Local economic decline has been presented as an explanation for populism, political alienation and geographic polarisation (Rodríguez-Pose, 2018; Broz et al., 2021; Carreras et al., 2019; Carreras, 2019; Greve et al., 2023). The economic vote is also known to encompass localised economic evaluations (Johnston and Pattie, 2002; Rogers, 2014; Healy & Lenz, 2017; Reeves & Gimpel, 2012), as well as national and household economics. Yet we lack theory and evidence about who is likely to perceive localised economic decline. Understanding the missing link connecting local economic decline and perceptions is the goal of this article, using novel survey data designed for that purpose from Great Britain.

‘Voters’ perceptions of economic realities… are more important than the realities themselves’ (Sanders, 2000: 290). However, perceiving national economic performance is not a straightforward exercise, and depends upon the clarity of signal (Chzhen et al., 2014). The political implications of local decline should likewise be rooted in perceptions, but we expect perceiving localised economic decline to be challenging, involving some assessment of the past and the present in a certain location, and the direction of travel over time. In this article, we ask: who perceives local economic decline? We offer five models that each focus on mechanisms to explain who is likely to perceive decline, and why.

Our results show that decline perceptions are heterogeneous. We reveal a story of personal economic factors, perceptions of the national economy, perceived increases in immigration, a strong relationship with partisanship, and greater likelihood to perceive decline among women. Our results highlight the importance of direct experience, heuristic reasoning, cultural resentment and cued partisanship, and less support for the expectation that nostalgia, or collective memory, is associated with greater perceived decline for older citizens. We do find correspondence between the measured objective reality of economic change in people’s local area, and their perceptions of decline. However, this relationship is substantially conditioned by a respondent’s partisanship. Specifically, in the case of Great Britain, Labour partisans’ perceptions (the party out of power in this period) of the local economy correspond to the economy measured using objective data, whereas this is not the case for Conservative partisans (supporters of the party in power). Partisanship also has a direct influence on decline perceptions. To the extent, then, that there is a known relationship between partisanship and evaluations of the national and personal economy (Wlezien et al., 1997; De Boef and Kellstedt, 2004; Evans and Andersen, 2006; Gerber & Huber, 2009, 2010; Pickup and Evans, 2013), our findings extend that observation to perceptions of localised economic decline.

The heterogeneity in local economic decline perceptions that we reveal, and their accuracy at the local level in relation to partisanship, offers useful insights into the possible limits and contingencies of localised economic effects. There remain analytical questions about the causal role of economic decline on political outcomes such as populism, or the economic vote. At a minimum, we should expect the updating of political preferences based on local economic experience to, at least in part, be conditioned by who forms perceptions of decline; partisans of the out-party, and those with worse personal economic experiences, evaluations of the national economy, and perceptions about rising immigration. Disentangling these bundles of perceptions would be useful in work that seeks to isolate effects of localised economic decline. Overall, we provide new insights into the variability of perceptions of localised economic decline and the potential for some citizens to be more or less responsive to economic decline. In so doing, we address a gap in the literature on localised economic decline; its relationship to the economic vote and its relationship to populism, political alienation and geographic polarisation; namely, the link between objectively measured decline and citizens’ subjective perceptions of decline.

Local Economic Decline: From Applications to Expectations

Decline is central to many accounts of political alienation and geographic polarisation. According to Elchardus and Spruyt (2016: 125–6), populism ‘is in the first place a consequence of declinism…The decline, perceived by many [our emphasis], is blamed on the establishment politicians.’ People come to take a negative view of changes in society, with some researchers putting more focus on economic change (Rodríguez-Pose, 2018; Broz et al., 2021), others on cultural change (Norris and Inglehart., 2019) and others a mix of the two (Carreras et al., 2019; Carreras, 2019; Greve et al., 2023). Rodríguez-Pose (2018) explains that populism took hold not among the poorest in society, but in a combination of poor regions and areas that had suffered long periods of decline. It has thus been the ‘places that don’t matter’, not the ‘people that don’t matter’, that have reacted. Sources of local economic grievance may be through a loss of manufacturing (Baccini and Weymouth, 2021), fewer job opportunities for younger generations (Zagórski et al., 2021), income stagnation (Carreras et al., 2019), import shocks and exposure to international trade (Colantone & Stanig, 2018; Dorn et al., 2020; Steiner and Harms, 2021), automation (Frey et al., 2018), austerity (Fetzer, 2019), loss of social amenities (Bolet, 2021), social status (Gidron and Hall, 2017; Carella and Ford, 2020; Kurer, 2020), nostalgia (Gest et al., 2018; Steenvoorden and Harteveld, 2018), beliefs that things could only get better (Carreras, 2019) and perceptions of gender discrimination through unemployment (Green and Shorrocks, 2021).

