MPs' Localness

Multiple studies have shown bias in MPs' responsiveness to citizens based on citizens' socio-demographic backgrounds (Butler Reference Butler2014; Butler and Broockman Reference Butler and Broockman2011; Costa Reference Costa2017; Dinesen, Dahl, and Schiøler Reference Dinesen, Dahl and Schiøler2021; Grohs, Adam, and Knill Reference Grohs, Adam and Knill2016; Habel and Birch Reference Habel and Birch2019) and political traits (Butler and Broockman Reference Butler and Broockman2011; Gell-Redman et al. Reference Gell-Redman2018; Rhinehart Reference Rhinehart2020). For example, elected officials in several polities have been found to be less responsive to working-class and ethnic minority constituents and more responsive to ‘co-partisans’. Furthermore, a ‘friends and neighbours’ effect has long been observed in multiple democracies where voters prefer ‘local’ politicians – those born in the constituency they representFootnote ¹ (Arzheimer and Evans Reference Arzheimer and Evans2012; Blais et al. Reference Blais2003; Gallagher Reference Gallagher1980; Górecki and Marsh Reference Górecki and Marsh2012; Key Reference Key1949; Lewis-Beck and Rice Reference Lewis-Beck and Rice1983).

The electoral advantage enjoyed by local MPs can be partially explained by behavioural localism; for example, the expectation that MPs with strong local connections will be more likely to prioritize the interests of the local area above those of the party or the nation (Campbell et al. Reference Campbell2019; Schulte-Cloos and Bauer Reference Schulte-Cloos and Bauer2023). Voters' preference towards local MPs may also reflect citizens' desires to express their place-based identity based on homophily (the tendency for individuals to connect with similar others) (Campbell et al. Reference Campbell2019; Schulte-Cloos and Bauer Reference Schulte-Cloos and Bauer2023).

Extant research provides some evidence that the extent to which politicians attempt to signal their constituency focus is strategically targeted. MPs representing more marginal constituencies prioritise their constituency duties more than those in safer seats (Campbell and Lovenduski Reference Campbell and Lovenduski2015; Sällberg and Hansen Reference Sällberg and Hansen2020). While there is extensive research in the literature on legislators' attempts to cultivate a ‘personal vote’ by signalling their commitment to the interests of the area they represent (Cain, Ferejohn, and Fiorina Reference Cain, Ferejohn and Fiorina1987; Zittel Reference Zittel, Arzheimer, Evans and Lewis-Beck2017), less attention has been paid to the role localism plays in mediated MPs' direct relationship with their constituents.

Transitioning from voters' preference for politicians with local connections, this letter explores how MPs' localness shapes their responsiveness to their constituents. Research has demonstrated that legislator responsiveness is influenced by the strategic incentives set by the electoral system; responsiveness is higher in majoritarian than in proportional representation systems (Breunig, Grossman, and Hänni Reference Breunig, Grossman and Hänni2022), where there are greater opportunities to cultivate a personal vote (Cain, Ferejohn, and Fiorina Reference Cain, Ferejohn and Fiorina1987). MPs' localness may influence their behaviour for strategic reasons to maximize their chances of re-election, from a psychological perspective or as a result of homophily, or both. From the perspective of homophily, we might expect MPs with strong local connections to be motivated by a sense of connection with their community and be more responsive than other MPs. However, we have seen that an electoral preference for local politicians is so well established that it is ‘bordering on banality’ (Pedersen, Kjaer, and Eliassen Reference Pedersen, Kjaer, Eliassen, Cotta and Best2007). Here, we speculate that, from a strategic perspective, perhaps the local advantage is so profound that MPs with strong local connections can better afford to shift their attention away from constituency work to promote their parliamentary career. Local MPs may be less responsive overall because they estimate that the electoral benefit their localism provides reduces the potential gains from responsiveness. Thus, our expectations could run in either direction. From the perspective of homophily, we may expect local MPs to be more responsive. Still, from a strategic approach, we may expect lower levels of responsiveness from local MPs, all else being equal. Alternatively, local MPs may not be less responsive overall, but they make no distinction between constituents from different backgrounds based on homophily, feeling that they share a local place-based identity with all their constituents and thus responding to all equally.

Regarding non-local MPs, strategically motivated ones may be responsive to all constituents, irrespective of their backgrounds, to signal behavioural localism. However, being hyper-responsive to constituents to convey behavioural localism might only partially remedy their disadvantage, as research demonstrates that there are gains for local connections beyond those delivered by assumed behavioural localism. Hence, given the wealth of evidence that local MPs have an electoral advantage, it is probable that non-local MPs will ‘play up’ other traits (André, Depauw, and Deschouwer Reference André, Depauw and Deschouwer2014, 905) to foster alternative mechanisms for securing voters' loyalty. In the absence of place-based identity, we consider whether non-local MPs may utilize other aspects of social identity to ensure a connection with voters. Fostering relationships driven by homophily, beyond place-based identity, may provide non-local MPs with an opportunity to develop and maintain support and build an incumbency advantage among specific sub-groups of voters. Thus, we explore whether non-local MPs are more responsive to constituents with whom they share non-local attributes than those with strong local connections.

