Territorial variation in territorial representation: the local base of Westminster MPs

Territorial representation, the representation of local entities in the Westminster Parliament, lies at the heart of British democracy. In the recent academic debate, it has been asked whether local representation also needs a local representative and which specific characteristics would constitute ‘localness’ in this context. Investigating the biographies of 1108 Westminster MPs between 2010 and 2019 and exploring different dimensions of local base, this study examines the extent to which the demand for a local representative is actually fulfilled in the British Parliament. We discover a slow yet stable increase in the descriptive representation of the local (DRL) over time as well as notable variations across party and regions, especially among the constituent nations of the United Kingdom. Applying a genuine territorial perspective to our results, we link the detected DRL increase to the general territorialisation of politics in the UK. The fact that the political divergence of its constituent parts affects even the implementation of a core principle of British democracy is finally depicted as another indicator of the deep territorial divisions pervading British politics.


Introduction
Territory lies at the heart of political representation, as 'almost every modern democratic government uses territory in some form to construct constituencies for their national legislature' (Rehfeld 2005, p. xii). This applies in particular to England and the United Kingdom, where the House of Commons has adhered to the territorial of the United Kingdom might be seen as another dimension of a more fundamental political divergence that is increasingly threatening the integrity of the British state.
Our paper is organised as follows. We start with theoretical deliberations about the territorialisation of British politics, followed by a conceptual discussion of localness and the opportunities for local candidates to become MPs ("The territorialisation of British politics and the local base of Westminster MPs: overview and theoretical starting points" section). In "DRL in Britain: the state of the art" section we discuss previous findings on the local base of British MPs and identify the research gap, we will address: territorial variation in territorial representation. Based on preliminary theoretical deliberations and prior empirical results, we then develop expectations as to possible DRL variations across party, territory and time ("A territorial politics perspective on DRL in Britain: expectations" section. "Data and operationalisation" section sets out our dataset in more detail, emphasising the wide range of variables and indicators deployed. This is followed in "Results" section by an examination and interpretation of the results. "Conclusion" section concludes.

The territorialisation of British politics and the local base of Westminster MPs: overview and theoretical starting points
A few decades ago, mainstream social science was dominated by notions of territorial integration and functional differentiation (Keating 1998, pp. 1-15). Globalisation and modernisation were seen as irreversible processes of de-territorialisation both within and beyond the nation state (Appadurai 1996;Papastergiadis 2000). Scholars of territorial politics, however, have since pointed out that 'territory' has remained a fundamental and rather ubiquitous factor in social and political life. It 'structures and gives meaning to social factors everywhere' (Keating 1998, p. 4). Its effect 'is the very combination of other factors within a given place', and thus, it is 'more than merely the sum of its parts' (Keating 1998, p. 5). This effect can be seen on various spatial levels from neighbourhoods to villages, towns, cities, regions, stateless nations up to the state and beyond.
Despite featuring an electoral system formally based on territorial representation, the UK has nevertheless long been depicted as a territorially homogeneous country where society and politics show very little meaningful spatial variation (for example Blondel 1963;Pulzer 1967). Despite the validity of this point for the post-war decades (for a critical position see Bulpitt 1983), at least since the 1970s this is no longer true. In fact, British politics have since undergone a profound process of territorialisation. Peripheral nationalism and devolution have transformed a seemingly territorially homogeneous nation state into a political system characterised by electoral divergence and the territorial disintegration of its party system. Since devolution, Englishness has also undergone processes of renegotiation and politicisation (Kenny 2014). While England still remains without any meaningful institutional representation, a plethora of new English national, regional and even local parties has emerged (Copus et al. 2008;Copus 2009). Today, voting patterns in the United Kingdom not only differ considerably between its constituent nations (Stolz 2019), but also between regions and localities within them (Curtice and Steed 1986;Johnston et al. 1988;Jennings and Stoker 2019), while Britain's traditional, unified two-party system has finally given way to different party systems operating on different levels of government (Dunleavy 2005) and in different areas (Johnston and Pattie 2011). According to Norris' (2000) three ages model, electoral campaigning in postmodern Britain has also returned to more localised forms. All in all, political competition in the UK is now strongly territorialised, i.e. it takes place under very different parameters in different parts of the country.
At the same time, British statecraft seems to have run out of integrating projects and ideas. Following Tom Nairn's (1977) early prediction of 'The Break-Up of Britain' a discourse of 'endism' (Aughey 2013, p. 24ff) is increasingly detecting the terminal decline of British party politics (Awan-Scully 2018), British parliamentary politics (Sheldon 2021) and even '[t]he end of British politics' itself (Moran 2017). Linking this discourse with the DRL debate, strong territorial variation on territorial representation (i.e. varying DRL scores across the constituent nations of the UK) may be seen as another expression of a diverging understanding of how democracy works or should work undermining the integrity of the British polity.