Localised decline builds a sense of alienation with political elites that have failed to represent people’s concerns and areas (McKay, 2019), resentment of more prosperous communities, groups, cities and other places (Cramer 2016; Green, Hellwig and Fieldhouse, 2022; Jacobs and Munis, 2022), heightened concern about immigration (Carreras et al., 2019) and, ultimately, a desire for political disruption. It has led to growing geographic polarisation between metropolitan areas and former industrial areas (Rickard, 2020; Rodden, 2010; Jennings and Stoker, 2019), and this resulting geographic polarisation maps onto support for populist causes (Ford & Goodwin 2014; Mudde, 2016).

Local economic decline is also linked to punishment of the incumbent government, through a socio-tropic economic vote (Johnston and Pattie, 2002; Rogers, 2014; Healy & Lenz, 2017). Voters evaluate the local economy alongside their own personal economic conditions, and those of the national economy. One mechanism through which socio-tropic concerns may affect vote choice is through greater understanding of the experiences of people similar to oneself, which – alongside a relationship with vote choice – inform perceptions of the national economy (Ansolabehere et al., 2014; Reeves & Gimpel, 2012). Local economic conditions have been shown to be related to reward or punishment of local government (Dassonneville et al., 2016), depending, in the US, on the state structure of the local economy (Ebeid & Rodden, 2006). Local economic conditions may even be more important sources of information about the national economy for those voters who pay relatively little attention to national economic indicators and national economic news, and to independents (Reeves & Gimpel, 2012).

It is highly likely that where it occurs, localised economic decline is indeed perceived by many. But it is also self-evident that not everyone will be similarly likely to perceive localised economic decline, and, as a result, some citizens may become more alienated or aggrieved, and others less so. This does not mean that a person could not be aggrieved without also holding clear and reportable perceptions of decline (that could be measured in a survey), but we should expect a stronger link between local conditions and behaviour for those who report perceptions in the first place; for those people who hold some assessment of the local economy over time.

Our starting point is that perceiving local economic decline is inherently difficult. These difficulties relate to: (i) cognition and recall – whether people are aware of prior economic conditions and conditions now, and specifically how those relate to a person’s experiences, their likelihood to gather information and context; (ii) separability – whether respondents derive perceptions of their local area from their own economic circumstances and other attitudes and evaluations; and (iii) endogeneity – whether, as is well-observed in national economic evaluations (e.g. Wlezien et al., 1997; De Boef and Kellstedt, 2004; Evans and Andersen, 2006; Gerber & Huber, 2009, 2010; Pickup and Evans, 2013), people evaluate their present local economic conditions and any decline on the basis of evaluations of the responsible political actors over that period, associated with their political preferences.

A further reason to focus on perceptions is that it is challenging to select an objective contextual measure that corresponds to what a voter might perceive as their relevant ‘local economy’. Such communities will be highly diverse and will not overlap with units of analysis such as electoral districts, local authorities, or regions (Wong, 2007; Wong et al., 2012). Analysis of objective contextual effects on political attitudes and electoral outcomes may therefore benefit from a comparison to subjective perceptions of local economics.

Who forms Perceptions of Local Economic Change?

Before we theorise who perceives decline (negative change over time, as opposed to positive), we can consider who is most likely to have any measurable opinion about the local economic situation currently and in the past; who answers survey questions about the local economy.

Local economic assessments may elicit specific sources of variation in survey response. One obvious way people could form assessments of economic decline is greater time spent living in an area. That may be associated with knowing more people who live there, having friends and relations in the area, knowing the impact of economic changes on different occupations, house prices, industry, etc. This relates both to how long a person has lived in the same place and to their age. A similar expectation relates to how embedded a person is in their local community. A predictor of such embeddedness, in addition to longevity and age, would be whether someone regularly uses amenities and services, such as local schools, nurseries, libraries, and health services. We would therefore expect caring responsibilities and home ownership to be associated with holding opinions about the local economy, alongside age and the length of time someone has lived in a local area.

