1 Introduction

Funding: research for this paper has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 682626). See for more information: https://www.milifestatus.com

Citizenship policies in Europe have been characterized by contrasting trends over the past decade with reforms such as dual citizenship acceptance or shorter residency requirements making citizenship more accessible to immigrants (Vink & de Groot, 2010; Vink et al., 2019). In contrast, the introduction of civic integration and economic requirements have provided new obstacles to immigrants’ naturalisation (Goodman, 2012). Economic requirements can take different forms. They can either be direct, such as proof of economic self-sufficiency, or indirect, such as the payment of substantial application fees. In the European context, especially the United Kingdom and the Netherlands have witnessed significant increases of application fees in the past decade (Stadlmair, 2018). This increase may well prejudice the changes of immigrants of becoming a citizen of the destination country.

While the overall impact of naturalisation requirements on citizenship acquisition rates is well understood (Huddleston, 2020; Huddleston & Falcke, 2020; Vink et al., 2013, 2021), the relevance of economic requirement remains largely understudied in Europe. Administrative fees contribute to the costs of naturalisation and may therefore impact the propensity to naturalise (Goodman, 2010). Existing research in the United States indicates that fees are a substantial barrier for low-income immigrants (Hainmueller et al., 2018; Hotard et al., 2019; Yasenov et al., 2019). However, in the European context the impact of fees on naturalisation propensities remains an open question.

To investigate the role of application fees in the naturalisation decision of immigrants in Europe, in this chapter we look at the case of the Netherlands, where fees have increased from 336 euro in 2003 to 901 euro in 2020 – an increase of 168% points – for a single application, with significant hikes in the fee in 2010 and 2011. Simultaneous changes in the civic integration requirements for permanent residence likely had a positive effect on naturalisation rates among non-EU immigrants and consequently may have obfuscated the impact of the higher fees. As EU immigrants can move freely within the EU and are thus not affected by requirements for permanent residence, we expect that for this group of immigrants, increased fees directly affected the cost-benefit calculation of applying for citizenship. Hence, in this chapter, we analyse naturalisation rates among EU immigrants in the context of increasing application fees by using longitudinal microdata from administrative registers on the complete immigrant population between 2007 and 2014. We use a two-step identification strategy. First, we apply a single-difference regression, based on a fixed-effects model, to investigate immigrant naturalisation rates in conjunction with increased application costs. We subsequently explore impact heterogeneity by household income and use a double-difference regression, based on a difference-in-differences model, to test whether the relevance of the fee increase is conditioned by income groups.

The remainder of the chapter is organised as follows: the next section provides an overview of the increased naturalisation fees in the Netherlands and discusses simultaneous changes in the context of civic integration policy that affect immigrants’ cost-benefit calculations. In the third section we provide a description of the dataset and the empirical strategy to estimate the impact of the fee increase on naturalisation propensities. In Sect. 3.4, we discuss the main results, and end the chapter with a summary of our main results and reflection on the wider implications of these findings.

2 Naturalisation Fees in the Netherlands

Access to citizenship in the Netherlands is regulated by the 1985 Dutch Nationality Act which defines immigrants as eligible for independent naturalisation if they are at least 18 years of age, in possession of a permanent resident permit, legally and uninterruptedly reside in the Netherlands for 5 years (or 3 years if married to a Dutch national) and have made an effort to renounce the citizenship of their country of origin, if they do not lose this automatically (van Oers et al., 2013). Since 2003, immigrants need to demonstrate that they are sufficiently integrated into Dutch society. They have to pass the civic integration exam which tests sufficient Dutch language capabilities (currently level A2, in the Common European Framework of Reference for Languages) and knowledge of Dutch society (see IND (2021a) for more information on current requirements).

In order to naturalise in the Netherlands there is no direct economic requirement, such as economic self-sufficiency. However, besides the costs for the naturalisation exam (which amount to 350 euro in 2020, increased from 260 euro when the exam was introduced in 2003, cf. van Oers, 2006, p. 30) and costs for preparatory courses, immigrants have to pay an application fee. This fee needs to be paid when submitting the application and is not reimbursed when an application is rejected or the applicant withdraws her or his application. In 2020, the fee for an individual naturalisation request in the Netherlands stood at 901 euro (see IND, 2021b for currently applicable fees). Reduced tariffs apply to stateless persons or holders of a residence permit asylum (670 euro in 2020) and for applicants submitting an application together with their partner (1150 euro in total).