Whether and to what extent MPs can be said 'to be from' the constituency they represent in parliament can be conceptualised in different ways. In their seminal treatment of this question, Childs and Cowley (2011) discuss many personal features that might characterise a representative as local. While we agree that there is no 'definitive definition of what "local" means in this context' (Childs and Cowley 2011, p. 6), we can nevertheless distinguish between different dimensions of territorial linkages and between different territorial scopes.
First, local roots or local linkages can be ascribed to representatives on purely nominal grounds: features like being born in, living or working in the locality allow for a clear-cut binary distinction. 2 While the first criterion could be seen as providing for an essentialist understanding of local roots, the latter two allow for individuals to 'become' local. However, on their own, neither of these criteria might be seen to reflect 'deep roots' as they fail to capture the major reasons for a potential preference of locals for local candidates: common experiences, mutual relationships and shared aims (Childs and Cowley 2011, p. 12). These qualities come in different degrees and different forms. More meaningful, strong or deep local connections may instead be either acquired via long-term residency ('localness by origin') or via active involvement in the social, cultural or political life of the locality ('localness by engagement') or via both at the same time. As to territorial scope, Childs and Cowley (2011, p. 7) point out that loyalty and identity of citizens may be attached to different territorial units. Localness could, thus, refer to the electoral constituency, the city, the region or any other smaller bounded territorial unit. These differentiations constitute a major challenge for any comprehensive analysis of DRL and its operationalisation.
We see the occurrence of DRL in a constituency (in whatever form) and its frequency in parliament as conditioned by a complex institutional opportunity structure consisting of three different stages and three different groups of decision makers (cf. Fig. 1), each with their own evaluation of the normative and strategic value of DRL. Whether a local MP will represent a constituency at Westminster depends on the supply of potential candidates (stage 1: personal decisions), the demand of party selectors (stage 2: party selection) and the demand of voters (stage 3: public election) (for a more comprehensive delineation of the supply and demand model of political recruitment see Norris 1997). The final outcome obviously depends on decisions at each stage, with actors' preferences being highly interdependent. Voters' preferences for local candidates might positively affect party selectors' preferences for them (demand), which, in turn, might encourage more local candidates to come forward (supply). Of course, this also works in the other direction: If only (non-)locals are standing, party and voters' preferences will not matter at all. Finally, it is important to stress that both voters and party selectors have to balance their general preference for local candidates with other major considerations such as ideology, social class, party faction or gender.

DRL in Britain: the state of the art
The literature dealing with DRL is diverse, yet there is no study covering all stages of the recruitment and electoral process, nor are the different stages represented in equal numbers of studies. The vast majority seems to focus either on the outcome in terms of parliamentary seats or on stage three, the voters' preferences. Furthermore, there is considerable dissent as to what constitutes localness and how to measure it.
The most recent comparative empirical study explicitly asking about such linkages, however, revealed that despite their perceived anachronism, the local roots of national parliamentarians in Western Europe have not yet been eradicated by the forces of modernisation (Pedersen et al. 2007). Instead, the authors found rather high numbers of locals representing the constituency in which they were born or live. The country with the lowest share of DRL, though, was the United Kingdom (Pedersen et al. 2007, p. 169).
The most comprehensive historical analysis of the local roots of British MPs has been undertaken by Rush (1994Rush ( , 2001. According to him, in the nineteenth century local connections of Westminster MPs were the norm rather than an exception, especially in the Conservative Party. The numbers decreased considerably during the twentieth century, prompting Wheare (1963, p. 44) and Rush (1994, p. 575 Rush). This time, though, the surge of localism came from the Labour Party and from third parties, while the Conservative Party lagged well behind. Empirical evidence for Scotland shows above-average levels of prior local government experience of Scottish MPs between 1945 and the 1970s and a relative decline vis-à-vis non-Scottish MPs since then (Keating and Cairney 2006, p. 50). Another study disclosed that the Scottish contingent at Westminster has become increasingly Scottish since the mid-twentieth century, with the numbers of non-Scottish 'carpet-baggers' considerably decreasing (Keating 1989, p. 90). Measurements of a true territorial linkage to the 'local', i.e. the constituency level, however, are missing. To our knowledge, separate figures of localness for Welsh and Northern Irish MPs are missing completely.
There has been only one recent study on the territorial base of British parliamentarians with a UK-wide perspective and a clear focus on territorial variation across the UK. Starting from the UK standard regions (which include the three non-English constituent nations), Gandy (2014Gandy ( , 2018 calculates MPs' and MEPs' 'localism' by linking the standard region of their place of birth to the standard region in which their constituencies are to be found. Looking at the 2009 European Elections and the 2010 General Election he finds 'a great deal of politician mobility' (Gandy 2014, p. 207) but also quite significant numbers of MPs with a regional territorial base, as almost half of all UK MPs (45%) represent constituencies from within their region of birth. This share increases to almost three quarters (74%) when neighbouring regions are included. Apart from party, gender and age differences, Gandy found considerable territorial variation. The North-East of England together with the three constituent nations Northern Ireland, Wales and Scotland were the regions with the highest rates of regionally born MPs, while regions in the south and east of England were those with the highest geographical mobility (Gandy 2014, p. 193). A replication of this study with data from the 2014 European Election and the 2015 General Election revealed even higher levels of regionally born MPs, with the highest scoring regions remaining the same (Gandy 2018).