A further set of expectations can be derived from existing generalised research suggesting that men (Mondak & Anderson, 2004), respondents with higher levels of political attention (Sturgis and Smith, 2010), information, education or newspaper readership (de Geus & Green, 2021) are more likely to report answers or certainty to political survey questions. This does not mean these respondents are more accurate in their responses, but they are more likely to express opinions. We would therefore expect these same demographics and habits of political news consumption, to be associated with providing any assessment of local economics.

Who Perceives Decline (as opposed to growth)?

To theorise the mechanisms associated with perceiving decline, we first present the implicit prevailing model of perceived decline, which we label ‘direct experience’ (see Table 1), based on the assumption that people notice the economic decline in their local area, directly through their own experience or through observing the economic situation of others. A strongly articulated example is provided by Broz et al., (2021:2), who write that the ‘direct impact on individuals who lose their jobs or have their wages cut initiates a more widespread impact on local communities. Jobs and income decline, property values fall, the local tax base erodes, more educated residents leave, and local public services deteriorate.’ In this way, because the impact is diffused, the facts of decline become virtually inescapable. A minimal expectation of this model would be that long-term changes in local economic contexts – measured at a range of geographic scales and using diverse indicators – would have a clear relationship with subjective perceptions of decline.

Table 1 Perception of local economic decline: five models
Full size table

Our first additional model of decline perceptions is one of heuristic reasoning. We assume that in all but the most obvious cases, there is considerable uncertainty over whether and to what degree a community is in decline and, as such, voters turn to heuristics. Chief among these, we expect, is drawing from other economic perceptions. On the one hand, voters may generalise from their own powerful experiences (Malmendier, 2021), possibly owing to some form of ‘projection’ or ‘false consensus’ bias: the tendency to believe that other people have similar attitudes, values, beliefs, and perceptions to their own (Krueger, 1998). On the other hand, people may generalise from perceptions about the national economy. This could be rooted in an ‘availability’ bias (Tversky & Kahneman, 1973) to generalise information which may be more accessible (recalled from memory) and more frequently encountered (such as in news media). The testable implication of the heuristic reasoning model is that standard survey measures of retrospective personal and national economic perceptions will predict perceptions of local economic decline.Footnote 1

The next set of explanations recognise the force of economic decline but see it as operating through the impact of collective memory, and nostalgia for lost glories, or lost social hierarchies. Here these models (collective memory and cultural resentment) are not competing models, but they are associated pathways. It is the underlying argument offered by Rodríguez-Pose (2018): ‘The areas left behind, those having witnessed long-periods of decline, migration and brain drain, those that have seen better times and remember them with nostalgia’. The idea of nostalgia capturing and shaping a response to economic decline is shared in other analyses too (Gest et al., 2018; Steenvoorden and Harteveld, 2018) and further developed by Greve et al. (2023: 412): ‘the mechanism leading to current voting behavior may be a place-based collective memory of past economic success, leadership, and economic well‐being, compared with a less favorable current situation…a social construct that reflects collective histories, memories, and identities’. This kind of economic nostalgia would, we expect, be felt most strongly among older people (Richards et al., 2020). It could be related to a nostalgia for a prior social status hierarchy related to race, employment and industry, (Gest et al., 2018), and is therefore related to our next model.

The fourth set of explanations finds expression through the visible identification of an out-group (‘cultural resentment’), and associated identification with white in-groups (Jardina, 2021). The literature strongly points to a relationship between perceptions of rising immigration and disaffection about the local economic situation (Carreras et al., 2019), even if those perceptions are acting as a source of bias. As alluded above, this may relate to nostalgia for times gone past, when, for example, white working class voters had a higher social status relative to ethnic minorities (Gest et al., 2018; Mutz, 2018), or through whites considering themselves to be outnumbered, or oppressed in contemporary society (Jardina, 2019). Furthermore, the direct experience of economic decline (model 1), and the heuristic of personal economic loss (model 2), is argued to have an indirect effect on support for populist causes via the related perceived threat of immigration, nativist sentiment and fear of outsiders, the role elite rhetoric plays in shaping mass attitudes toward immigration and cultural grievances more generally (Carreras et al., 2019) and the way in which direct experiences of economic transformation, such as deindustrialisation, are related to dominant group status threat (Baccini and Weymouth, 2021). Hence, these models are inherently overlapping, as well as having potential direct effects as well.