As Fig. 3.1 shows, the application fee for naturalisation requests has increased dramatically from 336 euro in 2003 to 901 euro in 2020 for a single application, with significant hikes in the fee in 2010 and 2011. A similar trend applies to reduced fees and to the fees for joint applications.Footnote 1 This trend of increased costs for naturalisation applications is part of a longer trend of increased restriction of Dutch naturalisation policy. During the early ninetees, Christian Democrats (CDA) and Conservative Liberals (VVD) developed an assimilationalist perspective on naturalisation. Where requirements previously had been relatively liberal, reflecting the notion of naturalisation as a stepping stone for integration (Heijs, 1995), this gradually shifted towards a view of naturalisation as the social, legal and emotional completion of the integration process (de Hart, 2007, p. 91). From that perspective, responsibility for successful integration increasingly fell to immigrants, whose integration would be tested rather than facilitated. This is evident in the (re)introduction of the renunciation requirement in 1997 and a language and civic integration test in 2003.

Fig. 3.1
A histogram of the application fee from 2000 to 2020. Over time, the values might well rise from 240 to 875. The values are estimated.

Application fee for a single naturalisation request in the Netherlands over time. (Source: Government Gazette of the Kingdom of the Netherlands)

Full size image

This restrictive turn in naturalisation policy is also characterised by a broader neoliberal trend in Dutch immigration and integration policy (van Houdt et al., 2011; Suvarierol & Kirk, 2015). This includes the notion that public services need to be ‘cost efficient’, which has gained increasing ground (IND, 2004). As a result, the Netherlands has witnessed growing marketisation regarding the implementation of integration policy where the government sets norms for exams, but the implementation of preparatory courses is left to private actors. In 2009, the government concluded that the costs of processing all naturalisation applications were higher than what was covered by the received income from fees. As a result, the fees needed to be substantially increased to ensure a closer approximation of the costs of the procedure (Department of Justice, 2009). After substantial jumps in 2010 and 2011, subsequent incremental increases are based on annual indexed wages.

These restrictions in requirements for naturalisation are mirrored in a decreasing number of naturalisations. While the policy changes were implemented with the aim of encouraging immigrants to integrate, it has been argued that these requirements in practice have led to exclusion (Groenendijk, 2003; van Oers et al., 2013, p. 46). Indeed, in particular vulnerable immigrants from economically less developed countries (Peters et al., 2016) or with lower levels of education (Vink et al., 2021) were deterred by restrictive citizenship policies. These are also the immigrants who stand to benefit most from naturalisation to mitigate their structurally disadvantaged position in for instance the labour (Peters et al., 2018, 2019) or housing market (Leclerc et al., 2022; Peters, 2020). While policy makers never explicitly intended these policies to serve a selective purpose, they paradoxically may hamper integration by obstructing a realistic pathway to citizenship for those immigrants who need it most. Whether the fee increase also had this stratifying impact remains an open question, however.

The fee increase coincided with a number of other policy changes. First, in January 2010, passing a civic integration exam became a requirement to receive a permanent residence permit for immigrants with an integration obligation under the Integration Act. The same is true for those who wish to receive a temporary residence permit for family reunification purposes. This requirement applied to most non-EU immigrants, but not to EU immigrants and their non-EU partners and neither to citizens from Norway, Iceland and Switzerland (Besselsen & de Hart, 2014, p. 16). Whether immigrants from Turkey could be obliged to do the civic integration exam as a condition for permanent residence was disputed. Initially, they were included in the new obligatory civic integration policy; yet in 2011 this was rejected in court with reference to the free movement rights of Turkish citizens under the Association Agreement between Turkey and the EU. Second, funded opportunities to prepare for the exams were offered, which was not the case previously (van Oers et al., 2013, pp. 31–32).

