Regional and Ethnic Minorities

While members of regional and ethnic minorities can reach the highest echelons of power, in most contexts they remain politically marginalized and under-represented in formal politics. The heterogeneity of regional and ethnic minority groups creates a challenge for the study of representation if one wants to avoid the traps of essentialism and unrealistic assumptions. The inclusion of regional and ethnic minorities in legislatures and government can increase trust, alleviate conflict, and provide substantive representation. Much evidence shows that, on average, representatives of regional and ethnic minorities work in the name of their respective groups, especially in ‘low-cost’ activities like asking parliamentary questions. Such substantive representation should be the guiding principle, but the relationship between descriptive and substantive representation seems moderated by the context in which legislators operate.


Introduction
The elections of Barack Obama as President of the United States in 2008 and Leo Varadkar as Taoiseach 1 in 2017 demonstrate that members of regional and ethnic minorities can reach the highest echelons of power. Yet, in many contexts regional and ethnic minorities remain politically marginalized and under-represented in formal politics (Bird, Saalfeld, and Wüst 2010;Ruedin 2013). This under-representation poses a fundamental threat to the legitimacy of liberal democracies, which partly draws on having representative bodies that reflect the population (Dahl 1985). In legislatures around the world, anything between the absence of minority representatives to -in rare cases -a numerical over-representation can be observed.
This chapter focuses on ethnic and regional minorities, a dimension the literature on political representation has long neglected despite ethnic cleavages being salient in many places (Lijphart 1999). Traditionally researchers have sometimes assumed that all so-called minorities of power are equivalent when it comes to political representation, but research has shown that this is not the case (Ruedin 2010). While there are similarities in some respects, clear differences exist in others. Similarities include the importance of trust in representatives from the same group, and (contested) claims to substantive group-specific interests. Particular to regional and ethnic minorities is that they are not numerically relevant in all countries: Some populations are ethnically homogeneous with no politically relevant minority groups. If such minority groups exist, the relevant cleavage is not always apparent: race, ethnicity, language, urbanity, geographical region, or a combination of these. With this comes the challenge of identifying minority groups without making the assumption that all members of the group are the same -an issue more commonly discussed in the context of the representation of women (McCall 2005). There are many reasons why we care about the inclusion of regional and ethnic minorities in legislatures and government. To start with, members of regional and ethnic minorities tend to trust members of their own group more (Schildkraut 2016). The inclusion of minority groups in legislatures also serves as a strong symbol of equality (Saideman et al. 2002). Researchers commonly assume that regional and ethnic minority groups have specific interests that are best represented by members of their group. The nature and existence of such group-specific interests is sometimes contested, but from a normative point of view, the perception of such interests matters, not their objective existence. Moreover, ethnic minorities may be geographically clustered, ethnicity often reinforces differences of social class, and it may be reflected in rather than cutting across party lines (Htun 2004). For these reasons, including regional and ethnic minorities in positions of power can alleviate open conflict (Van Cott 2005).

The chain of representation for regional and ethnic minorities
At its basic, the chain of representation for regional and ethnic minorities does not differ from other forms of representation: Citizens elect party candidates into legislatures to articulate their interests and influence policy (Introduction in this Handbook). Throughout the chapter, I argue that there are various challenges to studying representation when it comes to regional and ethnic minorities. To outline these challenges, the chapter follows the key actors along this chain of representation. To start with, the representation of regional and ethnic minorities assumes that there is such a thing as ethnic groups in society. Ethnicity refers to social entities that share areal or assumed -common origin, a cultural-linguistic legacy that collectively ties members of the group and is transmitted across generations (Ben-Rafael and Sternberg 2015;Jenkins 1997). The meaning of boundaries between regions and ethnic groups are socially constructed, and power plays a