience more terrorist activity to be more inclined to limit immigration out of fear, so I include data from the Global Terrorism Database about each country's yearly terrorist attacks as well (Global Terrorism Database, 2018).Finally, partisanship can also play a major role in influencing immigration policy (Givens & Luedtke, 2005;Money, 1999).In order to control for this, I include a variable that captures the political strength of the left by calculating the percentage of parliamentary seats held by parties on the left compared to the right.I obtain this data from the Parliaments and Governments Database using their elections dataset (ParlGov, 2018).These economic and ideological control variables allow me to mitigate the effect of omitted variable bias in my analysis.
Though I plan to carefully control for the most important confounding variables, I acknowledge that all research has constraints.Due to the limited scope and resources of this study, I cannot thoroughly investigate every possible variable that could affect immigration policy.For example, I would like to include a variable about public opinion on immigration, but during the years my study covers, no consistent measures exist.The Eurobarometer, European Social Survey, and other common sources of public opinion data began consistently including immigration questions only recently.Before they did, public opinion data on immigration was sparse and inconsistent.Trying to measure it would involve creat-Uneven Influence ing some kind of index based on multiple sources and inconsistent questions that exceeds the scope of this paper.However, by using established statistical measures and carefully planning my research design, I do address the most common factors discussed in the literature, as well as those with major theoretical importance.

Empirical Analysis
Though I would prefer to analyze every country in the IMPIC database, including my additional variables means I have to limit my study in order to maintain feasibility.Consequently, I analyze only the EU-15: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom.These countries are all well-established democracies, and they provide a diverse sampling of female representation, with Sweden being the highest in the EU and Greece being the lowest.For the sake of manageability, I also modify the index to include only its last decade of data.I extend the IMPIC database by adding female representation and the control variables discussed previously: GDP per capita, unemployment rate, growth rate, immigrant flows, terrorist attacks, and political strength of the left).
To analyze my dataset, I use panel fixed-effects regression, clustered by country, to evaluate the relationship between representation and restrictiveness, controlling for my additional variables.I use fixed-effects regression in order to mitigate autocorrelation error in my analysis.Because my data involves multiple countries over multiple years, a simple Ordinary Least Squares (OLS) regression would overestimate the relationships between restrictiveness and representation because each country's values would be highly correlated with their same values from the previous year.This would bias the relationship upward by making it appear stronger than it really is.Instead, using a fixed-effects regression allows me to automatically correct for correlation between each country's values.

Aggregated Immigration Policy Model
The results of my initial regression appear as Model 1 in Table 1.Though I include the most theoretically compelling variables in the literature, none has a significant impact on immigration policy restrictiveness in my analysis.Based on this surprising result, in Model 2, I investigate whether there are any interactions or nonlinear relationships among my variables that have conceptual significance.For example, having a high GDP with a slowing growth rate would likely affect a country's attitudes about immigration policy differently than having a low GDP with an accelerating growth rate.I account for these effects by including interactions between the three economic variables in addition to the other control variables.I have tested each interaction before adding it to the regression and find that all three improved the model's adjusted R-squared both individually and jointly.Based on this evidence and the theoretical justification, I am confident that the interactions improve the fit of my model.I have also tested several other interactive variables and non-linear relationships, but none are theoretically or substantively significant in the regressions.
https://scholarship.claremont.edu/urceu/vol2019/iss1/5Surprisingly, both models indicate that female representation in legislatures has no impact on immigration policy restrictiveness.This contradicts my hypothesis that fe-Uneven Influence male representation would significantly reduce policy restrictiveness.This unexpected result likely occurs because the regression only evaluates the relationship between female representation and the restrictiveness of immigration policy as a whole.However, due to the dual nature of immigration policy as both a security and a care issue, it is possible that women's greater focus on care issues has a greater impact on the care aspects of immigration.Lumping all five aspects into a single measure of policy restrictiveness likely obscures women's actual effect.
Based on this expectation, I analyze each of the five policy dimensions individually.In Table 2, I include five more fixed-effects regressions, replacing overall immigration policy restrictiveness as the dependent variable with the restrictiveness of the individual policy dimensions-family reunification, labor migration, asylum and refugees, co-ethnics, and enforcement.Though many of the control variables were insignificant in my initial regression, I still include them in the subsequent regressions in order to evaluate whether they affect individual policy dimensions differently.The results of these regressions, which appear in Table 2 on the following page, indicate that female representation does influence certain aspects of immigration policy, although clearly not others.