The cost-benefit calculation for naturalisation is likely affected by these changes because the impact of the increase in the application fees is offset by the financial support that was made available to prepare for the civic integration tests from 2010. Since then civic integration courses were offered locally and financially covered from the ‘participation budget’ allocated by the State to municipalities as part of a so-called ‘Deltaplan’ to ensure greater participation in language and integration courses (Algemene Rekenkamer, 2017, p. 16). The assumption that offering free participation changed the cost-benefit calculation of naturalisation decisions is supported by the observation that, after the funded training opportunities were introduced, the number of participants in preparatory courses and naturalisation tests increased (van Oers et al., 2013, p. 32). Moreover, since passing the now-required tests is also a pre-condition for naturalisation, ‘skipping the permanent residence stage in favour of naturalisation seems rather self-evident’ (van Oers et al., 2013, p. 31). For both reasons, for non-EU immigrants we may expect an increase in the propensity to naturalise, in contrast to an expected decrease due to the fee increase for EU immigrants.

In the next section, we introduce our data sources, clarify the empirical focus on EU immigrants and discuss our empirical strategy to identify how naturalisation rates may have changed for this group in the context of increased application fees.

3 Data and Methodology

3.1 Data

To study the impact of the increase in application fees for naturalisation in the Netherlands in 2010, this paper draws on administrative register data from Statistics Netherlands. These data include all legally registered individuals in the Netherlands over time, allowing for a comparative analysis of immigrant naturalisation rates before and after an increase in the application fees in 2010.

As outlined above, for most non-EU immigrants the fee increase coincided with simultaneous policy changes implying, on the one hand, that taking the civic integration test became obligatory for continued residence in the Netherlands and financial support for taking preparatory courses became available, on the other. Hence, these changes are expected to offset possible effects of higher application fees and likely increase the propensity to naturalise (van Oers et al., 2013, p. 31; Besselsen & de Hart, 2014, p. 31). To disentangle the impact of the application fee from simultaneous policy changes in civic integration policy, we focus on immigrants from EU countries, as well as those from associated states Iceland, Norway and Switzerland with which the EU shares a freedom of movement regime (hereafter we refer to both citizens of the European Union and from associated states as EU immigrants). These immigrants are exempted from the integration requirement for residence because this would violate their right to freedom of movement within the EU. Moreover, the application fee for naturalisation likely plays a relatively important role for these immigrants. Indeed, since EU immigrants already enjoy many of the rights that Dutch citizenship would provide, the benefits of naturalisation are few (e.g. voting rights at the national level, or the symbolic value of being a citizen of the country in which you reside). On balance, a high application fee may thus be particularly relevant to EU immigrants, whereas this is less likely to dissuade immigrants for whom the benefits of naturalisation are larger. In sum, we focus on EU immigrants to ensure that our estimation is not biased by coinciding integration policy changes that affect the naturalisation cost-benefit decision, and because application fees are likely to play an important role in their cost-benefit calculation for naturalisation.

In light of these considerations, the research population consists of all foreign-born EU citizens registered at a Dutch municipality between 2007 and 2014 (observations = 1,230,925; N = 203,962). We select this observation window to facilitate a comparison of the period before and after the increase of the application fees for naturalisation in 2010 and avoid confounding period shocks due to new origin countries entering the research population as a result of EU enlargement, which may interfere with our identification strategy (outlined below). To ensure stability in the sample, the observation window includes member states that joined in 2007 (Bulgaria and Romania) from the start. For the same reason, we exclude immigrants from Croatia (which joined in 2013) from our analysis (6134 observations). Note that the sample size grows over the observation period due to a substantial increase in the number of EU migrants in the Netherlands from 2005 onwards (CBS, 2020). Furthermore, we focus on immigrants who are born abroad and whose parents were born abroad, are 18 years or older and not a Dutch citizen at the moment of arrival in the Netherlands, and are eligible for naturalisation. These immigrants are observed annually on the first of January of each year.