The literature discussed so far focuses primarily on the electoral outcome, i.e. stage 4 in Fig. 1. A further productive line of empirical inquiry into DRL is the study of voter preferences and voting behaviour, referring to stage 3. Research in this field generally supports the relevance of localness. In line with Key's (1949) famous friends and neighbours hypothesis, recent electoral studies corroborate advantages of local candidates for different electoral systems (for example, Tavits 2010; Górecki and Marsh 2012;McAllister 2015;Jankowski 2016;Blais and Daoust 2017). For British voters, this seems to be of particular importance. Survey data show that ceteris paribus a considerable share of the British prefer an MP to have been brought up in their constituency. While patterns vary across time, they tend to surpass those for other attributes such as party loyalty, business or trade union experience or being well educated (Johnson and Rosenblatt 2007). Other studies found a small but significant effect of the distance between voter and candidate residence at General Elections (2010 and2015) in England (Arzheimer and Evans 2012;Evans et al. 2017). Similar effects are found in experiment-based studies (Campbell and Cowley 2014). According to Campbell et al. (2019), this preference for local candidates seems to be largely based on a positive inference about their likely legislative performance. Providing further insight into the complexity of this mechanism, Middleton (2019) recently revealed in a study on post-retirement electoral contests that different forms of local ties (residence, schooling, local party position, local business) have different effects on voters, while it also depends on the type of electoral contest (e.g. inheritor vs challenger or government vs opposition party), whether these local ties are beneficial or detrimental for the candidate. From a territorial politics perspective, however, the most important insight comes from Collignon and Sajuria's (2018) study on the UK general election 2015. They revealed that, generally, voters with a strong regional identity are more likely to prefer local candidates. They also found that Scottish and Welsh identity had different effects than English identity and that in the former case the opportunity to express regional identity via a nationalist party (i.e. the SNP and Plaid Cymru) actually moderated the local preference effect.
To the best of our knowledge, studies about the role of candidates' localness in party nomination processes (stage 2 in Fig. 1) do not exist. The same is true for the question if local candidates are more or less likely to declare their willingness to run for office than non-locals (stage 1), though some studies have at least analysed the candidate fields of some elections. Arzheimer and Evans (2012) and Evans et al. (2017), for example, report relatively high levels of local candidates. In almost half of the constituencies examined they found at least two local candidates, while only 7% exhibited no local candidate at all. However, their research is focused on a very specific theoretical question. Their data are, thus, limited to England, the major UKwide parties, geographical voter-candidate distance as measurement of localness and two separate General Elections (2010 and2015). Studying the effect of localness on electoral campaigning Sällberg and Hansen (2020) restrict their territorial, chronological and party-political focus in a similar way (England; 2015 General Election; Labour and the Conservatives), while using MPs' projection of locality in official communication as an indicator of localness.
While our literature review shows an increasing scholarly occupation with DRL, it also reveals existing research gaps with regard to all stages of legislative recruitment. Independent of which stage is researched, two desiderata stand out.
First, studies meant to provide a comprehensive overview of DLR at Westminster have made little effort to conceptualise localness in a differentiated way. Political career studies have long used prior local government experience as an indicator of a local base, usually ignoring the crucial question of territorial congruence. Rush (2001, p. 204) goes beyond this simple conceptualisation, merging several quite explicitly local links ('being born, educated, living or working in the constituency, having property interests or serving or having served as a member of a local government body in the constituency') into an aggregate indicator. While this conceptualisation constitutes a major improvement, it seems to cast the net too widely as localness would here include 'non-locals' who have recently moved into the constituency or even absentee landlords-neither of which would stand the test of our concept of localness (see "The territorialisation of British politics and the local base of Westminster MPs: overview and theoretical starting points" section). Others base their studies on a single indicator like 'residence' in Evans (2012, 2014) and 'birth' in Field (1997) and Gandy (2014Gandy ( , 2018 and may, thus, miss out on candidates who do exhibit other meaningful local connections. Furthermore, Gandy (2014Gandy ( , 2018 somewhat surprisingly confines localness to the territorial boundaries of the UK's standard regions, completely ignoring the constituency as the formal base of territorial representation in Britain. The second issue concerns the variations in terms of DRL that are actually explored. Here, unsurprisingly, political party is the dominant category. Investigating variations across regions, only Gandy includes a territorial dimension in his study. In our view, though, the consideration of territorial variation is important. Regional effects can superpose or eliminate party effects (or vice versa) or interact with them. Furthermore, substantial territorial variation regarding the core principle of representation may be seen as another indicator of a much wider political divergence within the United Kingdom. By taking into account different types of local base and both party and territorial variation, our study sheds light on central aspects of DRL in the Westminster system that have so far been neglected.

