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

Immigration’s effect on European welfare states is complicated. On one hand, increased immigration might undermine social solidarity and impose greater fiscal burdens on redistribution, reducing support for welfare spending. On the other, natives could respond to greater globalization with economic anxiety, increasing support for redistribution in order to mitigate risk. Welfare chauvinism predicts a mixed effect—increased spending for programs that middle-class natives use and reduced spending for programs that benefit immigrants disproportionately. I test this theory by analyzing (1) European attitudes towards immigration and welfare spending and (2) actual spending on these programs, particularly social housing. Additionally, I present a brief case study of France’s immigration/welfare relationship. Despite large increases in immigration, I find no significant increase in welfare chauvinistic attitudes and no systematic relationship between immigration and social spending. This surprising result—which contradicts recent empirical findings—suggests that immigration-based fears about Western European welfare states are overstated.


Introduction
Europe's populist resurgence suggests that the continent's appetite for globalization may be satiated.Immigration especially is testing European tolerance.Prominent academic voices claim that immigration has spurred the retreat of liberalism (Zielonka, 2018) and may even cause the "strange death of Europe" (Murray, 2018).They and others are especially worried about immigration-induced changes to welfare state attitudes and behaviors.Several theories support their concerns.Particularly, the welfare chauvinism hypothesis predicts increased spending for programs that many natives use and retrenchment for programs that benefit immigrants disproportionately.I build off previous research that supports the chauvinism hypothesis by extending the analysis to social housing spending.Surprisingly, I find that robust econometric specifications produce no significant causal relationship between immigration and social housing spending.In fact, I find no systematic relationship between immigration and any social spending program.This is puzzling given the correlation between recent surges in immigration and the resurgence of populism.These findings contradict apocalyptic voices who claim that immigration is an existential threat to European welfare states.
On the contrary, immigration may be an important part of their survival.The welfare state in Europe was built during a period of robust population growth but before the rapid increase in life expectancy.It was easy to finance these new programs when the workforce was growing, but since Europe's working-age population is now shrinking and will continue to do so for at least three decades, the welfare state is under increasing financial pressure (Vartiainen, 2018).Immigrants fill job vacancies and contribute to social insurance for the elderly.Even so, recent surges in immigration seem to be testing the limits of this potential solution.Understanding the causal relationship between immigration, populism, and welfare state spending is essential in order to sustain the viability of social protection schemes.

Theoretical Expectations
No collection of theories gives clear predictions about the impact of immigration on welfare state retrenchment, at least in part because the causal pathway linkages are extended and sensitive to many other factors (see Figure 1).A simplified causal chain begins with immigration flows affecting natives' perception of the economic and cultural security.These perceptions then influence political orientation and election outcomes.Finally, these governmental configurations should eventually translate into policy, although policy lags, noisy feedback loops, or opposition party resistance could distort legislation and implementation.I begin with a theoretical discussion on how immigration could influence native attitudes about welfare, then describe the theories predicting various spending reactions.From an economic standpoint, most citizens are better off because of migration.Current levels of immigration generally improve the economies of host countries and benefit most of their citizens (Collier, 2013).Several lines of research predict that opening migration flows further could bring even more economic benefits, potentially trillions of dollars of increased output (Clemens, 2011).Immigrants are usually much more productive in rich countries.Capital owners benefit from more labor to operate their machines.Even so, immigration creates losers.Standard economic modeling predicts that immigration of low-skilled workers puts downward pressure on native low-skilled workers' wages and can increase unemployment.Non-economic factors matter, too.The dilution of the host country's culture, acts of violence committed by immigrants, and problematic integration can all influence natives' perceptions and engender anti-immigration sentiment.
Indeed, recent levels of immigration seem to have rattled European natives.The positive correlation between immigration and right-wing populism voting has been tested and demonstrated in a variety of national and political contexts (Facchini & Mayda, 2009).Evidence from France, however, suggests that the association is not as simple as it may seem (Della Posta, 2013).At the department (regional) level, large immigration does indeed lead to higher vote share for the far-right National Front party.At the commune (town) level, however, large immigrant populations are associated with lower National Front votes.Intergroup contact potentially mitigates native anxiety as more frequent social contact with immigrants lessens social distance and alleviates tensions that lead to anti-immigrant sentiment (Della Posta, 2013).The overall ambiguity caused by these opposing pressures provides the basis for my null hypothesis that increased immigration does not systematically worsen natives' beliefs about immigrants or the welfare state.
Other theories, however, provide different predictions regarding natives' perceptions of immigration and suggest competing hypotheses for welfare support.Like Della Posta (2013), Burgoon et. al. (2012) examines how various frames of reference influence perceptions of immigration by comparing the nationwide immigrant stock to the number of immigrants in a native respondent's occupation.He finds that national-level immigration affects populist voting only marginally, if at all.However, workers in industries that have relatively high concentrations of immigrants are much more likely to vote for an extreme right party.If natives are uncomfortable with the current level of immigration, they could respond in several ways.The solidarity hypothesis suggests that higher levels of ethnicallyheterogeneous immigration increase the fiscal costs of redistribution and diminish social solidarity, therefore reducing welfare support (Burgoon, Koster, & van Egmond, 2012).On the other hand, the anxiety hypothesis suggests that, especially for low-skilled native workers, increasing levels of immigration might heighten economic anxiety and reinforce support for redistribution in order to mitigate risk (Burgoon, 2001).With both of these theories, we might observe that increased immigration lowers natives' willingness to fund welfare programs for immigrants (Vadlamannati & Kelly, 2017).