3.2 Identification Strategy

Our identification strategy is based on a two-step approach: a single-difference and double-difference regression. The single-difference regression analyses the effect of the fee increase for the immigrant population overall, as well as in sub-group analyses for immigrants from low and high-income households separately. The double-difference regression then provides a more robust test of the differential impact of the fee increase for immigrants from low-income households versus those from high-income households.

The single-difference regression is based on a fixed-effects model, and is formalized as follows:

$$ {Y}_{icmt}=\propto + Post{2010}_t+{X}_{icmt}+{\gamma}_c+{d}_{ct}+{p}_t+{\delta}_m+{\varepsilon}_{icmt} $$
(3.1)

where Yicmt indicates whether an immigrant i from origin country c and municipality m is a Dutch citizen in year t. Post2010 is a dummy that is set to unity in 2010 and all subsequent years, which is used to identify the impact of the fee increase. Xicmt is a vector of control variables at the individual level, including gender, age at migration, age at migration squared, the partner status (including whether the potential partner is a native-born, a naturalised or non-naturalised citizen), having minor children, employment, household income and the highest level of education. The model also includes origin country fixed-effects (γc), as well as a further control for dual citizenship toleration of origin country c at time t (dct). We include municipality fixed effects (δm) to account for local differences, in particular regarding potential differences in the coverage of fees from special welfare budgets. Finally, the model has two period controls: the share of votes for far-right parties and the annual employment rate (pt) (Alarian, 2017; Graeber, 2016). ∝ denotes the intercept and εicmt the error term. We account for potential heteroskedasticity by calculating robust standard errors clustered at the individual level.

As outlined above, we expect the relevance of the fee increase (as identified by Post2010t) to be particularly strong among households with lower levels of income. To test that expectation, we perform subgroup analyses for immigrants with below or equal to/above modal household income (€37,500 in 2010).

To test the robustness of the differential impact of the fee increase by household income group, we formulate a double-difference regression. This model draws on the logic of a Difference-in-Differences (DiD) design, and is formulated as formalized:

$$ {Y}_{icmt}=\propto + HHin{c}_{icmt}\ast t+t+{X}_{icmt}+{d}_{ct}+{\gamma}_c+{\delta}_m+{\varepsilon}_{icmt} $$
(3.2a)
$$ {Y}_{icmt}=\propto + HHin{c}_{icmt}\ast Post{2010}_t+ Post{2010}_t+{X}_{icmt}+{d}_{ct}+{\gamma}_c+{\delta}_m+{\varepsilon}_{icmt} $$
(3.2b)

Model (3.2a) tests the parallel trend assumption. More specifically, it draws on the expectation that if Post2010t indeed captures the relevance of the fee increase, and this matters in particular to low income households, then we should only observe a difference in the naturalisation rate between immigrants with below/above modal household income (HHincicmt) in the years 2010 and after (when the fee increased), all else constant. This is measured through the interaction term HHincicmt ∗ t (note that a control for HHincicmt is included in vector Xicmt). Statistically insignificant coefficients prior to 2010, and negative coefficients from 2010 onwards, are consistent with our expectation. This would indicate that the naturalisation rate prior to 2020 followed the same trends whereas it dropped more strongly among below modal income household from 2010 onwards. Model (3.2b) then estimates the overall DiD coefficient by replacing the individual year dummies in the interaction term (t) with a post 2010 dummy (Post2010t).

4 Analysis

4.1 Trends

Figure 3.2 shows naturalisation rates within the observation period for EU immigrants with below/above modal household income. The former group has a higher cumulative naturalisation rate than the latter (between 39 and 33 compared to 31 to 27%). This can be explained in part by the fact that immigrants with lower levels of income originate more often from countries with lower levels of economic development. It is well established that these immigrants have a higher propensity to naturalise (Graeber, 2016; Vink et al., 2021), although note that the discrepancy is limited here because levels of economic development are generally high within the EU. While we observe a drop in the naturalisation rate, this is gradual over time rather than concentrated around the year 2010. However, the downward trend is stronger among immigrants from below modal income households, in particular after 2010. This is consistent with the notion that the fee increase particularly affected immigrants who would have difficulty meeting those requirements. The decrease in the naturalisation rate of immigrants with lower household incomes is especially strong during the last observation years (2013 and 2014). A possible reason for this is that immigrants who wish to naturalise need to pass a number of language and civic integration tests. The decision to naturalise thus precedes the moment of naturalisation by several years, as immigrants prepare for the formal requirements. An increase in the application fee is less likely to dissuade immigrants who have already decided to naturalise in the past, and have started to prepare for the tests. The impact of the application fee is thus expected to be particularly visible among later observation years, as these contain more immigrants who had not yet decided to naturalise before the fee was increased.