The final model (‘cued partisanship’) takes its lead from mainstream models of public opinion formation in political science from Zaller (1992) onwards, which have seen voters as vulnerable to cue taking and particularly selective cue taking from political elites. While there are various ways in which a person’s local economic experiences may drive their partisanship, there is ample evidence that partisanship leads to misperceptions (Wlezien et al., 1997; De Boef and Kellstedt, 2004; Evans and Andersen, 2006; Gerber & Huber, 2009, 2010; Pickup and Evans, 2013; Bullock et al., 2013; Prior et al., 2015). Partisanship can colour the perception of economic decline as citizens who are too busy or uncertain about how to judge something so complex rely on cues from trusted political leaders and parties. They may simply be loyal in their survey response, reacting more positively (negatively) to any change that has occurred under a preferred (disliked) party. Their partisanship may influence which evidence is absorbed, and which is ignored, with partisans of the incumbent party ignoring negative outcomes, and partisans of rival parties being more attentive to them (Taber & Lodge, 2006). For the effects of partisanship, we have two expectations. We expect opposition partisans to be systematically more negative (to perceive more decline) than those of the governing party. In addition, as Bailey (2019) argues, a logical consequence of partisan motivated reasoning is that partisanship moderates the link between economic change and perceptions. We therefore expect partisanship to moderate the relationship between change in the local economic context and perceived decline.Footnote 2

Drawing on these expectations, our hypotheses correspond to the ‘testable implications’ column in Table 1 in respect of perceptions of decline:

H1:

Local contexts of economic change are associated with decline perceptions.

H2:

:Personal and national economic perceptions are associated with decline perceptions.

H3:

:Older people are more likely to perceive decline.

H4:

Perceptions of increased immigration are associated with perception of decline.

H5:

Political identities and preferences are associated with perceptions of decline.

H6:

The relationship between local contexts of economic change and decline perceptions is moderated by partisanship.

Research Design

We introduced new survey items into the British Election Study (BES) internet panel (Fieldhouse et al., 2017, wave 12), fielded in May to June 2017. This was one year after the Brexit referendum, before the UK’s formal exit from the EU (in January 2020) and before the subsequent inflation surge (from 2021 to 2023), and it preceded waning support for Brexit in the UK.

A survey measure of local economic perceptions should allow different drivers of what kinds of economic change are relevant to the respondent, and should allow respondents to refer to whichever local community they are thinking about and ideally not ask about the direction of change per se – which will exaggerate bias and error – but ask about specific time points. If the time period is too long it will be unrealistic for people to answer with any reliability. Too short and it might not be relevant to the major economic changes that should be associated with decline (for example, the period encompassing deindustrialisation or import shocks). Motivated by these considerations, we defined perceived economic decline as the observed difference between evaluations of the local economy in the past, ‘15 or 20 years ago’, and evaluations of the local economy ‘now’. The time period is not spuriously precise, it coincides with relevant economic changes, and it avoids coinciding with a change of government (which might enhance partisan associations). We define ‘local’ as whatever a respondent considers their local area to be.

The question wording was: “Thinking about the local area where you live at the moment, how is the economic situation now and how was it 15 or 20 years ago?” [bold in original]. We specified ‘where you live at the moment’ because many people may have moved from a different area (and to avoid picking up change due to respondents moving area). Possible response options ranged from 1 ‘very bad’ to 5 ‘very good’, with a ‘don’t know’ option for each.

We are interested in the response to the economic decline survey questions by age, gender, how long someone has lived in the area, a respondent’s attention or information-seeking, educational level, and caring responsibilities. All variables are outlined in the accompanying Appendix. In assessing decline, we are also interested in a person’s partisanship. Furthermore, our study, focusing on the UK, raises the possibility that other identities, such as ‘Brexit identities’ (Hobolt et al., 2021) are also a driver of economic perceptions. Although the Brexit example is a UK-specific and temporally contingent one, it acts as an informative test of the relationship of other political identities to decline perceptions. We include a standard measure of partisanship (see below) and also whether respondents supported Leave or Remain in the EU referendum.Footnote 3 We also control for respondent ethnicity (white British or ethnic minority).