A territorial politics perspective on DRL in Britain: expectations
To address the research desiderata identified so far, our study provides insights into the overall share of DRL and variations across territory, party and time. We assume the occurrence and frequency of DRL to be influenced by two major driving forces introduced in "The territorialisation of British politics and the local base of Westminster MPs: overview and theoretical starting points" section: the growing salience of territory in British politics and the increasing disintegration of its traditional two-party system. Both factors seem to influence legislative recruitment at different stages evoking two general expectations.
On the most general level, we expect to find rather substantive and possibly increasing shares of DRL among Westminster MPs. This expectation is based on the argument that the high salience of territory predominantly detected with regard to the regional level (especially with regard to the UK's constituent nations) might spill over to the local domain. Empirical evidence from Collignon and Sajuria (2018) to the effect that voters with a strong regional identity care significantly more about local representatives, strongly supports this assumption. An overall increase in regional identity might, thus, directly or indirectly affect all three stages of legislative recruitment as it might nudge both voters' (stage 3) and party-selectors' (stage 2) preferences towards local candidates and encourage a higher number of local candidates to come forward (stage 1). At the nomination stage (stage 2) the selection of local candidates might also be a strategic answer to the increasing challenges from populist and regionalist parties. This expectation is further supported by Britain's divergence from the two-party system since the 1970s. According to Key (1949, p. 38), '[i]n well-developed two-party situations, localism is minimised, if not erased, by a larger concern for party victory'. Conversely, de-alignment from the socio-economic cleavage and divergence from the binary logic of the two-party system may reinstate voters' 'natural' affinity to local candidates.
Second, and perhaps more importantly, we expect strong territorial variation with regard to the mode of territorial representation, i.e. DRL scores. The reason for this is that both driving forces-the political salience of territory and the divergence from the two-party format-are unevenly distributed across the UK's regions and constituent nations, but also across constituencies. Taking Collignon and Sajuria's and Key's theoretical insights and Gandy's empirical findings as our cues, we expect the three smaller constituent nations to show higher shares of localness than England. This goes specifically for Northern Ireland. Its completely distinct party system together with its strong antagonistic identities and a highly segregated political landscape suggest the highest levels of regional and local embeddedness. In Scotland, increasing nationalism and the dominance of the SNP also make for a high political salience of territory and long-term deviation from the Labour-Conservative duopoly and, thus, potentially higher shares of DRL. Wales also exhibits nationalism and party system divergence (a strong third party plus the sustained dominance of the Labour Party), though to a lesser extent. On the other hand, strong internal electoral divisions (Balsom 1985) might be a factor increasing Welsh DRL figures. England as the largest and political dominant British nation has traditionally been seen as the part of the UK where territory and national identity is least salient. We would, therefore, expect lower DRL scores than elsewhere. However, this assumption has to be qualified as Englishness and English nationalism have recently gained traction. A central trait of English national identity has been said to be the frequent reference to the local (Kenny 2014). And indeed, Cox and Jeffery (2014, pp. 5-6) show local identity in England to be the strongest of all territorial attachments, ahead of both English and British identity.
England, though, is far from homogeneous and we, thus, also expect territorial variation within England (cf. Gandy 2014Gandy , 2018. However, given their lack of distinct political institutions and identity (compared to the devolved nations), 3 we do not expect English standard regions to exert an independent and direct influence. Regional variation within England may instead follow major demographic differences between the regions, especially those regarding population density and geographical mobility. The growing 'bifurcation' of English politics, for example, might also be reflected in a higher salience of localness in largely rural, more 'inward looking' constituencies (Jennings and Stoker 2016). Similarly, constituencies with lower geographical mobility, i.e. lower immigration levels, might show a higher propensity to vote for local candidates (demand), as well as a higher supply of local candidates. Epitomised in the anomaly of the migration hotspot and highly populous metropolis London, both these considerations might point to rather low DRL levels in the English capital and much higher figures for the rural surrounding regions.
With regard to variations across parties, we expect our data to reveal patterns similar to those depicted by previous studies, which generally show Conservative MPs to be more geographically mobile, while Labour and third parties, especially nationalist and non-state wide parties, increasingly select local candidates (Rush 1994;Childs and Cowley 2011, p. 6;Gandy 2014, pp. 193, 196;Collignon and Sajuria 2018). However, our theoretical starting point of an increasing salience of territory suggests these party effects to be mediated by territory, and thus, we expect party patterns to vary across constituent nations and regions. 4

Data and operationalisation
For this study, we compiled biographical data on MPs from the House of Commons website supplemented by further trustworthy publicly available sources (for detailed information on sources see Online Appendix). This way, we could realise high utilisation rates above 90%. The full dataset consists of information on all 1108 MPs who served during the electoral periods of 2010-2015, 2015-2017, 2017-2019 and since 2019.