role in determining which of these are politically salient (Posner 2017). Boundaries can be drawn on the basis of race, ethnicity, religion, culture, urbanity, geography, immigration status, 2 or a combination of these. Particularly comparative studies need to justify the boundaries they examine and beware that a label like 'Asian' can mean different things in different contexts. In the United States, 'Asian' refers to minorities with an origin in countries such as China, Japan, or Korea. In the United Kingdom, by contrast, 'Asian' refers to minorities with an origin in India and Pakistan. Depending on the motivation of the research, it can be useful to focus on politically relevant groups, but doing so can contradict the observation that regional and ethnic identity may be particularly important for those largely excluded from formal politics. The nature and relevance of these boundaries can be controversial and should probably constitute a future line of inquiry in itself. Just counting the size of regional and ethnic minorities is challenging. Because regional and ethnic identities can be multi-dimensional, ethnic groups can be fluid and context dependent. Put differently, membership in a particular minority group is not necessarily fixed, with relevant ethnic boundaries and their importance open to change over time (Strijbis 2019). 3 Context dependence means that in some situations membership is highlighted and salient -e.g. to secure resources -while in other situations it may be subdued or even denied. To make matters more challenging for researchers, ethnic groups are often nested or hierarchical in the sense of groups and subgroups that may be distinguished depending on the context (Marquardt and Herrera 2015). With this in mind, it should be clear that it is often inappropriate to assume that the number of minority representatives is equivalent to the number of representatives from an ethnic minority party where such parties exist. Ruedin (2013) argued that with a focus on political representation, ethnic differences can be regarded as sufficiently fixed in time to count the number of representatives belonging to ethnic and regional minoritiesbut we cannot assume that they stay the same over time. He also demonstrated that different sources tend to agree on the salient divisions and how many representatives fall into a particular group. Because scope and boundary issues cannot always be resolved, future work should try to capture the uncertainty in measurement, and include this uncertainty in empirical models whenever possible. This encompasses efforts to verify that potential markers like skin colour are salient, or whether ethnic boundaries draw on other forms of difference in a particular case. If we take the literature on intersectionality seriously on better capturing the complexity of social life (McCall 2005), there are serious consequences for work on the representation of regional and ethnic minorities. This literature highlights the diversity and heterogeneity within regional and ethnic minorities, suggesting that not all members of these groups are represented or excluded to the same extent (Celis and Mügge 2017). This differential effect is referred to as intersectionality-plus by Weldon (2008), suggesting that there are effects for being a woman and for belonging to a minority group, but also separate effects for belonging to both. Indeed, political inclusion and marginalization can occur for different subgroups of regional and ethnic minorities at the same time (Mügge 2016;Mügge and Erzeel 2016). For the study of political representation, intersectionality means capturing membership in different socially and politically relevant dimensions. This is a challenge only few studies, to date, have taken political representation by chance alone (Ruedin 2013). up directly (Hughes 2011(Hughes , 2013(Hughes , 2016Krook and O'Brien 2010;Murray 2016) or indirectly (Ruedin 2010), especially for dimensions like sexuality that are not readily visible. One solution is to bring the study of representation to the individual level (Ruedin 2012), although this approach should be regarded a complement rather than a replacement of existing approaches. Whether political representation should focus on the public, citizens, or voters (as opposed to non-voters) is a topic that requires further theoretical work. Lack of representation and political marginalization can be a reason for regional and ethnic minorities to abstain from formal politics, directly affecting the first link of the chain of representation. The robust descriptions of this phenomenon suggest that the viability of candidates, identity, cross-pressures, tactical voting, trust, and differences in interest and participation may all play a role (Fisher et al. 2015;Kolpinskaya 2017;Martin 2016;Ruedin 2018;Sobolewska 2013;Thrasher et al. 2013). More research and theory building is necessary to understand why this is so in some cases and not in others.