We include a measure of whether respondents think immigration is getting higher or lower, which should be related to economic decline perceptions (and support for Leave or Remain), and a standard retrospective measure of how the national economy is perceived to be doing. Demographics are measured in standard ways, including economic measures of household income, whether the respondent owns their own home (with or without a mortgage), whether household finances have got better or worse, and whether respondents are in the labour market (employed part-time or full-time). We capture whether someone is new to the local area through assessing whether they changed Local Authority District (LAD) between waves 10 and 12 of the BES panel (a five to seven month period – the only BES waves for which we can calculate this).

Our theoretical mechanisms imply the importance of local contexts of economic change. We use Census and Office for National Statistics (ONS) data linked to BES data (the respondent’s current residence) to explore whether perceptions of decline correspond to official data over an equivalent period. We use data across a full date range spanning 2002 to 2017, capturing decline at the LAD level (using the lower tier of authorities where there are multiple tiers): there are 355 local authorities in England and Wales. Using this level of geography parallels the existing literature (see, e.g. Ansell and Adler, 2019) and makes multiple economic indicators available. We use three common economic indicators for which there are relatively localised data. The first is the increase in levels of unemployment between 2004 (when the measures are first available at this level of geography) and 2017, measured via the proxy of the percentage of working-age adults who are claiming benefits principally for the reason of being unemployed.Footnote 4 The second is the change in median house prices in the local authority, based on ONS records of the (inflation-adjusted) price paid for all dwellings in 2002 and 2017. The third is the change in mean disposable incomes at LAD level between 2002 and 2017, based on ONS records of (inflation-adjusted) gross disposable household income at each time point. Using these three indicators, we compute a latent objective decline score to simplify analysis and capture decline in overall economic conditions. To calculate this, we first z-transform each of the indicators, and use structural equation modelling (SEM) to fit a latent model (see Appendix Figure A1 for the SEM diagram). Our latent decline score is the factor score for each area calculated on the basis of this model. Since we mean to distinguish between the effects of the state of the economy per se, and that of decline, we also use this approach to create an indicator as a control capturing the latent distress of the economy as of mid-2017, based on the static indicators of unemployment, disposable income, and median house prices.

Finally, we explore the degree to which respondents’ partisan affiliations are associated with perceptions of local economic decline. The governments in this period were Labour (until 2010) and then a Conservative-led coalition, majority, and minority administrations (2010–2017). Since respondents are asked the questions under a (majority) Conservative government, we expect Labour partisans to be most likely to report decline. We assess this by including a variable for partisanship (whether the respondent reports identifying as a Labour partisan, Conservative partisan, a partisan of another party, or expresses no partisan affiliation) and we use an interaction of latent objective economic decline and partisanship to assess whether different partisans are more or less likely to exhibit an association between perceptions and objective economic measures. We fit mixed-effects multilevel models with random intercepts by Local Authority District (matching the local economic statistics). Our sample size varies from 6,755 in the models of having/not having change perceptions to 5,474 in the models of perceived decline (i.e. the direction of change perceptions), yielding around 16–20 respondents per level-2 unit.

Results: Who Perceives Decline?

Figure 1 presents the coefficient plot of results of a model of respondents’ answering both the past and present questions about the economic conditions in the local area (Table A1 of the appendix, model 1).Footnote 5 The findings in Fig. 1, in addition to the fact that 22 per cent of our sample could not answer both questions (with 90 per cent of respondents answering the ‘now’ version and 79 per cent answering the ‘15 or 20 years ago’ version), provide an interesting caution to direct or simplified assumption about local decline effects.

Younger respondents are significantly less likely to answer both questions than older respondents, and women are less likely than men, but there are higher rates of response to both questions among respondents with caring responsibilities, higher levels of education (to A-level, not degree level), high use of news and current affairs information, people who own their home and people who have lived in the area for longer, supporting our initial expectations.

Fig. 1
figure 1

Coefficient plot of covariates on answering both past and present questions about the local economy

Full size image

Attention to politics and information-seeking behaviour have effects in the expected direction, both being positively associated with answering the questions. We did not expect that partisanship would be correlated with the propensity to respond to both questions, but people without a partisan affiliation are less likely to answer both questions.

We next provide the equivalent coefficient plot with the difference in the past and current questions as the dependent variable (perceived economic decline) in Fig. 2 (see also Table A1, model 2).Footnote 6

Fig. 2
figure 2

Coefficient plot of effects of covariates on perceived decline

Full size image

Figure 2 shows us who is likely to perceive decline among those who answered both questions; namely women, those on lower incomes, people who do not own their home (either living with family or renting privately or through a housing association), people whose own household finances are getting worse, and people who perceive increases in immigration and worse national economic outcomes. We also find a very substantial relationship with partisanship. Interestingly, perceived decline tends to be found among people who report greater attention to politics, but lower consumption of information about news/current affairs sources.