H1: Increased immigration lowers natives' willingness to fund welfare programs for immigrants.
Recent work by Dennis Spies (Schmidt-Catran & Spies, 2016;Spies, 2018;Spies & Schmidt-Catran, 2016) synthesizes these theories into the welfare chauvinism hypothesis and links them with actual spending behavior using econometric analysis.He suggests that social protection schemes are too complex to garner either categorical support or rejection Europe's (Lack of) Welfare Chauvinism in the wake of immigration.Spies considers three particular protection schemes: old-age insurance, health insurance, and unemployment insurance.He finds that immigration has a positive correlation with old-age insurance and health insurance spending, but a negative correlation with unemployment insurance spending.His theory is that many workers in the middle class either benefit from or anticipate benefitting from the first two programs, and immigrant participation in them is relatively low.Unemployment insurance, on the other hand, disproportionately benefits immigrants.

H2: More immigration causes increased spending on social programs that most natives use or anticipate needing (e.g., old age and health insurance) but decreased spending on programs they perceive to benefit immigrants disproportionately (e.g., unemployment insurance).
I further evaluate this theory by examining another realm of social protection: public housing assistance.Spies (2018) suggests this area for further study since public housing assistance also disproportionately benefits immigrants (Fougère, Kramarz, Rathelot, & Safi, 2013;Scanlon, Fernández Arrigoitia, & Whitehead, 2015, 3).In France, 15 percent of French natives lived in public housing in 1999, while the participation rate for immigrants was closer to 50 percent (Scanlon et al., 2015).Further, while unemployment insurance is funded at least in part through worker contributions, public housing assistance is financed completely through taxes.These two facts indicate that this more redistributive program may be particularly vulnerable to retrenchment or welfare chauvinism with high rates of immigration.An example from Austria is useful.Before 2003, non-European Union (EU) nationals were ineligible for access to public housing, regardless of length of stay.A 2003 EU directive mandated member states to permit individuals with foreign residence permits to access social services and transfers.In Austrian cities with large stocks of immigrants, this mandate fueled the belief that natives were competing with immigrants for access to a limited public housing stock, and populist parties won greater shares of votes in those areas (Cavaille & Ferwerda, 2016).Since it is legally difficult to exclude immigrants from welfare transfers altogether, societies with welfare chauvinistic attitudes could instead reduce total funding for specific types of programs that (they perceive to) benefit immigrants disproportionately.In line with H2, I expect public housing expenditures as a percent of gross domestic product (GDP) to eventually decrease as a result of higher immigration.
Certain factors complicate the analysis by distorting the relationship between immigration, native attitudes, and actual spending.First, since immigrants are typically poor relative to citizens of the host country (Burgoon, 2014), an increase in immigration should eventually lead to an increase in social expenditure.This is true if immigrants become eligible for social assistance in the years after their arrival and if the political status quo remains the same.Thus, even if natives succeed in legislating lower welfare allowances per head, the number of heads increases because of immigration.Second, there are likely policy lags between native desires to curb spending and legislated retrenchment.Because different countries may have different lags, uncovering a systematic relationship might be difficult.Finally, even if natives elect politicians who actually want to reduce spending, political and practical opposition might be significant.Natives also use social services that benefit immigrants, and they might be successful in blocking attempts to curb spending.https://scholarship.claremont.edu/urceu/vol2019/iss1/3