National Legislatures
National legislatures have been the focus of much work on the political representation of regional and ethnic minorities. Directly elected by the voters, they constitute the second link in the chain of representation. 4 A central assumption is that group membership of representatives can be perceived by voters, matters to voters, and that boundaries of group membership correspond to those at the population level. Within countries these assumptions are probably not unreasonable (Ruedin 2009), but they remain assumptions. On this basis, there is by now a significant body of literature exploring the descriptive representation of regional and ethnic minorities -much of it focusing on individual cases (Barker and Coffé 2018;Bergh and Bjørklund 2003;Garbaye 2005;Geisser 1997;Morden 2017;Saggar and Geddes 2000;Schönwälder 2013) with comparative research emerging (Bird 2003;Bird, Saalfeld, and Wüst 2010;Hänni 2018;Hughes 2013;Ruedin 2009Ruedin , 2013. These studies outline numerical underrepresentation of regional and ethnic minorities, although the degree to which minority groups are under-represented seems to vary, and there are a few instances where minority groups are over-represented like Whites in South Africa or Chinese in Trinidad and Tobago (Ruedin 2013). While some studies engaged in head counting of representatives from regional and ethnic minorities highlight that descriptive representation is valuables in itself (Saideman et al. 2002), many assert that numerical under-representation reflects structures that exclude minorities (Leyenaar 2004;Phillips 1995). A stronger argument asserts that descriptive representation is the best way to ensure substantive representation (Mansbridge 1999): We insist on the presence of regional and ethnic minorities to have their interests reflected when new policies and laws are written. The existence and nature of these substantive interests, however, is a subject of considerable controversy. On the one hand, there is the position that group membership and interests are different things, focusing on the individual and basic needs like jobs, security, or cultural expression. To some extent, this view is informed by the ideal that the issues preferences behind substantive representation should dominate politics, and that different groups in society should not be disadvantaged. On the other hand, there is the assertion that groups have inherent interests, perhaps most obvious when it comes to control over land with regional minorities. Mansbridge (1999) also highlights that when trust is low, such as in situations of ethnic conflict, descriptive representation is normatively desirable. In either situation, perceptions of group interests are relevant, not whether these objectively exist. Others focus on the experience of having grown up as a member of a regional or ethnic minority group -an experience difficult to relate to for those not directly affected (Walby 2009). There is much evidence that on average representatives of regional and ethnic minorities work in the name of their respective groups (Bird 2010;Broockman 2013;Casellas and Leal 2010;Juenke and Preuhs 2012;Minta 2011;Saalfeld and Bischof 2013;Saalfeld and Kyriakopoulou 2010;Wüst 2010Wüst , 2014. This can be shown by an increased number of parliamentary actions related to these groups, like putting forward motions, being part of commissions, asking parliamentary questions, votes in the chamber, or oversight work (Minta 2011;Saalfeld and Kyriakopoulou 2010;Wüst 2014). In this sense, representatives from regional and ethnic minorities articulate public interests. Bailer and Ohmura (2018) show that in Germany parliamentarians from under-represented groups use parliamentary questions to represent 'their' group particularly at the beginning of their political career. Where studies are willing to identify minority interests, they often find an association between descriptive and substantive representation. Bird (2010) examined how parliamentarians from visible minorities are more likely to intervene on issue that she argued represent the interests of these regional and ethnic minorities. Casellas and Leal (2010) make similar observations for the US Congress. In the same direction, Juenke and Preuhs (2012) demonstrate that the voting behaviour of Black and Latino legislators in the US differs systematically from other legislators, but they stop short of specifying the actual minority interests they assume this to reflect. Saalfeld and Kyriakopoulou (2010) observed minority parliamentarians in the UK asking more questions related to minority interests. Kolpinskaya (2017) highlights, however, that parliamentary questions are a 'low-cost' activity for parliamentarians: Interventions on behalf of minority groups may be more symbolic than substantive. For this reason, work by Minta (2011), who focused on parliamentary oversight as more 'costly' activities, is important. Minta shows that Black and Latino legislators are more likely to testify in favour of minority interests, or write letters to officials to urge them to take action in the interest of members of minority groups. A challenge for these studies is to separate out electoral calculus from intrinsic motivation to represent a specific group, because often the two cannot be separated. The importance of electoral incentives is probably best illustrated with work in the United States that majority legislators have become less responsive to the interests of their Black constituents because they assume that other (Black) legislators would represent these interests (Lublin 1997;Overby and Cosgrove 1996). Another challenge of these studies is that they assume that representatives have an opportunity to voice these interests in the legislature (compare Goodin 2004). Future work should examine these opportunities to represent, which may include questions of framing: how politicians present and justify issues in political debates in addition to which issues they address (Morden 2017). Research on immigrant minorities in Europe shows that group membership can be an important -even dominant -political issue for the affected parliamentarians. For example, Saalfeld (2011) finds that in Europe ethnic minorities are more likely to ask immigration-related questions in parliament (see also Saalfeld and Bischof 2013;Saalfeld and Kyriakopoulou 2010;Wüst 2010). The visibility of immigration origin may be more important than the immigrant origin in itself. Furthermore, minority representatives are more likely to act for immigrants if they represent a constituency with a relatively large share of minorities (Wüst and Saalfeld 2011). These