Fig. 3.2
A line graph depicts the naturalization rate from 2007 to 2014 for below-modal and above-modal income. Both lines are dwindling. The modal points are 38.68, 37.05, and 32.66. The points indicated in the modal above are 31.36, 29.85, and 26.99.

Cumulative naturalisation rate of immigrants from the EU (incl. CH, IS and NO but excl. HR) in the Netherlands between 2007–2014 by modal household income. (Source: Statistics Netherlands)

Full size image

5 Single-Difference Regression: Main Effect and Impact Heterogeneity

While the trends are consistent with our expectations, no firm conclusions can be drawn about the specific impact of the fee increase due to potential compositional, regional or period confounders. To account for this, we perform a single-difference regression for the full sample based on Model specification (1). As outlined above, we interpret the post-2010 dummy as the impact of the fee increase by holding all other variation at the individual, municipal and origin level constant (see Sect. 3.2 for a list of controls). Results in Fig. 3.3 show that the naturalisation rate decreased by 6.9% points from 2010 onwards compared to the preceding period, all else constant (see Table 3.1 for details). In other words, the general downward trend observed in Fig. 3.2 cannot be fully attributed to variation at the individual, municipal and origin level, or by period effects that we control for in our model.

Fig. 3.3
A plot is divided into the main effect, below modal income and above modal income in the y-axis and x-axis gives the difference in naturalization rate of 2010 to 2014 compared to 2007 to 2009. A vertical line passes through 0.

Linear fixed-effects regression on the heterogeneous effect of the increase in application fees for naturalisation in the Netherlands in 2010 on naturalisation rates (immigrants from the EU, incl. CH, IS and NO but excl. HR; 2007–2014). Dots denote point estimates and horizontal lines correspond to 95% confidence intervals. (Source: Table 3.1; Statistics Netherlands)

Full size image
Table 3.1 Linear fixed-effects regression on the heterogeneous effect of the increase in application fees for naturalisation in the Netherlands in 2010 on naturalisation rates (immigrants from the EU, incl. CH, IS and NO but excl. HR; 2007–2014)a
Full size table

To test our expectation that an increase in the application fee for naturalisation in particular affects immigrants with limited financial means, we perform subgroup analyses for those with a household income below/above modal household income. Results in Fig. 3.3 confirm that expectation (Table 3.1). More specifically, while the naturalisation rate among those with high household incomes decreases by 5.6% points after 2010, it drops by 7.2% points among those with lower incomes. In other words, the negative coefficient in the main model is predominantly driven by those with below modal household incomes.

6 Double-Difference Regression: Conditioned Relevance of the Fee Increase

To delve deeper into the conditioned relevance of the fee increase, we next perform a double-difference regression based on Model (3.2a) and (3.2b), which has two advantages. First, we include a control for the annual employment rate and share of votes for far-right parties in our single-difference model to disentangle the fee increase from coinciding period shocks. This still leaves open the possibility that other such confounders for which we do not control are picked up by our period dummy, and thus bias our estimation of the impact of the application fee. The most robust approach would be to include a control for year fixed-effects, but this is not possible because of collinearity with the period dummy we rely on for impact identification. The double-difference model addresses this problem by drawing on the assumption (as confirmed in Fig. 3.3) that the fee increase mattered in particular to immigrants with lower household incomes. More specifically, it identifies the impact of the fee increase not through a period dummy but with an interaction between time and a ‘treatment group’. While strictly speaking, we do not have a treatment group because the entire sample was subject to the fee increase, our findings in the subgroup single-difference regression confirms that the fee increase mattered in particular to immigrants from below modal income households. By interacting time with a dummy indicating below modal household income, we are free to include year fixed-effects as a control. Second, while the subgroup single-difference models show that the naturalisation rate decreased more post-2010 within the group of immigrants from below modal income households compared to immigrants with higher household incomes, it is difficult to identify the relative difference between these two groups post-2010. Since the double-difference model is based on the full sample, the interaction terms tell us more about how the impact of the fee increase differed between income groups.