Methodology and Data
The first step in investigating this black box of a causal chain is to see what comes out at the end.To test the effect of immigration on actual spending, I extend Spies' model using public housing data: where W is the welfare position of the government and X is a vector of control variables (Spies 2018, p. 130).I use data from 1980 to 2015 and include thirteen Western European countries in the analysis (Austria, Belgium, Denmark, Finland, France, Germany, Italy, Ireland, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom).
The necessary data are available from multiple sources.I utilize the Organisation for Economic Co-operation and Development's (OECD) Social Protection Database, which tracks annual expenditure by social protection program category for European countries (OECD, 2018).The OECD International Migration Database tracks the foreign-born population in each country of interest, including refugees (OECD, 2017).The welfare position of government is proxied using information from the Comparative Manifesto Project (CMP), which "isolates quasi-sentences in a party's manifesto and pairs them with fifty-six predefined policy categories" to estimate the priority of each party based on the percentages in each category (Volkens et al., 2018).This is an important control variable because the welfare position of a country influences its spending on social programs, but is also likely correlated with the number of immigrants it accepts.The vector of other control variables includes unemployment, GDP growth, and public debt (OECD, World Bank).
My identification strategy is a fixed effects model with country dummies and cluster-corrected standard errors.The dependent variable is the percent change in spending as a percent of GDP from the previous year.These specification choices aim to reduce potential problems with a panel data set: endogeneity and autocorrelation.The main measure for immigration is the percent change in the inflow of foreigners as a percent of the population.The primary variables are summarized below: Europe's (Lack of) Welfare Chauvinism Although I do use foreign stock as a percent of the population in one model, I argue that primarily using inflows is a better way to gauge natives' perceptions of contemporary immigration, since areas with large stocks might be more 'used' to them-the same flow in country A with a large stock and in country B with a small stock might increase native anxiety more in country B since it is not accustomed to having a large immigrant community.Using percent changes rather than levels mitigates autocorrelation problems that arise from standard panel data models.This choice departs from Spies' methodology, which uses only levels, since models with percent changes are more "harsh."A broader discussion of this divergence is included in the empirical results section.
After understanding how immigration is correlated with welfare spending, I investigate the first part of the causal chain: immigration's impact on attitudes towards welfare.I report several findings from the European Social Survey, a broad and comprehensive survey administered in Europe every two years.Round 4 of this survey included a module measuring welfare attitudes in 23 European countries in 2008 and featured 56,752 respondents.The same module was conducted again in Round 8 in 2016 with 44,387 respondents.The survey includes dozens of questions on a variety of topics, and two particular questions are useful in this paper's analysis.The first gauges respondents' views about the timeline for immigrant eligibility of social benefits.Europeans were asked at what point immigrants should obtain the "same rights to social benefits and services as citizens already living here" (European Social Survey ERIC, 2009& 2017).The second asks if the current level of social spending is damaging to the economy.Since the period between 2008 and 2016 experienced mass immigration, significant changes in these answers would suggest that immigration indeed changes natives' attitudes towards welfare.

Spending
Obviously, immigration in Europe has increased significantly in recent decades.In 1989, foreigners accounted for an average of 3.5 percent of the population for the 13 Western European countries previously identified.By 2014, that number swelled to 13.4 percent.Immigration inflows have ballooned as well, from an average of 0.48 percent of the population in 1984 to 1.2 percent of the population in 2015.Figure 3 shows the correlation between immigration and housing expenditure specifically, and yields a puzzling result.On average, there appears to be a roughly inverse relationship between the average flow of immigrants as a percent of the population and the percent of GDP spent on housing benefits (consistent with H2).This trend does not hold for every country in the sample, but it is present for many of the countries (see the appendix for a selection of country-specific graphs).Alternatively-and not so obviously-the relationship may be describing the lag between immigrants' arrival to the host nation and their eligibility to receive social benefits, which varies by country.With this lag, the trends could in fact be parallel.This would support H3 that immigration increases social spending with an eligibility lag.

