When Does National Identity Matter?
Economic Context and the Conditional Activation of Immigration Attitudes
The Puzzle
National identity is one of the strongest predictors of restrictive immigration attitudes across all known surveys.
Yet national identities and immigration attitudes appear to be stable (Hess & Torney 1968; Kustov, Laaker & Reller 2021).
But look at the BES panel over eleven years:
- The association fluctuates substantially wave to wave
- The strength of the relationship varies over time
- Yet national identity itself is relatively stable
The question: Why does a stable predisposition generate such variable political consequences?
What We Know (and Don’t)
Established:
- National identity → restrictive ATI (robust cross-sectional evidence; Hainmueller & Hopkins 2014; Toruńczyk-Ruiz et al. 2025)
- Economic conditions → ATI (as direct predictors; Dancygier & Donnelly 2013; Chen 2024)
- GDP and globalization moderate the nationalism–ATI link across countries (Pehrson, Vignoles & Brown 2009; Ariely 2012)
The gap:
- No study has asked systematically: do economic conditions determine when national identity becomes politically potent?
- Most of existing studies compare different people or different countries
- They show that identity matters
- But they cannot isolate how changing economic conditions shape its effects within the same individuals
Our contribution: A framework where economic conditions determine when identity matters, tested using panel data that tracks the same individuals over time, across three levels: individual, local, and national
The Framework: Conditional Activation
National identity = stable predisposition
Economic conditions = situational trigger (Sniderman et al. 2004)
Three mechanisms linking economic adversity to stronger identity–ATI slopes:
- Threat salience: economic hardship heightens the psychological salience of outgroup threats, making nationalist predispositions more consequential for attitudes toward immigration (Gheorghiev and Collini, 2025)
- Status compensation: economic disadvantage increases investment in national group as prestige source (Shayo 2009)
- Politicized framing: local conditions interact with political rhetoric to make immigration salient (Hopkins 2010)
Three levels of economic context
| Level | Empirical focus | Activation mechanism |
|---|---|---|
| Micro | Individual conditions | Personal threat perception |
| Meso | Local conditions | Group boundary salience |
| Macro | National conditions | Sociotropic evaluation |
Hypothesis
Core expectation: economic context conditions when national identity becomes politically consequential.
H1: Baseline
- Stronger nationalist orientations will be associated with more restrictive immigration attitudes.
H2: Micro activation
- The nationalism–ATI relationship will be stronger among individuals experiencing greater personal economic insecurity: lower income and unemployment risk
H3:Meso activation
- The nationalism–ATI relationship will be stronger in constituencies marked by local economic deterioration: higher unemployment, greater deprivation, or sharper economic decline.
H4: Macro activation
- The nationalism–ATI relationship will be stronger during periods of national economic deterioration: rising unemployment, negative GDP growth, or heightened macroeconomic uncertainty.
The Britishness–Englishness Asymmetry
Economic conditions are expected to affect identity-based attitudes unevenly, depending on the type of identity (Gidron & Shayo 2026)
Britishness: broad, civic-compatible, distributed across UK population
Englishness: more ethnically exclusive, politically reactive, tightly coupled to immigration/sovereignty discourse post-Brexit (Ford & Goodwin 2014; Sobolewska & Ford 2020)
Expectation: Englishness should be more sensitive to economic contextual activation than Britishness, both because its political content is more exclusionary and because the status-compensation and politicized-framing mechanisms apply more forcefully to it
Two rival predictions for which identifiers are activated (H5a vs H5b):
- H5a: Strong identifiers: adversity reinforces already-central identity
- H5b: Weak/moderate identifiers: strong identifiers already hold stable, restrictive views; those with attitudinal latitude are more responsive to cues (Converse 1964)
Data and Design
British Election Study Internet Panel (2014–2025)
- 30 waves, ~86,000 unique respondents, ~453,000 person-wave observations
- 18 waves with immigration attitudes; both Britishness and Englishness across most waves
Three DVs: Self-placement on immigration levels (10-pt, main), economic ATI (7-pt), cultural ATI (7-pt). Higher values = more permissive
Two key independent variables: Britishness and Englishness, both measured on a 1–7 scale capturing the intensity of national identification. Higher values = stronger national identification
Main Identification strategy: Two-way fixed effects
\[y_{it} = \alpha_i + \lambda_t + \beta \cdot \text{Identity}_{it} + \gamma \cdot (\text{Identity}_{it} \times \text{Econ}_{it/t}) + \varepsilon_{it}\]
- \(\alpha_i\): absorbs all time-invariant individual differences
- \(\lambda_t\): absorbs all common temporal shocks (including political events, elite rhetoric)
- Identification = within-person change only
Macro indicators vary only at wave level → main effects absorbed by \(\lambda_t\) → identification rests entirely on interactions
Economic Variables
| Level | Variable | Source |
|---|---|---|
| Micro | Equivalised household income | BES survey |
| Micro | Perceived unemployment risk | BES survey |
| Meso | Constituency unemployment rate | ONS/APS |
| Meso | Claimant count | ONS benefits records |
| Meso | Median weekly earnings | ASHE |
| Macro | National unemployment rate | ONS |
| Macro | CPIH inflation | ONS |
| Macro | GDP per capita | ONS |
Meso variables merged by parliamentary constituency and fieldwork year. Macro variables merged at wave level.