In our study we investigate the two dimensions of localness that have been set out in "The territorialisation of British politics and the local base of Westminster MPs: overview and theoretical starting points" section: local origin and local engagement. An MP of local origin would ideally be somebody who is born and raised in her constituency and has been living there all her life. As data on the duration of local residence are not available, we limit our set of indicators to an MPs' place of birth and place of school 5 (as a proxy for upbringing), both reflecting some long-term connection and arguably also some identity link. Place of birth or place of school was coded as local if it was inside the MP's constituency 6 or in the same city or town as the constituency. 7 This poses a considerable problem in London, where our definition for a match between constituency and place of birth or place of school extends to the whole of Greater London and, thus, to a much larger and more populous area than for other constituencies (thereby increasing the chance of such a match). We need to keep this in mind when interpreting our results.
Local engagement would ideally include any prior institutionalised civic or political activities of MPs in their constituency. This might include party office, mandates in local councils, positions in trade unions and other interest groups or honorary 5 More precisely, we use the schools in which MPs completed their secondary education. 6 Correspondence was checked using the constituency boundaries at election time. This might deprive MPs who had been elected before the last boundary review in 2010 of their 'localness' status, if their place of birth, school or local council area has been cut out of the new constituency they represent. However, a check of 15 MPs first elected before 2010 and lacking localness in 2010 showed not a single case that would have been classified as 'local' if we apply the old constituency boundaries. 7 The latter is mainly relevant for bigger cities including multiple constituencies. For example, the birthplace of the MP for Glasgow Cathcart can be anywhere in Glasgow to be coded as a local base. posts in voluntary organisations. Again, the nature of our data prohibits such a broad approach, as only local government service seems to be reported with reasonable consistency. Our sole measure for local engagement is, thus, a mandate in a local council whose jurisdictional borders at least overlap with constituency boundaries. As other valid forms of local engagements are not considered, this measure (as well as our combined measures, see below) systematically underrates local engagement and should, thus, be interpreted as the lower limit rather than an exact estimate.
In order to account for the wider multilayered territorial embeddedness of MPs, we also look at linkages to other territorial units. Thus, besides identifying a territorial match on the constituency level, we have also searched for congruence on the level of the four constituent nations (England, Scotland, Wales and Northern Ireland) and on the UK level (that is identifying MPs born or raised abroad). Geographical congruence on all levels (or the lack thereof) was established using the Ordnance Survey Election Maps (https:// www. ordna ncesu rvey. co. uk/ elect ion-maps/ gb/).
Besides information on the local base, we added information about an MP's party affiliation and the constituency she was elected for. The latter includes data about when the MP represented the constituency, in which constituent nation or English region the constituency is located as well as data on the constituencies' population density and migration levels (UK Parliament 2020). For clarity, our tables display only the figures for the four constituent nations plus London and 'England without London', as those territorial units comprise the most important results. Scores for the English regions can be found in the Online Appendix and will be referred to when necessary.
Of course, all MP-related data can change over time. MPs can, for example, leave or change their party, or they can run for different electoral districts at different points in time-with the consequence that they might be coded as 'local' in one case but not in the other. For this reason, our units of analysis are not individual MPs but MPs during a specified electoral period. This decision has the advantage of allowing for potential changes of constituencies or parties as each situation would be treated as a separate case. Hence, an MP who served from 2010 to 2020 would constitute four distinct cases. Furthermore, this definition allows for a more adequate comparison with studies that refer to single electoral periods as respective results give a more accurate reflection of the situation in parliament at any given time. 8 As a technical consequence, our number of cases increases considerably. The 1108 MPs of these four periods make for a total of 2627 cases. Percentage values in this kind of categorisation, thus, do not represent a particular share of all MPs but a share of all MPs per electoral period. To check the robustness of our findings, we duplicated all results shown in "Data and operationalisation" section on the basis of MPs. Overall, these results differ only marginally, as the tables in the Online Appendix demonstrate.

Overview
We start the results section with an overview of the different types of local base British MPs exhibit (Table 1). Overall, we find local bases in 41% of our cases. The largest contribution to this rate comes from local government experience in the elected district (and, thus, from local political engagement) which we observe in roughly 27% of cases. The shares of cases with MPs born (20%) or educated (19%) within a constituency are clearly lower. While the latter rates might slightly underestimate real numbers because of missing data (roughly 6% of birthplaces and 13% of places of school are missing), 9 a whole range of political offices at the local level (party office, interest group position, etc.) are excluded because of difficulties in systematically collecting data. Table 1 also shows the share of MPs with multiple facets of local base. While most MPs born in their constituency have also attended a local school (and vice versa, see the Online Appendix), exhibiting a local base by origin (place of birth or place of school) and by engagement (local government office) at the same time is a rather rare phenomenon. Taken together, 59% of all cases show no local base at all, 41% show some local base referring to either origin or engagement, and only about 13% show strong local roots that entail both local birth or local upbringing and local political engagement.
To get a better impression of British MPs' roots, Table 1 also provides more detailed information for those parliamentarians who are not born or raised in their constituency. This information relates to the question whether these MPs are born or schooled in the UK or abroad and, for the first case, whether this happened in the nation in which their district is located. The table shows that almost all Westminster MPs are born and raised in the UK, the large majority of them within their nation.