Results in Fig. 3.4 based on Model (3.2a) which tests the parallel trend assumption, confirm our expectation: there is no statistically significant difference in the naturalisation rate between immigrants from below or above modal income households prior to 2010 (see Table 3.2). Only after the introduction of the fee increase does the difference appear, from 0.7% points in 2010 to 3.1% points in 2014. Overall, according to our estimates from Model (3.2b), the naturalisation gap between both income groups during the post-2010 period increases with 1.5% points. Two conclusions can be drawn from these results. First, the findings from the single-difference regression models cannot be fully attributed to confounding period effects. Second, the impact of the increase in application fees for naturalisation is indeed stronger for immigrants from households with below modal levels of income, as the single-difference models suggested.

Fig. 3.4
A linear graph plots the difference in naturalization rate between migrants from below or above-modal-income households from 2007 to 2014. The second plot represents the main effect marked between minus 2.5 and 0.0.

Linear difference-in-differences regression on the effect of the increase in application fees for naturalisation in the Netherlands in 2010 on naturalisation rates among immigrants from below modal income households relative to immigrants from above modal income households (immigrants from the EU, incl. CH, IS and NO but excl. HR; 2007–2014). The left panel shows the analysis of the parallel trend assumption, and the right panel shows the overall difference-in-differences coefficient. Dots denote point estimates and vertical lines correspond to 95% confidence intervals

Full size image
Table 3.2 Linear difference-in-differences regression on the effect of the increase in application fees for naturalisation in the Netherlands in 2010 on naturalisation rates among immigrants from below modal income households relative to immigrants from above modal income households (immigrants from the EU, incl. CH, IS and NO but excl. HR; 2007–2014)a
Full size table

To get a sense of the impact magnitude of the fee increase, consider the year 2014. In that year, 81,042 individuals from below modal income households were eligible for naturalisation in the sample. If we multiply the DiD coefficient of that year from the double-difference regression to the affected population, we obtain the number of immigrants from below modal income households who decided not to naturalise in 2014 because of the fee increase, based on EU immigrants’ propensity to naturalise in 2007 (the reference category). This calculation suggests that an estimated 2512 immigrants from below modal income households (81,042 * –0.031) did not naturalise because of the policy change. More generally, if we contrast the overall DiD coefficient to the covariates in the same model (Table 3.3), then the impact of the fee increase is slightly smaller than the difference between having a low rather than high level of education.

Table 3.3 Linear fixed-effects regression on the heterogeneous effect of the increase in application fees for naturalisation in the Netherlands in 2010 on naturalisation rates, including coefficients for covariates (immigrants from the EU, incl. CH, IS and NO but excl. HR)a
Full size table

7 Conclusion

Substantial variation in citizenship policies across Europe (Goodman, 2010; Vink & de Groot, 2010) has given rise to a large field of literature analysing the impact of these institutional conditions for immigrants’ propensity to naturalise. Over the last decade, scholars have increasingly drawn on panel data and quasi-experimental methodologies for that purpose. Although robust identification strategies often place limits on the comparative scope of such studies, there is a growing understanding of the relevance of requirements for naturalisation, including language and integration tests (Vink et al., 2021), nationality procedures (Huddleston & Falcke, 2020) and residence or renunciation requirements (Mazzolari, 2009; Vink et al., 2021). What has remained understudied in the European context, however, are economic requirements, varying from demands on self-sufficiency to application fees. In this chapter, we focus on the latter in the Dutch context, where the application fee has risen markedly over the last decades. We exploit a significant increase of the fee in 2010 to analyse whether, and if so for whom, such requirements matter for the propensity to naturalise. We use Dutch administrative data between 2007 and 2014 on immigrants from the EU and associated states with whom the EU shares a freedom of movement regime. These immigrants were exempted from integration requirements that were implemented in parallel with the fee increase. Moreover, the application fee for naturalisation likely plays an important role for these immigrants given the relatively limited added benefit of a Dutch passport compared to their EU citizenship.