Result 1: Baseline (H1) ✓
Within individuals, increases in national identification predict more restrictive immigration preferences:
- Britishness: β = 0.042* per point on 7-pt scale
- Englishness: β = 0.068* — consistently and larger
The Englishness–Britishness gap holds across all specifications, all DVs, all robustness checks, including categorical re-specification which shows a monotonically ordered gradient for both identities, with Englishness uniformly larger at every position
The wave-by-wave plot reveals substantial temporal variation in the association, motivating the moderation analysis
Result 2: Micro Level (H2) ✗
No significant results
Null finding: Neither household income nor perceived unemployment risk conditions the identity–ATI relationship in any specification. Interaction terms ≈ 10⁻⁸, SE ≈ 10× larger.
But this null is theoretically informative, not just a non-finding:
- Perceived unemployment risk has a small negative effect on national identification itself (β ≈ −0.006 to −0.008)
- When individuals feel personally economically insecure, they become marginally less identified, opposite of Shayo (2009) status-compensation prediction at the individual level
Implication: The activation mechanism does not operate through personalized material circumstances. It operates through collectively experienced economic conditions, consistent with the sociotropic voting literature (Lewis-Beck & Stegmaier 2007)
Result 3: Meso Level — H3 Partial ✓
. . .
Britishness:
- Responds only to unemployment.
- As unemployment rise, the identity-ATI link shifts downward (driven by weak identifiers).
- Effect concentrated among weak identifiers
Englishness:
- Responds more strongly to unemployment and claimant count
- Economic context reshape the identity-ATI link, also driven by weaker identifiers
- Effects concentrated among weak identifiers
Result 4: Macro Level — H4 ✓
Unemployment and inflation:
Strengthen the relationship between identity and ATI (both identities)
Englishness:
Stronger response to inflation than Britishness (~1.6× larger)
Macro-level activation is the most consistent pattern.
Macroeconomic conditions amplify how strongly identity translates into immigration attitudes, widening identity-based differences.
The H5a/H5b Verdict
Across levels:
| Level | H5a (strong IDs activated) | H5b (weak/moderate IDs activated) |
|---|---|---|
| Micro | — (no conditional effects) | — (no conditional effects) |
| Meso | ✗ | ✓ Effects concentrated among weak identifiers (unemployment, claimant) |
| Macro | Partial (effects not limited to strong IDs) | ✓ Effects strongest among weak/moderate identifiers, but not exclusively concentrated |
H5b receives stronger support. Economic context primarily expands the base of restrictive attitudes by shifting those with weaker or moderate identification, rather than intensifying positions among strong identifiers.
Summary of Findings
H1 ✓ Both Britishness and Englishness predict restrictive ATI within individuals. Englishness effect is consistently ~60% larger.
H2 ✗ No micro-level activation. Individual insecurity marginally reduces national identification — opposite of status-compensation logic at personal level.
H3 Partial ✓ Constituency unemployment activates Britishness; Englishness responds additionally to claimants and earnings. Effects concentrate among weaker identifiers.
H4 ✓ National unemployment and inflation are the strongest activators for both identities.
H5b ✓ Activation concentrates among weaker/intermediate identifiers, not the already-committed.
Three Broader Implications
1. Reconciling cultural vs. economic accounts of ATI
The debate is misfocused. Economic conditions do not compete with cultural identity, they determine when cultural identity becomes politically consequential. Cross-sectional dominance of cultural factors is fully consistent with the within-individual evidence here.
2. How economic crises work politically
Economic crises operate not only through direct grievance, but by activating existing national identity. The key is not just who suffers economically, but which individuals identify nationally among those exposed to economic strain. . . .
3. Temporal dynamics of public opinion
A predisposition can be stable while its political expression is not. Theories of opinion stability (Kustov, Laaker & Reller 2021) and instability are both partially right, operating at different levels of the attitude system.
Limitations and Next Steps
Current limitations:
- Single-country panel: generalisability to other contexts remains to be established
- Limited number of wave-level macro observations (~18–30) constrains the precision of interaction estimates
- Cannot distinguish weather activation is driven by economic conditions themselves, or by how these conditions are politically framed
Future directions:
- Multi-country panel replication (ESS rotating panel; BHPS)
- Quasi-experimental designs exploiting local economic shocks (plant closures, regional deindustrialisation)
- Supply-side variation: does political salience of immigration rhetoric condition the economic activation mechanism?
Conclusion
The question is no longer whether national identity predicts immigration preferences.
The question is under which economic conditions its attitudinal consequences become most pronounced, through which mechanisms, and for which identifiers.
Economic context does not simply add to or subtract from immigration attitudes. It determines when the identity–attitude link is switched on.
Thank you!