There is a vast literature investigating the substantive representation of women, demonstrating that women legislators are more likely to attempt to substantively represent the interests of women voters (Franceschet and Piscopo Reference Franceschet and Piscopo2008; Schwindt-Bayer and Mishler Reference Schwindt-Bayer and Mishler2005). We hypothesize that women MPs may be more likely to seek the votes of women voters when they do not benefit from the electoral advantage of a local connection. There is also some research that MPs may be more responsive to their co-partisans; again, a sense of shared identity (homophily) or a strategic incentive to mobilize the base could be motivational factors (Schakel et al. Reference Schakel2024). For example, non-local women MPs may be more likely to respond to women constituents, and non-local MPs may be more likely to respond to co-partisans.

Research Design

We used original data from an audit experiment conducted in the United Kingdom from 2 November 2020 to 18 December 2020.Footnote ² For a discussion of the ethical implications of conducting audit studies of MPs, see Zittel et al. Reference Zittel2023. As previously noted, studies have shown the importance localness plays in electoral politics. This is especially true in the UK, where perceived candidate localness is either the most important or the second most important feature voters find desirable in their MPs (Johnson and Rosenblatt Reference Johnson and Rosenblatt2007). The United Kingdom is an excellent base to explore the relationship between MP's localness and responsiveness. The electoral system (single-member majoritarian) encourages strong linkages between representatives and constituencies. The issue of candidate localness has grown in political significance, and there has been an increase in MPs with a direct constituency connection across the major parties (Cowley, Gandy, and Foster Reference Cowley, Gandy and Foster2022). As local orientation is such a critical aspect of British politics, we should expect the localness of British MPs to influence their behaviour towards constituents. Thus, the UK is a likely case to explore the inverse of the traditional ‘friends and neighbours’ effect and see if MPs’ behaviour towards their constituents is shaped by their localness.

The audit experimentFootnote ³ involved sending policy queries to legislators via emails where fictitious constituents varied according to ethnicity, gender, class, status, and partisanship. This experimental design enabled us to observe whether local and non-local MPs respond differently depending on the constituents’ socio-demographics and political identities and whether their response level is contingent on how similar the constituents are to themselves. One significant limitation of our design, common in audit experiments with MPs, is that responses to our emails may have been managed by staff rather than the MPs themselves, making it empirically challenging to distinguish between these responses (staff may sign off documents using MPs' names). Using MPs' official email addresses suggests that staff responded on behalf of their MPs, considering their MPs' characteristics.Footnote ⁴ The emails (see Supplementary Material A) asked how legislators and their parties would respond to the economic and social impact of the COVID-19 pandemic. We chose this issue because it was relevant and salient to all MPs and parties at the time of the study. For ethical reasons, we kept the email concise, with an open-ended question to avoid sending signals to MPs that could sway their reply.Footnote ⁵

We used a 2 × 4 factorial design that simultaneously manipulated these socio-demographic and political features.Footnote ⁶ This allowed us to test how various constituents' characteristics affect local versus non-local MPs' response rates without reducing statistical power or requiring more emails to be sent to MPs. To increase the power of the study, the emails were sent out in two waves, meaning legislators received two short emails with a random combination of socio-demographic and political treatments. We minimized the risk of detection by waiting at least two weeks before submitting the second email for each MP and by having two versions of the email that differed in wording but not in substance.

Data and Method

Empirical Strategy

We pool observations across waves, which leads to an overall number of 944 observations.Footnote ¹⁵ Our effects are first estimated with a standard regression with the various measures of localness in the following equation:

(1)$$Y_i = \alpha _i + \beta \;{\rm Localness}_i + {\rm Constituents}^{\prime}\;{\rm Cues}_i + Z_i + \varepsilon _i$$

We then estimated our two-way interactions between the localness variable and the treatment factors in the following equation:

(2)$$Y = \alpha _i + \beta \;{\rm Localness}_i + \beta \;{\rm Localness}_i\;\times \;{\rm Constituents}^{\prime}\;{\rm Cues}_i + Z_i + \varepsilon _i$$

where Y is the outcome variable mentioned above. We use logistic regressions for the response rate with standard errors that are clustered by MPs. α is the constant and ɛ is the error term. Z represents the various covariates, which include the wave, version and names of constituents, MPs' personal features and the contextual variables. In Supplemental Material F, we test different operationalisations of the dependent and independent variables and various modelling strategies to bolster the robustness of our findings.