Across regions and nations
A look at the territorial variation across Britain's four nations (and London) mainly fulfils our expectations. All values clearly show that the selection and election of local candidates are highly contingent on the location of the constituency in question. This applies to combined as well as individual measures. In Northern Ireland and Scotland, MPs tend to have a local base: 65% (Scotland) and 74% (Northern Ireland) of cases show at least one type of local base. The exceptionally high value in Northern Ireland has even been reached despite a high share of missing data (almost a third of school places could not be traced). In Scotland, this local base is more likely to be one of local origin (41% born in constituency, 41% schooled in constituency) than of local engagement (33% former local government office), while Northern Ireland features by far the highest ratio (58%) of former local councillors turned local constituency MP. These figures contrast strongly with average values for Wales (42% any local base) and rather low values for England excluding London (EeL) (31% any local base). In EeL only 14% of constituencies are represented by MPs who were born in their own constituency and only 13% are represented by MPs schooled in their constituency. Although figures for local government experience in the constituency are much higher (23%), EeL is clearly the country of the carpet-bagger. The lowest rate for former local councillors (only 15%), is to be found in Wales.
Surprisingly, figures for London are much closer to those for Scotland and Northern Ireland than to those for surrounding England. Roughly a third of London's constituencies exhibit MPs born in London. The same goes for MPs schooled in the capital. These values, though, have to be interpreted with caution, as our coding scheme recognises the whole of Greater London as a single locality (see "Data and operationalisation" section). However, figures for prior local government service in general (63%) and for local government experience with borough council area and constituency overlap (50%) are also much higher than in the rest of England, suggesting that the stronger local base of MPs from London constituencies (70% any local base) is more than a methodological artefact.
As expected, there is also a fairly large variation across the other English regions, with a range from 13.7% (East of England) to 53.4% (North-East) for any local base (see Online Appendix, Table A2). This result is in line with Gandy's (2014Gandy's ( , 2018 empirical findings, yet does not correspond to the fairly equally distributed sense of local identity (Cox and Jeffery 2014, pp. 5-6). In contrast to our expectations, there is also no clear correspondence with the two demographic variables we expected to condition MPs' local base: English constituencies with lower population density are hardly more (or less) likely to elect local candidates and neither are English constituencies with lower migration rates. 10 It is remarkable, though, that despite this enormous (and difficult to interpret) intra-English variation no English region (except London) comes even close to the much higher scores in Northern Ireland, Scotland and London.
Broadening our analysis of MPs' territorial linkages from the local, to the national and finally the UK level reveals a strong territorial rootedness of MPs in their respective nation (England, Scotland, Wales, Northern Ireland). A mere 13% of all constituencies are represented by MPs born outside the nation in which their constituency is located. Values for the place of school are even lower (8%). Despite the unitary character of the UK parliamentary system, political career mobility across internal national boundaries is, thus, rather limited. As expected, this pattern is particularly pronounced in Northern Ireland (only 4% are born and 1% schooled outside Northern Ireland), followed by Scotland (7% and 3%) and England (13% and 8%). Wales, however, constitutes a marked exception to this rule. Almost a third of its constituencies have MPs who are born outside Wales (mostly in England) with roughly the same share schooled outside Wales. This high value reflects the rather high share of English-born people among the general population in Wales (21% in 2011). It also contradicts Childs and Cowley's (2011, p. 7) claim that hardly any English MPs sit for Welsh seats.
Unsurprisingly, MPs' mobility across the UK border is even less frequent. Variation between the constituent nations is low, ranging from 1 to 5% (MPs born outside the UK) and 0 to 2% (MPs schooled outside the UK). Only the constituencies in highly globalised London feature 10% of MPs born outside the UK, with PM Boris Johnson the most prominent of them. A look at their place of school, though, shows only very few (2%) to be actually brought up abroad.

Across parties and time
A look at variations across parties (Table 2) generally confirms previous findings. The highest localness scores can be found for nationalist and non-state-wide parties (SNP, DUP, UUP, SF and SDLP, though not PC), followed by Labour and the Liberal Democrats, while the Conservatives show strikingly low values. Only in a quarter of Conservative seats do we find MPs who have at least one form of local base. Only 11% of Tories elected to Westminster are elected in the constituency in which they were born, only 9% in the constituency in which they went to school. A Tory MP with previous experience in a local council within the constituency is to be found in 19% of all cases.
The shares for Labour and the Liberal Democrats are considerably higher. Roughly half of their cases feature some local base. While Labour exhibits rather similar individual values for all forms of local base (at around 30%), Liberal Democrat MPs are more likely to have served in the local council (36%) than to have attended a local school (19%). Only 11% of all cases (and, thus, almost as few as in the Conservative Party) show Liberal Democrat MPs born in their constituency.