Figure 3. Immigration and Housing Spending (sample average)
To truly test the hypotheses, it is important to examine the regression results, not just the descriptive correlations.Table 1 lays out the results of various regression specifications.The first seven models evaluate the relationship between immigration and social housing expenditures, while the final three examine expenditures on old age, housing, and unemployment insurance.The first housing model's main independent variable is the inflow of foreigners as a percentage of population.Consistent with H2 (immigration leading to lower welfare spending), the coefficient's sign is negative and statistically significant at the 0.01 level.The coefficient's magnitude is tricky to interpret since the main independent and dependent variables are both percentages.For Model (1), an increase in the inflow of immigrants from one percent to two percent of the population (roughly double the sample average) is associated with a seven percentage point decline in social housing spending's rate of change.For example, if other factors would normally cause a 10 percent increase in social Europe's (Lack of) Welfare Chauvinism housing expenditures, a one percentage point increase in migrant inflows (as a percent of the population) would bring the increase in social housing expenditures down to roughly three percent.This magnitude is small but not negligible.The next four models include lags of foreign immigrant inflow at one year, three years, and five years respectively.The lag at one year is still negative and barely significant at the 0.10 level.The lag at three years is actually positive, but not statistically significant.The lag at five years is negative again, but also not statistically significant.

Table 1. Regression Estimates
Model (5) includes both the current level of immigration and the three-year lag.Interestingly, the sign of the three-year lag is positive while the sign of the contemporary variable is negative.This might suggest that it takes several years for immigrants to become eligible to receive social housing benefits, so a surge in immigration would not translate into an increase in spending until later (H3).Counteracting this upward pressure could be mass contemporary immigration that makes natives uneasy, therefore applying downward pressure to spending immediately.Since neither coefficient is statistically significant, the data do not give robust support to this theory.Models ( 6) and ( 7) use different measures of immigration.All previous models examined the impact of yearly immigration flows as a percent of the population.Model (6) instead uses the stock of foreigners as a percent of the population.This is important if relatively small flows of immigration to a small base of immigrants is perceived as unproblematic or even beneficial.On the other hand, even with small annual flows, a large immigrant stock could unsettle anxious voters who observe overcrowded schools, hospitals, and neighborhoods.The coefficient has a negative sign but is small and not statistically significant.Model (7) measures the annual flow of migrants as a percentage of the total stock.This variation is insightful because it essentially interacts flows with stocks.A flow that is small compared to the native population may be large relative to the existing stock.As mentioned earlier, regions with little exposure to migrants are more anxious about migrant inflows than areas https://scholarship.claremont.edu/urceu/vol2019/iss1/3 with larger stocks.This coefficient is positive, but small and not statistically significant.
Finally, models ( 8)-( 10) report the standard specification for old age, health, and unemployment insurance, respectively.None are statistically significant.These results are admittedly troubling.There is no real theoretical framework to suggest that immigration should have a strongly significant effect on housing but not on other programs.This suggests the possibility that the results from model ( 1) are sensitive to a specification that just happened to be significant.There are several reasons why this could be so.
The research upon which this analysis is based found statistically significant effects of immigration on old age, health, and unemployment insurance, but no significant effect on overall social spending (largely because, as H2 predicts, immigration may have diverging effects on various programs).My model specifications are different.First, I use slightly different data.Though the countries in the sample are the same, mine spans through 2015 while Spies' data stops in 2007.Dramatic increases in immigration have occurred since then, so a less than equivalently dramatic change in welfare spending would attenuate the estimates.Further, compared to Spies' approach, my model is more "harsh."His approach does not utilize a fixed effects model that removes large parts of variation attributed to country-specific unobservable characteristics.Finally, instead of using changes in spending, his approach uses yearly levels of spending as a percent of GDP and does not include lags of the dependent variable.These choices expose the specification to autocorrelation problems and do not remove variation that might be caused by country-specific effects.Perhaps this is why my harsher model fails to find significant effects.It may be that there really is an effect between immigration and social spending, but not one that is significant enough to survive robust standard errors-when I use Spies's specification with my data, I find statistically significant results consistent with his findings.
Further, the possible existence of country-specific lags is problematic.Different countries likely have different eligibility and political process lags.Because regressions require making a fixed set of choices for all countries in the sample, different lags could attenuate my results.Finally, the measure of welfare spending is not perfect.Since it is a ratio of actual spending to GDP, it essentially measures two things.Steady social spending but increasing GDP would cause the ratio to decrease even though there are no substantive changes in welfare institutions.
Notwithstanding specification limitations, there are certainly theoretical reasons why a systematic relationship does not exist between immigration and social spending.It might be that, despite opinion surveys revealing native anxiety about immigration generally, natives have not (yet) significantly changed their voting patterns or even their attitudes towards immigrants' access to the welfare state.Populist parties have gained footing in Europe in recent years, but they have not yet reached parliamentary or executive majorities (at least in this sample).Further, even if many natives wish to curb benefits, it may be politically difficult to do so since many natives still rely on programs that disproportionately benefit migrants (H4).To investigate the plausibility of these hypotheses, I analyze immigration's impact on attitudes towards social protection programs-the first part of the causal chain.