We use a two-step identification strategy, formulating a single-difference and double-difference regression model based on the logic of a fixed-effects and DiD regression respectively. Results from our single-difference models reveal a decrease in EU immigrants’ naturalisation rate after the fee increase in 2010, all else constant. Consistent with our expectation that economic requirements matter particularly to immigrants with limited financial means, subgroup analyses show a stronger decrease among those with below modal household incomes compared to immigrants with higher incomes. To delve deeper into the conditioned relevance of the fee increase, we exploit the observed impact heterogeneity in double-difference models, which confirm that the main findings cannot be fully attributed to unmeasured period shocks, and that there is indeed a statistically significant difference in the relevance of the fee increase by household income. Our tentative interpretation of the stronger impact in later observation years is that application fees in the context of restrictive requirements for naturalisation are particularly relevant early in the decision-making process, resulting in a delayed effect. Immigrants who were already preparing for language and civic integration requirements were less likely to be dissuaded by an increase in the fees than those who still had to decide whether they would invest in becoming a citizen in the future. From that perspective, the impact of the fee increase should be less visible in the initial years after the fee increase, as many migrants will have decided to naturalise before then. The individual year-coefficients in the double-difference regression are consistent with that expectation, but more specific analyses need to confirm the presumed mechanism.

These findings align with conclusions from existing research on the relevance of economic requirements for naturalisation in the United States (Hainmueller et al., 2018; Hotard et al., 2019; Yasenov et al., 2019). They also present several avenues for further research into the impact of economic requirements for naturalisation in Europe. First, future research can investigate the role of different types of economic requirements on naturalisation propensities. While some countries, like the Netherlands and the United Kingdom, have high application fees, other countries, such as Belgium, Denmark, Finland and Germany, demand economic activity, a minimum level of income or no reliance on welfare benefits over a given period prior to the application for citizenship. These requirements are not mutually exclusive, as is evident in the case of Austria which combines strict economic naturalisation criteria with high fees (Stadlmair, 2018). To what extent and for whom specific economic requirements matter for the propensity to naturalise remains an open question. Second, given that economic requirements are only one aspect of citizenship policies governing access to nationality, the question remains how various requirements interact. For instance, immigrants with lower levels of education or from less developed countries of origin are most deterred by restrictive language and civic integration tests (Vink et al., 2021). Since these are typically also immigrants with limited financial means, is the impact of economic requirements in countries with demanding naturalisation tests nullified by selection into naturalisation? In other words, do economic requirements matter more in countries whose overall citizenship policies are relatively liberal? Third, due to the coinciding policy changes for non-EU immigrants, our analysis focused on EU immigrants residing in the Netherlands. This raises the question whether the findings can be generalized to the immigrant population more broadly. Naturalisation rates in the Netherlands are average in the EU (Eurostat, 2021) due to relatively accessible citizenship policies. However, EU immigrants show generally lower propensities to naturalise, and this is particularly true in the Netherlands, where the renunciation requirement is an important deterrent for these immigrants (Vink et al., 2021, p. 11). Similar to the differential impact of restrictive dual citizenship regulations, which affect EU migrants more strongly than non-EU migrants, we expect that due to the limited benefits citizenship acquisition provides to EU immigrants, the costs associated with naturalisation (such as application fees) will weigh relatively heavy in the decision to naturalise. In other words, if we did not observe an impact of the application fees among EU immigrants, it is unlikely that we would observe such an effect for non-EU immigrants. Whether this expectation holds empirically remains to be tested in other studies, in the Netherlands or elsewhere, given that the design of our study that is set around the specific policy context of 2010 only allows a focus on EU migrants. Future research should assess whether application fees have a depressing effect on naturalisation rates for immigrants in general, or whether fees matter most to those who stand to gain least from citizenship acquisition.