Disaggregated Immigration Policy Model
As the Table 2 demonstrates, most of the disaggregated models had higher adjusted R-squared values than the initial model, which indicates that breaking immigration policy into its individual dimensions offers a better fit for the data.In interpreting this data, I mostly focus my analysis on the direction and significance of each variable.Because restrictiveness is measured from zero to one as less restrictive and more restrictive, a quantitative interpretation of the relationship has little real world significance.For example, it is not very meaningful or helpful to say that as GDP per capita increases by one US dollar, enforcement policy restrictiveness increases by .124points.In contrast, the direction and significance of the relationships are extremely instructive, because they indicate whether female representation makes policies significantly more or less restrictive.I therefore focus my analysis on those aspects, rather than the numerical values.
The disaggregated regression indicates that female representation has a significant negative relationship with the restrictiveness of family reunification policies in Europe.As female representation in legislatures increases, family reunification policies become significantly less restrictive.The above-mentioned research about care issues in immigration policy reveals why this would be the case.Family reunification is more concerned with individual and family well-being than with state well-being, so it exhibits more characteristics of a care issue than a security issue.It is therefore unsurprising that the female effect would emerge in this area.
https://scholarship.claremont.edu/urceu/vol2019/iss1/5Besides female representation, only the interactive economic variables had a statistically significant effect on the restrictiveness of family reunification policy.For example, no matter how many immigrants enter a country, there is no significant effect on the restrictiveness of family reunification policy.Likewise, political strength of the left, terrorist attacks, and the individual economic variables have no effect on the restrictiveness of family reunification policies.This result is both instructive and encouraging, because while citizens and policymakers cannot truly control immigrant flows, terrorist attacks, the political strength of the left, or economic indicators, they can control female representation in legislatures.Though less restrictive may not necessarily be better than more restrictive, this evidence does indicate that immigration policies do not respond only to forces outside a country's control.
In contrast to family reunification policy, female representation has no statistically significant impact on the restrictiveness of labor migration policies.This result appears consistent with the literature about women's focus on care issues mentioned above.Though labor migration does offer some benefits to individuals, politicians generally advocate for it because it brings economic benefits to the state, not to the individuals.Thus, labor migration is not typically considered a care issue, so the insignificant effect of female representation is unsurprising.Instead, the political strength of the left and the interactive economic variables are the dominant influences.The effect of the economic variables needs little theoretical explanation.Labor migration laws directly affect the economic interests of native populations as immigrants compete for jobs and resources.If a country experiences economic difficulty, this would likely foster greater anti-immigrant sentiment, which could lead to increased restrictiveness in labor migration policies.
The effect of the political strength of the left is somewhat more intriguing.The results of the regression indicate that as the political strength of the left increases in a country, labor migration policies become less restrictive.This could be because left-wing parties are potentially more open to immigration in general.This relationship could also indicate that when left-wing parties gain more political power, right wing anti-immigrant populist parties have less influence over immigration policies (Ivarsflaten 2008, Rydgren 2008).
Of all the dimensions of immigration policy, female representation has the most significant effect on asylum and refugee policies.As female representation increases, the restrictiveness of asylum and refugee policies decreases significantly.This result is rather unsurprising.Asylum and refugee policy explicitly aims to improve the well-being of individuals, so it strongly exhibits the characteristics of a care issue.Interestingly, this result supports the results of recent research.Emmenegger and Stigwall (2019) found that countries with higher female representation have more women-friendly asylum policies (Emmenegger & Stigwall, 2019).Using completely separate datasets and significantly different methods, we both found statistically significant evidence that female representation in legislatures affects asylum and refugee policy.
It is also interesting that female representation is the only variable in the regression https://scholarship.claremont.edu/urceu/vol2019/iss1/5 that had any significant effect on asylum and refugee policy.None of the other variables that the current literature typically highlights had any impact, including economic concerns, partisanship, terrorist attacks, or immigrant flows.If no other factors matter, this evidence indicates a serious need to evaluate how female representation shapes asylum and refugee policy.Is women's effect on asylum and refugee policies positive or negative for the individual countries?Is it positive or negative for the refugees?These questions highlight the need for further research on this subject.
In contrast with asylum and refugee policy, female representation has no significant impact on the restrictiveness of co-ethnic policies.It is unsurprising that female representation had no significant impact in this area because it does not appear to be a care issue that specifically or directly promotes individual well-being.In fact, none of the variables included in the regression had a significant impact on co-ethnic policies.This indicates that there are some omitted variables, such as a history of colonialism in a country, that could uniquely affect the restrictiveness of co-ethnic policies.Future research about co-ethnic policies could identify what these are.
The final dimension of immigration policy, enforcement, is more perplexing than the other dimensions.Female representation has a significant positive relationship with the restrictiveness of enforcement policy.This is puzzling for two reasons.First, my theoretical framework indicates that the influence of female representation is strongest for care issues.However, enforcement policy arguably contributes more to state well-being than to individual well-being.It does not, therefore, appear to be a care issue, yet its relationship with female representation is statistically significant.Second, in contrast to family reunification policies and asylum and refugee policies, the relationship between female representation and enforcement policy is positive, not negative.This means that as female representation increases, enforcement policy restrictiveness actually increases.Future qualitative research could investigate why this occurs, but one possible explanation is that women are willing to help immigrants that already reside within their country, but they fear letting in more immigrants because of the problems associated with immigration.