We cannot say for sure why women are more likely to perceive decline. This finding holds controlling for caregiving responsibilities, the number of children in the household, personal economics and work status. However, we know that women tend to report greater concern for other policy areas, including for the environment (Xiao & McCright, 2012), and also greater risk aversion (Croson and Gneezy, 2009) and so it seems plausible that women are in general more concerned about the local economy, although less likely to answer both ‘now’ and ‘then’ questions in the first instance.

To a substantial degree, people’s perceptions of local decline are working through their own individual experiences, as we hypothesised (H2).Footnote 7 Our findings give support to interpretations that link people’s personal economic hardships to localised economic perceptions. In so doing, they suggest some caution on interpretation of how such relationships could have a causal effect. That is to say, if localised decline perceptions are – at least partly – working through one’s own financial circumstances, those local effects may be at least partly an outcome of aggregation or composition.

We do not find the effects that might have been expected from the literature on populism – and in the British case – on Brexit. That is to say, older respondents are no more likely than younger respondents to perceive decline (although older respondents are more likely to answer both questions), there is no effect for education level, and only being in the highest income bracket is associated with perceiving less local decline. Moreover, there is no association with having voted Leave or Remain in Britain’s 2016 referendum on membership of the EU.Footnote 8 These findings caution use of some of the assumptions that might be associated with voting Leave; that older, retired individuals in particular might be more nostalgic for a bygone and rosier past, particularly in respect to their local communities. That also stands in contradiction to our hypothesis (H3). Note that this does not refute the relationship of age to nostalgia, but more specifically that this nostalgia does not appear to be related to perceptions of local economic decline.

Of interest to the literature on the association between immigration concerns and economics (see, for example, Carreras et al., 2019), respondents who perceive local economic decline are more likely to believe immigration levels are getting higher (confirming H4), although this relationship is substantively quite small. The effect of moving up one category (for instance, going from ‘Getting a little higher’ to ‘getting a lot higher’) is associated with a 0.06 point increase in perceived decline (0.75% of the scale). This compares to a 0.41 point (5.12%) increase in perceived decline associated with believing one’s personal finances had got ‘worse’ as opposed to having stayed the same.

How Perceptions Correspond to Objectively Measured Decline

Figure 3 presents the results of a regression model of subjective perceptions of economic decline that introduces our measure of latent local objective economic conditions (which is composed of unemployment data, house price data, and wage data, each observed between 2002 and 2017), and an interaction with partisanship. This model includes all earlier variables (not presented in Fig. 3), which can be found below in Table 2.Footnote 9

We control for current economic conditions (in 2017) in order to isolate any decline effects from the local economic conditions near to the time point of the survey, measured objectively, and estimate two separate models – one in which we also estimate an interaction between objective economic decline and partisanship (light grey in Fig. 3, from model 2 in Table 2), and one without this interaction (dark grey in Fig. 3, from model 1).

Fig. 3
figure 3

Coefficient plot of effects of context and partisanship on perceived economic decline

Full size image
Table 2 Regression models of perceived decline with contextual predictors and partisanship
Full size table

We find a significant relationship between our latent objective economic measure and perceptions of decline (β = 0.11, p < 0.001), which is an encouraging finding in respect of the growing body of research using aggregate data to estimate the link between local context and individual-level behavioral outcomes and attitudes (and supporting H1). This occurs controlling for other correlates of perceiving decline (being a woman, experiencing a deterioration in household finances, thinking that immigration is getting higher, the national economy is getting worse), which remain significant. Relationships between local decline and its purported outcomes are not only a function of individual experiences (which would suggest compositional rather than contextual effects), although these individual experiences drive considerable heterogeneity in perceptions even when people experience the same local economy.

Across model specifications, Conservative party and ‘other’ (combining the Liberal Democrats, the Greens, and nationalist parties) identifiers are significantly less likely to perceive decline compared to Labour party identifiers, which is also true for people with no identification. This reflects the potential for endogeneity in these evaluations (as is well demonstrated for other economic perceptions), although could also indicate a possible causal effect of perceiving decline on people’s partisanship. Note that the model already controls for the different economic (household) experiences of individuals, income, home ownership, and it also controls for the current objective economic conditions in the respondent’s local area.