findings are in line with the results by Kroeber (2018) who showed that substantive representation is more likely when group membership of the politicians and the composition of their constituency align, and Hänni (2018) who examined ethnic and regional minorities more broadly. In a comparative study, Hänni (2018) showed that presence in legislative chambersdescriptive representation -does not necessarily lead to substantive representation of ethnic and regional minorities. She highlights that the relationship between descriptive and substantive representation is moderated by the context in which legislators operate, like having a sizeable minority group in the population or a supportive government. At the same time, without descriptive representation we do not seem to observe substantive representation. Rather than focusing on the overall relationship between descriptive and substantive representation in government, it might be more fruitful to focus on specific instances where substantive representation can be said to take place (or not). From this point we can work towards understanding when and how descriptive and substantive representation relate, rather than trying to establish a link at the aggregate level where other influences may mask the underlying processes leading to representation. Future research should also seek innovative designs to better capture the substantive interests of minority groups, especially ways to move beyond assumptions and towards possibilities to treat differences in interests within groups. Experiments where voters send fictitious requests to representatives have been used to show that members of regional and ethnic minorities are often to some extent intrinsically motivated to represent group interests (Broockman 2013;Costa 2017;McClendon 2016). Yet, electoral calculus can ensure that non-minority representatives also work in the interest of regional and ethnic minority groups in the legislature (Sobolewska 2013). This is particularly relevant in cases where electoral districts, or the geographical distribution of ethnic minorities mean, that minorities are not the majority in a particular district. While we have empirical evidence for intrinsic motivation and electoral incentives as Mansbridge (1999) predicted, future research should complement these findings with a better understanding of when and how representatives work in the name of regional and ethnic minorities. Indeed, Giger et al. (2019) suggest that intrinsic and extrinsic motivations may interact to some extent. Furthermore, there is some evidence that many minority representatives are reluctant to represent their own group, but feel pressed to do so (Wüst 2014;Wüst and Saalfeld 2011). Similar pressure has been noted for women to represent the interests of women (Reingold and Swers 2011;Schwindt-Bayer and Palmer 2007). Experimental approaches like Broockman (2013) are unable to differentiate intrinsic motivations from feeling obliged into representing groups, and other approaches are necessary to understand the role of such pressures. Focusing on representatives in the legislature, more work is needed to examine the expectations of different actors that members of regional and ethnic minorities should represent the interests of 'their' group. While there is evidence of expectations from the party leadership (Murray 2016), there are also expectations from the population and voters. A careful examination of substantive representation would link these expectations in the population with the work of representatives -rather than assume what the interests of the groups are. This means that future research should focus on how substantive representation takes place, rather than be preoccupied with the question whether (on average) it does take place and whether it is linked to descriptive representation. This way, future research can move beyond the assumption of a simple direct link that reduces ethnic and regional minorities to a single homogeneous group, and also consider the possibility that legislators represent a privileged subgroup only (Mügge and Erzeel 2016). Novel approaches like process tracing are needed to complement the commonly studied policy outcomes in order to have stronger claims that we truly observe substantive representation. The heterogeneity of regional and ethnic minority groups is a challenge for the study of representation more generally. To date, only few studies have examined differences within groups (see Celis and Mügge 2017 for a discussion). When it comes to asserting or measuring the representation of substantive interests, these differences within groups are an important aspect that future research should examine. On the one hand, these considerations highlight that just counting parliamentary interventions that mention the interests of a regional or minority group will fail to capture whether the group is represented in its diversity. On the other hand, a focus on these differences within groups can highlight that what are purportedly group interests, are in some instances interests that transcend groups and may have been associated with groups due to historical reasons. For example, economic interests of Black South Africans probably do not constitute an interest of the group other that they are vastly more affected by poverty than other population groups, but even a rich Black South African needs recognition and protection from racial discrimination. A different approach has been championed by Saward (2006Saward ( , 2010 to focus on claims to representation. Here counting individual representatives and considerations of heterogeneity take a back seat, with political actors claiming to represent a particular regional or ethnic group in focus (see also Montanaro 2012). This approach is compatible with intersectionality -a legislator can claim to represent Black women in particular -and fluid ethnic boundaries in the sense that depending on the context the same politician can claim to represent different groups. As Rehfeld (2006) highlighted, however, claims to representation are limited because there is also an audience that needs to accept a claim to representation as legitimate. In free elections, this is apparent for political parties, but less so for group membership because voters did not explicitly legitimize a candidate because of their ethnicity. This brings us back to questions of trust: whether regional and ethnic minorities trust legislators to represent the interests of their group. In this sense, there is room for studies capturing claims to representation alongside feelings of representation and possibly descriptive representation.

