Intra-EU Labour Migration and Attitudes Towards the Welfare State 111 In Model 4, I test the direct relationship between foreign-born as a percentage of the population and attitudes towards the welfare state through including the indicator in our main model. The analysis finds that the general indicator of migration has a direct association with preferences for redistribution, higher levels of foreign-born increase the likelihood of supporting redistribution. Model 5 tests the indicator for CEE labour migration and finds no effect and Model 6 combines the two indicators to ensure that any result in Model 5 is not as a result of excluding an indicator for general level of immigration. In both Models, CEE labour migration is insignificant, there is no direct influence of this specific type of migration on attitudes towards redistribution. However, it is possible that this type of migration may influence a different part of the causal mechanisms of the compensation hypothesis, or that there is an indirect influence through the unemployment rate, presented in Table 5.3. Table 5.3 presents the marginal effects of our two migration indicators when interacted with the unemployment rate. Model 7 presents the results for foreign-born interacted with the unemployment rate and excluding GDP per capita (due to concerns about degrees of freedom), while Model 8 reintroduces GDP per capita as a robustness test. I follow the same procedure for CEE labour migration, Model 9 excludes GDP per capita while Model 10 includes it. In Model 11 I include both migration indicators. For foreignborn as a percentage of the population, there seems to be a minor relationship using the interaction variable, however this does not hold up when GDP per capita is reintroduced. For CEE labour migration, this interaction with unemployment appears to be more important. The analysis finds that at high levels of CEE labour migration and unemployment that there is a moderating effect on preferences for redistribution – the coefficient for CEE labour migration is positive while the interaction is negative. This suggests that an individual’s concerns for compensation may only extend so far and that beyond a certain limit or threshold of immigration and unemployment, then there are alternate mechanisms (e.g. anti-solidarity, symbolic and/or economic threat) driving preferences. It may be that the compensation hypothesis cannot explain attitudes in EU states that have high levels of both CEE labour migration and unemployment – although this combination is rare. To explore this in more detail I plot marginal effects graph for Model 9 (see Figure 5.2).
RkJQdWJsaXNoZXIy MjY0ODMw