Most frequent local links are to be found in Scottish and Northern Irish parties. 11 The SNP features high values for any local base (roughly two thirds of its MPs) including comparably high numbers for place of birth (43%) and place of school (39%). The local government experience of SNP MPs (34%), however, is slightly lower than that of Liberal Democrats. The small number of cases and the high rates of missing information (specifically for Sinn Fein) make the interpretation of the situation in Northern Ireland more difficult. What seems clear, though, is that on both sides of the divide, some form of local base is the norm rather than the exception. Despite its high ratio of missing data, Sinn Fein actually features the staggering value of 92% for the overall measure, while the DUP has 59%. Furthermore, these two parties feature by far the highest values of all parties for prior local government service (59% and 54%). In Northern Ireland, service on the local council still seems to be an important prerequisite to winning the nomination for a Westminster seat. While the two rivals show no difference in this respect, they are at opposite ends when turning to place of birth. In more than half of Sinn Fein constituencies (54%), MPs are born locally, while the DUP ratio for this indicator is around 9%. 12 Broadening the view to the question of territorial embeddedness within the constituent nation, party-political variations are less pronounced. Overall, though, MPs from all-British parties are slightly more likely to be born outside of the nation in which their constituency is located than MPs from non-state-wide parties. In 14% of the cases Labour MPs are either born in another UK nation (9%) or outside the UK (5%). The same figure applies to the Conservatives (14%), while the Liberal Democrats feature 11%. Values for nationalist and non-state-wide parties are generally lower. The DUP contingent is without any MP born outside Northern Ireland, Sinn Fein exhibits two such MPs (8%) and the SNP also shows a ratio of 8%.
This difference is considerably more pronounced regarding place of school. 7% of Conservative cases, 8% of Labour cases and an outstanding 18% of all Liberal Democrat cases show MPs not raised in the nation in which they stood for election. This picture contrasts with the situation for Sinn Fein and the DUP, where not a single MP was raised outside Northern Ireland and for the SNP, which exhibits a share of 2% of their cases not raised in Scotland. Overall, these results suggest that for non-state wide parties an orientation to the nation is more important than for all-British parties.
In line with the increasing territorialisation of British politics, we expected DRL figures to rise across time. For the sake of readability, we restrict our analysis to the values for 'any local base' here and in the following. Indeed, Table 3 shows a slow yet continuous increase of MPs with any kind of local base, starting at 38% in the 2010-2015 period and reaching 43% in the period starting 2019. This tendency seems to be driven by both major parties, as Labour's steady increase perfectly reflects the overall development, while figures for the Conservative's receded slightly in 2017 before rising again in 2019. Numbers for the smaller parties, though, are less consistent (SNP) or even show a decrease (Liberal Democrats).
The general rise since 2010 is perhaps best reflected in the fact that for each electoral cohort-and indeed for both major parties-the DRL share of newly elected MPs exceeds that of incumbent MPs (see Online Appendix, Table A4). Interestingly, though, the DRL share of newly elected MPs in 2019 is below that of 2017. This clearly shows the party effect, as the largest intake of new MPs in 2017 were Labour MPs (56) scoring 66.1%, while in 2019 the largest group of newcomers came from the Conservatives (117 MPs-35.9%).

Territorial and/or party effect
The party-political variation with respect to the local base of MPs in conjunction with the diverse electoral support of each party in the different nations and regions of the UK implies that variation across regions/constituent nations (Table 1) and variation across parties (Table 2) cannot be treated as independent of each other. Perhaps the different local base scores in these nations and regions are simply a function of the diverse party-political support in each part of the UK. The question is whether we should interpret the observed variations as the result of a party or territorial (national/regional) effect. To answer this question, we cross-tabulate the territorial and party results of the UK-wide parties (Table 4). If a party effect is dominant, individual party values should be quite similar across regions and nations. If this constitutes a territorial effect, we would expect similar values in each region and nation across parties.
A look at Table 4 shows both territorial and party effects at work. The party effect becomes apparent in the columns of the table. In all territorial units shown here, the party with the lowest DRL score and that with the highest differ by more than 20 percentage points-except for Wales where the difference is slightly lower and where patterns should be interpreted with reluctance because of very small case numbers for the Liberals. Importantly, the parties with highest or lowest scores vary across the territorial units. While in London and in Scotland, the highest values can be found for Labour, maximum DRL scores in EeL go to the Liberals and in Wales to the Tories. Table A3 in the Online Appendix shows the respective values for the English standard regions and confirms both strong variation within regions and no clear patterns across regions.
Territorial effects on party's local base figures are even more apparent. A look at Table 4 shows that the highest range across the British nations can be found for the Conservative Party which scores a mere 25% in England (22% in EeL) yet 67% in Scotland, reflecting a total range of 42 (45) percentage points. The range of Labour is clearly lower but still huge with 41% in Wales, 68% in Scotland and 76% in London. The Liberal Democrats show comparably constant values roughly around 50%, if we disregard the small case numbers in Wales. Table A3 (Online Appendix) corroborates higher regional variation for the Conservatives than for Labour. Regional numbers for the Liberals are too small for a meaningful interpretation.