Governments
Questions of substantive representation come to the fore when we look at representation in governments. Here arguments of numbers are typically less obvious, though normatively it is desirable to include regional and ethnic minorities in government. Especially when it comes to symbolic representation, the inclusion in government seems of great importance (Sharp 2014). Because governments implement policies, they play a particularly important role for ethnic and regional minorities. A central question is whether the inclusion of ethnic and regional minorities in government (and in legislatures) shapes policies with substantive impact on these ethnic and regional minorities. Establishing such a link empirically is fraught with difficulties to do with the many factors that influence policies: party politics, commissions, lobbying behind the scenes, oversight, but also economic constraints and limited budgets (Minta 2011;Ruedin 2013). These challenges notwithstanding, the relationship between budget allocations and policies on the one hand, and substantive interests of regional and ethnic minorities on the other hand is not always straightforward. While many studies in the United States suggest that descriptive representation and substantive representation are linked, other studies are more careful in claiming an association: Much depends on the willingness to identify minority interests, with the risk of treating ethnic and regional minority groups as homogeneous groups with unified interests. At the level of government, a tension becomes apparent between demands for representing regional and ethnic minorities on the one hand, and clientelistic behaviour on the other hand. While this tension may indeed exist at the level of the legislature, it is readily apparent when it comes to governments. In liberal democracies, both the inclusion of different groups in society, as well as the absence of clientelistic politics are considered desirable. For politicians finding the right balance can be difficult, and future work should examine how these different demands affect the work of government.

Institutions
Irrespective of the particular workings of representation in legislatures and governments, the literature on regional and ethnic minorities examines how different institutions affect the chain of representation. With this, institutional differences are often cited as reasons for the numerical under-representation of ethnic and regional minorities in formal politics. Electoral institutions are an obvious target for investigation because they affect which candidates and parties are elected into the legislature and government. Many contributions have underscored the role of the electoral system, arguing that proportional representation should benefit the inclusion of regional and ethnic minorities (Barker and Coffé 2018;Lijphart 1999). What is at work here is a combination of the district magnitude and the behaviour of parties in selecting candidates (see also Kook 2017 on electoral thresholds). A larger district means that a larger number of candidates is selected, which is relevant for regional and ethnic minorities when we bring in the second part: the tendency for parties to place majority candidates at the top of lists. With more candidates selected, the likelihood increases that a (minority) candidate further down the list is elected (Ruedin 2013). When minorities are geographically concentrated, the purported advantages of proportional representation systems are no longer obvious (Ruedin 2009;Wagner 2014). Moser (2008) argues that in the case of ethnic and regional minorities, proportional representation systems need to be combined with ethnic parties to positively affect representation. What is more, as stable institutions, electoral systems are not well suited to explain changes over time. Quotas and reserved seats constitute other electoral institutions that commend themselves for increasing the number of ethnic and regional minorities in formal politics. Contrary to gender, when it comes to regional and ethnic minorities, reserved seats are more common than party quotas (Htun 2004;Zuber 2015). On the one hand, the mechanism is relatively straightforward, and we can expect a direct impact on the number of minority representatives as a consequence. On the other hand, the way such quotas and reserved seats are implemented often means that they are ineffective to increase the number of minority representatives to a level we would expect from their share in the population (Ruedin 2009(Ruedin , 2013. More research on the origins and implementation of quotas is needed, requiring better longitudinal data. We know little about when reservations for regional and ethnic minorities are initiated, what groups are considered, and in what circumstances colour-blindness can lead to better inclusion of minorities. This research could also relate to the consequences of quotas in the style Clayton and Zetterberg (2018) who have looked at gender quotas. Clayton and Zetterberg examined the effects of quotas on public spending, but importantly worked from the purported causes to the effects in the analysis rather than trying to identify causes of observed effects (see also Jensenius 2015). On a normative level, there is a tension between including regional and ethnic minorities on the one hand, and individual freedom highlighted in liberal democracies. There is a danger that quotas and reserved seats relegate