Attitudes
I report two key questions from the European Social Survey (ESS) in 2008 and 2016, the two years when the welfare module was included in the survey.The first is a question that asks when the respondent believes an immigrant should obtain the same rights Europe's (Lack of) Welfare Chauvinism to social services as citizens: immediately, after a year (whether or not he/she works), after at least a year of working and paying taxes, after acquiring citizenship, or never.Secondly, I report on the question: "Do social benefits place too great a strain on the economy?"For clarity, Figure 4 aggregates the results across the sample of thirteen countries and contrasts them between 2008 and 2016, a period of mass immigration1 .
In the aggregate, perceptions of the majority do not change significantly.The most noteworthy change is the percentage of respondents who believe that immigrants should never get the same rights to social services as natives, from 4.17 percent to 6.86 percent.This 64 percent increase provides some evidence that increased immigration has indeed hardened attitudes for a part of the population.Even so, purely chauvinistic attitudes are still the minority, with most citizens (65 percent) believing that immigrants should gain access to social services immediately or after working and paying taxes for a year.Similarly, there is little change in the proportion of respondents who believe that social benefits place too great a strain on the economy.Feelings on this matter have somewhat polarized-the percentages of people who strongly agree or strongly disagree each increased by a little less than one percentage point.These results are robust even when excluding all non-citizen respondents from the sample 2 .

Figure 4. ESS Data
These results are surprising.Though only two questions on a survey, these results support the null hypothesis and suggest that, perhaps contrary to popular belief (Eichengreen, 2018;Murray, 2018;Zielonka, 2018), large-scale migration has not brought dramatic changes to Europeans' views about the necessary scope of social protection schemes or about immigrants' eligibility.Besides the slight increase in welfare chauvinism among a very small portion of the population, these results are not consistent with H1 that immigration undermines social solidarity and support for redistribution.Perhaps this is because the survey was administered in 2016, not 2015.The year after the height of the refugee crisis, natives breathed a sigh of relief when lower flows showed that 2015-level mass immigration was not the new normal.Further, although the European Social Survey is a carefully administered and widely-respected survey, sampling error and other data problems might soften a harsher reaction.Nevertheless, given these findings, we might also anticipate finding little, if any, effect in actual social spending.

