74 Chapter 4 cover a longer period. However, the data is restricted for two reasons. First, the variables we created to express EU labour migration cannot precede 2004 because the anonymisation process of the EU-LFS prior to 2004 restricts access to information regarding the country of birth of migrants. Even if we had data before this period, theoretically, it would not make sense to include it because EU13 migration under the legal conditions of nominally equal access to the labour market only commenced following this date. Second, the EU-LFS data provided for the creation of the migration indicators end in 2016 and consequently we cannot extend the time frame beyond this point. A final point regarding the sample is that the social assistance replacement rate ends in 2013. 4.3.2 Dependent Variables The “Dependent Variable Problem” has been a debated issue in the welfare state literature for some time now (Allan & Scruggs, 2004; Clasen & Siegel, 2007; Starke, 2006). In particular, Green-Pedersen (2004) argues that depending on how you theoretically conceptualise the welfare state, then the data chosen in order to operationalise it will impact the evaluation of welfare state change. With this in mind, we complement welfare state spending from the OECD’s Social Expenditure (SOCX) Database (OECD, 2024b) – a traditional indicator used to measure ‘welfare state effort’ – with the OECD’s unemployment replacement rate (OECD, 2024a) and a social assistance replacement rate updated from Wang and van Vliet (2016a; 2016b). Even though we control for the number of benefit recipients in the models of social expenditure, as will be discussed below, replacement rates are arguably better indicators of generosity than spending as they are not directly related to the number of beneficiaries, amongst other reasons (Iversen, 2001; Scruggs, 2006). In studies on migration, this might be even more relevant than in other welfare state research because according to the fiscal pressure argument, there might also be a mechanical link between immigration and social expenditure (Römer, 2023). Whether or not this is the case, and the extent to which migrants are net contributors to the welfare state, are debated in the literature (Boeri, 2010). Thus, the use of replacement rates to remove the direct link with number of recipients is useful additional tool. Additionally, Iversen (2001) argues that replacement rates are a good indicator of the extent to which risks in the labour market, such as becoming unemployed, are redistributed. By using a variety of dependent variables in order to operationalise welfare state effort, we aim to capture more dimensions of the welfare state and thus provide a more nuanced view of how immigration influences the welfare states of 16 European countries.
RkJQdWJsaXNoZXIy MjY0ODMw