regional and ethnic minorities to the positions put aside for them -be this reserved seats or ethnic parties -, and overall marginalizes their position in formal politics, especially in commissions and ministries with consequential powers. Murray's (2014) suggestion to use quotas that limit the majority group is difficult to implement when there are more than two regional and ethnic groups in a country, but may work well for gender. Research on other structural factors is less common. Spicer et al. (2017) highlighted that the political opportunity structures commonly examined in the protest literature can be fruitfully applied to questions of political representation. Both institutional and cultural aspects of the political opportunity structure are highlighted, but the link to specific social mechanisms and how these relate to the chain of representation are not yet welldeveloped. Cultural aspects describe attitudes in the population and among the party leadership that are conducive to including a diverse group of representatives, like the view that women are equally competent as political leaders. Ruedin (2009Ruedin ( , 2013 has highlighted how cultural differences between countries can affect all parts of the chain of representation -from voters to candidates, from parties to members of government. An important question is how such cultural attitudes can be changed to encourage the inclusion of minorities. Future research should examine how cultural aspects and formal political institutions interact to shape political representation.
Candidates, Political Parties, and Party Leaders Candidates from regional and ethnic minorities constitute and important link in the chain of representation. When too few candidates from regional and ethnic minorities come forward, voters are constrained in the ability to elect them, which reduces the level of representation in the legislature and government. While the number of representative seats is always small enough that there are certainly enough qualified members of regional and ethnic minorities who could run in national elections, there might be too few motivated to run. Just like many voters choose not to vote, regional and ethnic minorities may choose not to stand (compare Ohmura et al. 2018). There is a need for studies on this decision not to stand, but it is likely that marginalized groups are also politically demobilized -which could lead to a vicious circle where nobody works in the interest of the marginalized group. Mansbridge (1999) highlighted that regional and ethnic minorities often only trust members of their 'own' group to represent their interests. It is clear that representatives who are not members of these groups can represent the interests of these groups by voting in favour of certain policies for instance, but such representation will be incomplete without accompanying trust. In this constellation, politicians from regional and ethnic minorities may be under intensified pressure to stand for their 'own' group, even if they would prefer to focus on other policy areas. Such pressures have not been studied in detail, but they may be particularly strong when there are few other minority representatives who can share the 'burden' of representing minority interests. These pressures may increase because of expectations from other politicians (Lublin 1997;Murray 2016), conformity with a specific social role, concerns over political reputation, and indeed electoral incentives (Saalfeld and Bischof 2013;Sobolewska 2013). Future research should pay attention to political networks and interest groups that may be particularly influential because candidates and elected representatives are likely to value their political reputation. Pressure on regional and ethnic minorities to represent particular groups and interests may be particularly strong when these interests are organized and can lobby politicians. Political parties and the party leadership play an important role in shaping the political representation of regional and ethnic minorities (Geddes 1998;Kittilson and Tate 2004;Messina 1989). They are gatekeepers, and their actions can encourage or discourage minority candidates to put their name forward for an election (Chaney 2015;Murray 2016;Wüst 2010;Wüst and Saalfeld 2011). Parties may choose regional and ethnic minority candidates with a clear expectation that they represent these groups -at least outwardly in terms of symbolic representation. Some parties react to growing diversity in the population with recruitment drives of minority candidates (Fisher et al. 2015). The literature, however, continues to rely on studies on women candidates to draw inferences about regional and ethnic minorities, something that should be complemented with dedicated studies, especially in view of findings by Ruedin (2010Ruedin ( , 2013 that the mechanisms may not be the same for women and other minority groups. Where they are allowed, ethnic parties can positively affect the inclusion of regional and ethnic minorities. Ethnic parties can be defined by the priority they give to issues related to ethnic groups (Zollinger and Bochsler 2012). The presence of ethnic parties, however, cannot be equated with inclusion or indeed substantive representation and legitimacy (Hänni 2018). We also need to look at the geographic distribution of minority populations, the electoral system in place which may restrict the success of ethnic parties, or the inclusion of members from