The French Case
The results presented thus far characterize the average effect of immigration on various social spending categories for the entire sample of 13 Western European countries.While large-n studies bring many advantages, they also lack the context and specificity of individual case studies and may mask significantly heterogeneous effects.Analyzing each country in the sample individually is beyond the scope of any article, but a brief sketch of the French case adds a level of nuance to my analysis.In short, the relationship between immigration and social housing support in France is similar to the sample average effects.
French social housing has been a fundamental part of its welfare state for decades.After the decimation of hundreds of thousands of buildings during World War II, millions of French people lived in squalor.The death of a mother on the street in Paris in 1954 sparked public outcry.Soon after, construction of HLMs (habitations à loyer modéré) skyrocketed and continued for decades.According to Article 55 of the 1998 "Loi relative à la solidarité et au renouvelement urbains," French towns with more than 3,500 inhabitants must designate at least 20 percent of their housing stock as social housing.Current president Emmanuel Macron has proposed 1.7 billion euros of social housing benefit cuts as part of an effort to balance France's finances.Tenants are not affected, so the hundreds of housing associations throughout the country that will be vulnerable to the cuts have taken to the streets in protest.Social housing continues to be a contentious political issue in France today.
France-specific results for the European Social Survey question on immigration and welfare benefits are presented in Figure 5.They roughly match the sample average change from 2008 to 2016.The proportion of pure welfare chauvinists-those who believe migrants should never get the same rights to social benefits as natives-is slightly higher in France with 4.90 percent in 2008 and 7.72 in 2016 (compared to the sample average of 4.17 percent and 6.86 percent).Even so, the French are more likely to believe that migrants should get access after working and paying taxes for one year or even sooner, from 69.49 percent in 2008 to 72.08 percent in 2016 (compared to the sample average of 64.18 percent and 64.79 percent).Pure chauvinism increased slightly, but an even larger increase in proimmigrant sentiment more than mitigated its effect.The relationship between immigration and social housing expenditures as a percent of GDP is also consistent with the sample average.The trendlines suggest the same competing causal explanations: immigration either negatively affects housing expenditures immediately or positively affects them with eligibility lags.
Europe's (Lack of) Welfare Chauvinism This is not to say that immigration in France is harmonious.The rise of the rightwing National Front suggests that native French people are growing weary of sustained immigration.Marine Le Pen, the National Front's presidential candidate in 2017, won 33.9 percent of the popular vote in the run-off round.Though a minority, almost 11 million French people sided with the notoriously anti-immigrant party.In 2015, the National Front clarified its immigration policy goal-net legal migration of 10,000, significantly less than 2013's 140,000 (AFP, 2015).Further, there is evidence that ethnic minority immigrants face significant discrimination.Bonnal, Boumahdi, and Favard (2012) find that non-European foreigners on waiting lists for social housing experience much lengthier waits than others, despite the legal obligation to base social housing allocations solely on welfare criteria.
Altogether, France mirrors the sample average findings regarding immigration's impact on the welfare state.In my (albeit limited) case study, I find no evidence of overtly welfare chauvinistic public attitudes or policy changes in France.Immigration does not appear to be a significant threat to welfare state solidarity, but it could be problematic precisely because immigrants enjoy access.Immigrants' disproportional use of certain welfare programs, like unemployment insurance and social housing, might continue to add pressure to France's ability to finance these welfare programs.

Conclusion
Mixed results surround my hypotheses of immigration's effect on welfare spending.Although several specifications do point to negative pressure on housing spending, the findings' sensitivity to model specification is a red flag that these results are not robust.The other specifications that find no statistically significant relationship between immigration and welfare spending are likely more accurate.The data most strongly support the null hypothesis or H4: increased immigration does not significantly worsen natives' beliefs about immigrants or the welfare state, and even if it did, political opposition successfully blocks attempts to cut spending.Another alternative explanation is that natives express anti-immigrant sentiment not by depriving them of social assistance, but by attempting to limit future immigration altogether.They might be sympathetic to the struggling immigrants who have https://scholarship.claremont.edu/urceu/vol2019/iss1/3already come while supporting increasingly restrictive immigration policy.
This area of research is ripe for future study.Comprehensive data on social spending in Europe is currently only available through 2015, the year in which massive spikes in immigration (and specifically asylum seekers) skyrocketed.If there are eligibility and policy lags, it may take several years to see the effects in social spending data.Further, more reliable estimation techniques could shed light on the issue, like correctly identified instrumental variables which mitigate endogeneity problems.The 2015 immigration surge could be a suitable candidate for an event study.More country-specific case studies could examine the relationship between immigration and welfare support by harnessing regional variation within a country.Further, investigating this relationship in Southern and Eastern Europe would be instructive since these regions tend to adopt more restrictive immigration policies.Finally, future work could explore alternative theoretical modeling if the true relationship between immigration and social spending is non-linear.
Altogether, the practical implication of these findings is that there is no reliable historical evidence to suggest that immigration is a menacing threat to Western Europe's social welfare systems.This holds true for both the sample average of 13 Western European countries and the France-specific experience.On the contrary, immigration might be part of the solution to Europe's demographic pressures on the welfare state (although Murray [2018] criticizes this idea as a Ponzi scheme).Europeans seem to have experienced a period of increasing immigration without developing overtly welfare chauvinistic sentiments.At least not yet.Recent surges in immigration and contemporary populist tendencies may eventually expose social spending programs to more concrete threats of retrenchment.Until then, these results provide a contrasting voice to the current literature that predicts Europe's demise.Like that of Mau, Steffen, and Burkhardt (2009), my analysis demonstrates that public attitudes towards welfare are not just a simple reflex reaction to the degree of ethnic diversity or the influx of immigrants into a country.They are mediated through institutional, economic, and program-specific factors.Even though there might be significant challenges for Europe's welfare state, these findings suggest that immigration does not yet pose an existential threat.

Figure 1 .
Figure 1.Causal Chain between Immigration and Welfare Spending

Figure 2 .
Figure 2. Immigration and Social Spending (sample average)