Conclusion
Our study of the local base of British MPs provides a comprehensive overview of the patterns and variations of DRL across party, across region and nation and across time. While its approach remains mainly descriptive, the results are far from selfevident. First, we can show that the British parliamentary system's core principle of territorial representation is indeed empirically underpinned by the selection and election of local candidates. Since 2010 in more than 41% of cases constituencies have been represented by MPs who have some meaningful local base in the constituency they represent-either by origin or by local engagement. Thus, both normative claims deriving from democratic theory and empirical demands of the British public to this effect can be seen to be met-at least to an extent. Whether this share should be considered particularly high or low is difficult to tell, as we lack international comparison. The growing political salience of territory that triggered our study might have suggested even higher figures, but then we should not underestimate the number and significance of possible trade-offs (gender, social class, ideology, etc.) that might inhibit the nomination and/or election of locals.
However, taking a diachronic perspective, our data point to an ongoing trend towards increased DRL shares. While our figures are slightly lower than Rush's results at the turn of the century (45-48%), they are based on an idea of localness that is deliberately centred around strong and long-term links rather than weaker indicators like residence or property interests. Furthermore, missing data mean that our numbers have to be seen as lower bounds rather than exact values. Finally, our longitudinal analysis shows a slow yet steady increase from 2010 to 2019, despite some sensitivity as regards varying party strengths.
The latter are noteworthy, since the share of local MPs varies tremendously across political parties. For Conservative MPs a local base is still a rather rare characteristic, while Labour, the Liberal Democrats and especially nationalist and nonstate-wide parties exhibit considerably higher shares of local MPs. In the latter case, the high salience attached to their national base seems to translate also into a high relevancy of a meaningful local base for their MPs. While party clearly matters for the likelihood of getting local MPs elected, our study has shown the party effect to be highly contingent on other factors, most notably territory.
Territorial effects can be seen in the huge variations across English regions and across the constituent nations of the UK. The most striking result in this respect is the immense gap between rather low DRL scores for EeL, and to a lesser extent Wales, and the much higher rates for Scotland, Northern Ireland and London. While the exceptional status of Northern Ireland with its distinct political culture and party system is hardly surprising, the enormous difference between England and Scotland and the special case of London are in need of explanation. It is important to note that the much higher scores in Scotland are definitely not due to a party effect, as SNP, Labour and most notably the Conservatives produce almost identical DRL rates in Scotland. The much lower DRL figures for England are partly the result of a party effect, i.e. the much higher number of seats for the party with a lowest DRL score (the Conservatives), but the fact remains that in England (except for London) both major parties show much lower DRL scores than in Scotland.
A more detailed causal analysis of DRL, which is beyond the scope of this paper, would have to include a whole range of supply and demand side factors on all stages of parliamentary recruitment (see Fig. 1). This regards in particular constituencylevel political variables such as marginality, incumbency etc., which are highly likely to condition the strategic choices of would-be candidates as well of party selectors and voters.
Our study, however, suggests that even the effect of these factors would be territorially mediated. Given the enormous variation across the four constituent nations of the UK, it is rather unlikely that constituency-level factors can account for these differences. Instead, in line with studies on electoral and party system divergence (Awan Scully 2018; Stolz 2019) we maintain that the territorialisation of UK politics means that Scotland (in addition to Northern Ireland) and England (and to a lesser extent Wales) have become distinct polities, where the same individual variables might have rather different effects. Conditioned by living and practising politics in these polities, potential candidates, party selectors and/or voters seem to develop quite diverse understandings of the principle of territorial representation that cannot be explained with reference to single isolated variables.
Reflecting the exploratory nature of this investigation our results are still rather provisional. We see our study as a starting point for a more comprehensive account of DRL in the UK that would not only have to test more and other potential causal variables, but should also focus on the mechanics of the recruitment of locally based MPs. To grasp underlying differences in the understanding of localness and the salience of DRL among key actors, studies would have to include surveys, but also in-depth interviews with candidates (including losing candidates), party selectors and voters as well as close observations of nomination processes. This article also stops short of identifying concrete consequences of the detected variations. Effects might be expected on an individual level, i.e. MPs' legislative behaviour and career maintenance, but also on an aggregate level, i.e. parties' public appeal or a region's/ nation's (self-)image.
Finally, our findings are casting a new light on the current constitutional debate about the United Kingdom's territorial structure. Regarding the rivalling interpretations of the UK as a unitary state composed of 650 constituencies or as a union state composed of four constituent nations, our analysis rather corroborates the latter view. First, we observe low levels of MPs' movements across national boundaries, in particular in Northern Ireland and Scotland. Secondly, the immense gap between England and Scotland/Northern Ireland with regard to DRL reveals that the political divergence between the UK's constituent nations does not even stop at the implementation of its core representational principle. While this in itself does not constitute a constitutional problem, it might nevertheless be seen as another indication for the deeply divided United Kingdom. Furthermore, we might also interpret the comparably low levels of DRL in England as a failure of the two major British parties to connect meaningfully with English voters, thus, adding to grievance and political disenchantment in the dominant nation of the UK and in the nation that has long been regarded as its sole 'stable and secure heartland' (Kenny 2014, p. 232).