ethnic minorities in mainstream politics. While ethnic parties can provide a concentrated effort to represent substantive interests of regional and ethnic minorities, there is also a danger that they represent the interests of a particular subgroup and reduce mobilization along other interests that cut across ethnic lines, like poverty. By politicizing ethnic differences, ethnic community leaders and ethnic parties may project social cleavages through an ethnic lens, which may encourage ethnic conflict (Zuber 2015). In this context, claims to representation by ethnic parties (compare Saward 2006), but also the acceptance of these claims by mainstream parties deserve careful study. In many places ethnic parties are outlawed for the fear of ethnic conflict. Sometimes we can find quasi-ethnic parties in the sense that certain parties receive an overwhelming part of their support from particular regional or ethnic minorities. Parties will be more likely to put forward candidates if minority voters show a willingness to switch their support (Anwar 1990;Ruedin 2009), or if they are a pivotal electorate. Different motivations of candidates may be revealing here (Sobolewska 2013), and indeed how parties react to these different motivations among the candidates. Rather different is the role of left-wing parties who have traditionally championed the rights and interests of disadvantaged groups who also tend to be politically marginalized. This universalist focus tends to include regional and ethnic minorities, including immigrants (Carvalho and Ruedin 2018). Indeed, left-wing parties may choose (certain) minority candidates to highlight the fact that they champion diversity (Htun and Ossa 2013;Weldon 2006;Wüst 2010). On the one hand, this means that left-wing parties may reduce the pressure on minority politicians to necessarily represent the interest of regional and ethnic minorities, because there are other representatives watching out for these substantive interests. Whether they represent all facets of minority interests, or focus on representing the 'fact of diversity' (Goodin 2004) is a question that has been left unaddressed. On the other hand, questions of trust may limit the extent to which leftwing parties can legitimately speak in the name of regional and ethnic minorities.

Conclusion
This chapter has shown that the political representation of regional and ethnic minorities follows a chain of representation that links citizens to parties to legislatures and governments. The nature of ethnicity, however, makes it difficult to precisely capture representation at each link of the chain of representation. Indeed, throughout the chapter I have argued that much research is needed to understand the nature of representation when it comes to regional and ethnic minorities. It is inappropriate to simply assume that the mechanisms reflect those of the representation of women because both are minorities of power (Ruedin 2010), and only dedicated research on regional and ethnic minorities can establish to what extent there are similarities with gender. At the same time, if one wants to avoid the traps of essentialism and unrealistic assumptions, research on regional and ethnic minorities is plainly difficult (Celis and Mügge 2017). Rather than focusing on descriptive representation, as much of the literature on regional and ethnic minorities has done, the chapter argued that substantive representation should be the guiding principle -acknowledging how difficult it can be to ascertain that (perceived) substantive representation takes place. In this sense, however, future research should explore the dynamic nature of political representation. Regarding substantive representation, this means moving from the question whether substantive representation takes place (and whether it is linked to descriptive representation), to questions of how and when it takes place. In this sense, the pressures on politicians to represent their 'own' group versus anti-clientilistic norms in liberal democracies deserve more attention. The literature presents both institutional and cultural change as possible ways to improve the inclusion of ethnic and regional minorities in formal politics, and future work should also consider how these two forces interact. Longitudinal data, combined with a careful examination of the purported mechanisms at play may help identify causal order between relevant attitudes and levels of representation, and ascertain what happens when quotas are implemented and indeed what happens when quotas and reserved seats are removed. Even if quotas and reserved seats are accepted as a 'necessary evil' to address the historic under-representation of regional and ethnic minorities, the principles of liberal democracy suggest that we should work towards their eventual removal. Another significant question is the consequences of representation (or lack thereof) on trust, conflict, identification, but also social and political participation. Once again, longitudinal data combined with different methods -qualitative, quantitative, and experimental -are likely to be necessary to better understand how regional and ethnic minorities are represented at the different links of the chain of representation. With the challenges of counting who should be considered part of regional and ethnic minorities to start with, we are likely to see many small steps towards a better understanding of political representation in liberal democracies.