The second way in which partisanship shapes decline perceptions is in the responsiveness to localized decline, as proposed in H6. This can be seen in the interaction between the latent objective decline measure and the respondent’s partisan affiliation. Figure 3 reports the interaction of partisanship with objective economic decline (shown in Table 2, model 2), and Fig. 4 (below) reports the predicted values of decline for this interaction.

Fig. 4
figure 4

Predicted values of perceived decline, by latent decline and party ID

Full size image

Conservative identifiers are substantially less likely to report decline given variations in objective local circumstances, people with no partisanship are more likely to report decline regardless of variations in local circumstances, whereas Labour partisans – and those expressing identification with other parties - are more likely to report decline given variations in their objective local economies. Together, this suggests that, to a notable degree, partisanship is related to people’s responsiveness to their local circumstances. If local economic decline matters, it does so less for Conservative party identifiers in Great Britain – even Conservatives who live in areas of higher local decline. For Labour partisans, the worse the objective economic conditions of the local area (in terms of over-time decline), the more these individuals report local economic decline. Being a partisan identifier of the incumbent (the Conservatives) – even controlling for the covariates such as age and education level that are strongly associated with supporting the Conservatives (Fieldhouse et al., 2021), as well as the economic circumstances of the individual – makes these individuals less responsive to the local reality in which they live. While survey error and endogeneity are possibilities, more substantively interesting explanations are also plausible: perhaps partisans of the governing party are genuinely less responsive and attentive to ‘bad’ local economic outcomes, or out-party partisans (Labour supporters) are more attentive to local economic decline, following elite cues about local economic difficulties and outcomes.

While these findings may provide some reassurance for researchers who use objective economic indicators in their analyses of voting, populism and political discontent, they also emphasize the systematically different ways in which people observe and report decline around them, in particular structured by partisan identity.

Robustness of Results

We conducted a series of checks to ensure the robustness of our findings, particularly around the effects of objectively-measured decline. First, we tested for the consistency of results with three different objective indicators of long-term change in the local economy: rises or falls in unemployment, house prices, and household disposable income. Each had a statistically significant effect consistent with that of the latent measure. Second, we tested the effects of latent decline at a different, more granular level of geography: the Middle Super Output Area (MSOA). Research in political geography can be sensitive to measurement level due to issues such as the Modifiable Areal Unit Problem (Wong et al., 2012). Bisbee and Zilinsky (2023) find that the effect of the local economy on economic perceptions varies with scale, stronger at more specific geographies. Tables A2 to A4 of the appendix summarise the results. We find that change in income, and even more so changes in unemployment, are associated with decline perceptions. The larger effect of unemployment is also in line with Bisbee and Zilinsky (2023). Changes in house prices do not have a statistically significant effect in this model.

We also tested whether the heterogenous effects apply across measures, in particular, whether Conservative identifiers are less responsive to decline. We find that this is indeed the case for wages and house prices, though not for unemployment. At the MSOA-level, we confirm that decline perceptions are associated with latent decline, but we do not find evidence of partisan heterogeneity in this effect. This suggests that the link between perceptions and objective data is equal across partisanship at smaller units of geographic analysis, but less stable when compared to larger units of geographic analysis. These robustness checks support our main finding that ‘real’ and perceived decline are linked (i.e., the effects described above are not merely a function either of geographic level or of constructing a latent measure), but the partisan relationship to accurate decline perceptions depends, to a small degree, on the level of geography we compare perceptions against. This suggests that partisan differences tend to emerge, in our data, for larger units than they do for smaller units of analysis, which is itself an interesting insight.

Conclusions

At the heart of many accounts of populism and political alienation is the idea that ‘geographies of discontent’ emerge through local economic decline. Starting from this premise, these are perceived to lead to geographic differences in disaffection and alienation, perceptions of status threat/loss, and associated perceptions of cultural threat, out-group animus and in-group identification. The role of the local economy is also important in accounts of economic voting, influencing the reward-punishment dynamic for incumbent governments. Despite the importance of local economic decline perceptions, there is one assumed causal link that has, so far, gone untested. Can people form opinions about local economic decline, and who is most likely to do so? We began with the premise that these kinds of assessments are not especially easy. We proposed five mechanisms that relate to local economic decline assessments specifically, either as alternative pathways or in combination. We explored the association of local economic decline perceptions with objective measures of the local economy over time, the relationship of partisanship and other heuristics to perceptions of local economic decline, and the role of collective memory and cultural nostalgia.