Conclusions and Future Research
By disaggregating immigration policy into its separate dimensions, I was able to uncover relationships that were obscured at the aggregated level.I concluded that female representation has a significant impact on the restrictiveness of only certain dimensions of immigration policy, particularly those that are generally considered care issues.This is a crucial finding in a world that desperately needs to evaluate its immigration policies.Though policy restrictiveness does not necessarily correlate with effectiveness, my analysis indicates that women's perspective makes a difference.This study does not make any normative claims about whether that difference is positive or negative-whether more restrictive policy is better or worse than less restrictive policy.Future research will need to evaluate whether female policies are more or less effective, compassionate, beneficial, etc.However, if female representation does make a perceptible difference in shaping immigration policy, policymakers concerned with the effectiveness of their policies need to consider how the female influence is affecting those policies.
Though the results of this research were interesting and instructive, I have only begun to examine the relationship between female representation and immigration policy restrictiveness.My research involved primarily large-n, quantitative analysis.However, to Uneven Influence further establish the causal mechanisms at work and to evaluate the relative merit of the female perspective on immigration policy, future research will need to examine additional qualitative evidence that offers insight about the exact causal mechanisms that make female representation matter.Such evidence could include parliamentary records, news sources, political speeches, and other primary sources.
One limitation I faced in this study was that the IMPIC database only includes records through the year 2010.Though the causal mechanisms likely remain consistent across time, recent events, most notably the 2015 immigration crisis, might alter the precise relationship between female representation and immigration policy restrictiveness.In one scenario, the rapid increase of refugees could cause the female perspective to become even more relevant, potentially having a greater effect in some of the other policy dimensions that are not typically care issues.Alternatively, it is also possible that the female perspective would become less relevant because increased immigrant flows would cause more securityrelated problems at home.Future research with an extended dataset could better examine how this relationship between immigration policy restrictiveness and female representation was affected after the 2015 immigration crisis.
Another constraint I experienced was that I had to maintain feasibility by limiting the number of countries I analyzed.Because I was adding six additional variables for each country per year, I only had the resources to evaluate fifteen countries.In the future, I would like to look at other countries in the EU as well as countries outside the EU to confirm how my theoretical framework applies in other immigration settings.I am especially interested in how female representation would affect the restrictiveness of immigration policies in the United States.
A final limitation I faced was that I only had access to observational data.Because randomly assigning female representation ratios or immigration policies to the countries in Europe is not possible, I could not manipulate reality in order to establish causality.I acknowledge that the same social movements and forces that produce increased female representation in legislatures could also prompt changes in immigration policy.I controlled for partisanship in order to limit one major source of this distortion, but others likely exist.However, the related literature in the field supports my causal argument that having women in the legislature affects immigration policy (Crage et al., 2013;Emmenegger & Stigwall, 2019).Beyond the correlation vs. causation problem, any observational research design must also address the possibility of reverse causality.In the case of this research, it seems extremely unlikely that the restrictiveness of immigration policy changes female representation, unless perhaps women grow frustrated with male immigration policies.However, there is little real world evidence that this kind of causality actually occurs, so I maintain my original causal sequence.
The disappointing gap in the literature on gender and immigration indicates that much research still remains.However, my analysis has certainly added to the current literature on gender and immigration and has provided insight into what areas need more investigation.The results of this study will be immensely valuable for policymakers as they seek to combat the fractionalization and hostility that threaten the liberal consensus of post-war Europe.I have demonstrated that female representation is negatively correlated with the restrictiveness of certain dimensions of immigration policy.
In light of this evidence, scholars and politicians need to do more to understand women's effect on immigration policy and the implications for their respective countries.
https://scholarship.claremont.edu/urceu/vol2019/iss1/5Female participation is certainly not the only factor affecting immigration policies, but its influence is more significant than the current literature suggests.The task that remains is determining whether that influence is helping or hurting the countries of Western Europe.Women's distinct perspective could be the key to easing the immigration crisis and restoring stability and harmony to Europe.

Table 1 .
Dependent Variable: Immigration Policy Restrictiveness country level to allow for serial correlation in the error within a state.Coefficients are individually statistically significant at the *10% **5%, ***1% significance level.Dependent variables are measured on a scale of 0-1, with higher numbers being more restrictive.GDP per capita is calculated as the natural logarithm of GDP per capita to account for distortion from large values.

Table 2
Notes:Standard errors appear in parentheses beneath coefficients and are heteroskedasticity-robust and clustered at the country level to allow for serial correlation in the error within a state.Coefficients are individually statistically significant at the *10% **5%, ***1% significance level.Dependent variables are measured on a scale of 0-1, with higher numbers being more restrictive.GDP per capita is calculated as the natural logarithm of GDP per capita to account for distortion from large values.