Our findings point to considerable heterogeneity in people’s ability to assess local economic decline. There is systematic variation in people’s ability to answer questions about the economy ‘now’ and the local economy ‘15 to 20 years ago’. That is not a surprising finding, but it is of importance for assumed effects from experience to measurable perceptions to political expression. We find that women, younger respondents, people who do not own their home, have not lived in an area for as long, who do not have caring responsibilities and who have lower political attention or information consumption, are less likely to report these perceptions. That is of note, especially if we consider areas characterised by high mobility, lower home ownership rates, and having more young people. Perhaps those areas are less likely to give rise to economic grievances that would be expected to develop when decline occurs.

Turning to subjective perceptions of economic decline, we first pointed to the importance of direct experience, and the forceful arguments made in the populism literature for widespread awareness of decline. We assessed this model comparing subjective economic decline perceptions to available corresponding objective data and found a statistically meaningful correlation, controlling for individual-level characteristics. This is an encouraging finding for studies that rely on contextual data to estimate decline and, to the best of our knowledge, has not been evidenced to date. However, it is far from the whole story.

We theorised and found evidence of the importance of heuristic reasoning through a close correspondence between decline perceptions, evaluations of the national economy and the economic circumstances of the individual; cultural resentment through a relationship of perceived increases in immigration and perceptions of local economic decline; and cued partisanship. We found a direct association between partisanship and decline perceptions, as well as a significant moderating effect of partisanship, with identifiers of non-incumbent parties being more likely to hold perceptions that correspond to objective data on economic conditions over our time period. We also hypothesised that older respondents would perceive more decline, according to a potential for a localised nostalgia of things being better in the past, but this hypothesis was not supported.

These results help to overcome some potential problems of the ecological fallacy in inferring local economic effects only from objective data. We expect these effects to be rooted in people’s perceptions, although at the same time noting considerable heterogeneity and potential partisan bias. To the degree that objective economic conditions may matter via measured perceptions, we would expect any effects to be heightened among individuals whose personal economic circumstances are most difficult. These people will have more negative personal economic evaluations and more negative perceptions of local decline in their area. And we propose that by measuring perceptions of the local economy, we can – to some degree – overcome the disputes about the causal role of economics and culture in driving people’s backlash against populism and political grievances. The finding that perceptions of increasing immigration is associated with perceptions of local decline has implications for the literature linking cultural grievances to local economic grievance (Carreras et al., 2019), and the wider link between ‘culture’ and economics (Carreras, 2019; Greve et al. 2023; Green, Hellwig and Fieldhouse, 2022).

Our findings are relevant to debates about directing political resources into alleviating local economic decline. There may be indirect benefits to politicians for doing so, such as communicating that politicians care about certain areas and represent certain voters. But if governments hope to boost their fortunes through overturning perceptions of decline, whether voters deem them to be successful in doing so is likely to be partisan, related to how the national economy is doing and contingent on the personal economic experiences of households. Politicians may find themselves responding to different demands from their supporters, and governments may be absolved of some responsibility for decline among their own voters – a finding with important implications. We also found that women were more likely to perceive decline, compared to men. This might suggest that political efforts to alleviate local economic difficulties would find more resonance for female voters, even taking into account the range of individual level economic and other experiences (such as caregiving responsibilities) that we included in our analyses. Understanding this tendency would be an important area for future research, corresponding, perhaps, to greater risk aversion among women (Croson and Gneezy, 2009) and greater reported concerns in other unrelated policy areas (Xiao & McCright, 2012).

Our analyses highlight the usefulness of this novel measure of local economic decline. To better understand the implications of these findings, future research should investigate how subjective perceptions of economic decline are formed in other political and economic contexts, considering, for example, the role of political institutions in mediating these relationships, such as the degree of local governance and variations in clarity of responsibility for local and national economic performance. Researchers might also further consider how local economic experience translates into political discontent, and under what conditions. For now, our conclusions call for nuance in any assumption that observing and experiencing long-term local economic decline should lead uniformly or otherwise directly to localised discontent, anti-incumbent voting, populism, economic grievances and political distrust. It has the potential to do so for some individuals, but this is much more heterogeneous - influenced by personal circumstances and political considerations - than has been assumed to date.