An ethno-linguistic dimension in transnational activity space measured with mobile phone data

The study situates ethnic segregation and inequality within a mobility-centered perspective, arguing that activity spaces frequently span national borders. Traditional single-country, residence-focused approaches risk overlooking how mobility between activity sites (home, work, school, leisure) connects opportunities and reproduces or mitigates inequalities. The paper aims to: (1) extend segregation research to transnational activity spaces, and (2) assess differences in cross-border mobility between ethno-linguistic groups in Estonia, with particular attention to how often and how long people travel abroad, days spent abroad, number of countries visited, and membership in visitor groups (tourists, commuters, transnationals, long-term stayers). Estonia provides a pertinent context due to pronounced ethno-linguistic divisions (Estonian majority, Russian-speaking minority) across residence, schools, labor market, and leisure, and the availability of rich mobile phone data enabling longitudinal tracking of multiple forms of cross-border mobility. Research questions: (1) What ethno-linguistic differences exist in cross-border mobility regarding number and duration of trips, days abroad, and countries visited? (2) How do ethno-linguistic differences manifest across visitor groups and destination countries?

Theoretical background integrates activity-space-based segregation, mobility justice, and transnationalism. Activity space research shows segregation varies across sites (residence typically highest; workplaces lower; leisure mixed) and that mobility connects these sites, influencing exposure to opportunities and social networks. Vicious circles of segregation are shaped by policies, resources, discrimination, and lived experience; mobility access is a key prerequisite yet unequally distributed. Mobility justice posits that (im)mobility realizes (in)equalities and can shift disparities across space and borders. Transnational activities encompass physical cross-border movements and digital ties, including return visits to countries of ancestry, work/study abroad, and second-home use. Distance matters but is not linear; many ties are regional, though long-distance links can be strong. Return visits sustain identity, language, social capital, and may evolve across generations from VFR to heritage tourism; ethnic minorities may engage in cross-border business leveraging local knowledge. Prior quantitative work on ethno-linguistic differences is limited, often small-sample or single-mobility type. Evidence suggests minorities may travel more frequently, often to ancestral homelands, with destination preferences shaped by emotional ties and generational change. Estonia’s historical Soviet-era migration created a large Russian-speaking community with persistent segregation across domains; citizenship and visa regimes (Estonian, Russian, Alien’s passport) shape cross-border frictions. EU accession in 2004 expanded ties with Finland, while Russia remains a key ancestral destination for Russian-speakers.

Data and study design: Passive mobile positioning data (Call Detail Records, CDRs) from an Estonian mobile network operator for 2014–2016. Two datasets: (1) Outbound roaming CDRs for Estonian SIMs when abroad (country-level location via ISO alpha-2); (2) Domestic CDRs for the same users within Estonia. Records include timestamp, cell location, and pseudonymous SIM ID; anonymity preserved. Anchor point model (Ahas et al., 2010) applied monthly to domestic data to infer meaningful locations (home, work), with residence assigned to the county having the highest count of monthly home anchor points, aggregated to five NUTS3 regions (North, West, Central, South, East Estonia). Personal attributes from operator contracts: gender, birth year, and preferred communication language (Estonian, Russian, English). Language used as a proxy for ethno-linguistic background; English speakers excluded. Datasets linked via pseudonymous IDs. Inclusion criteria: users with at least one outbound trip, available personal characteristics (Estonian or Russian language, gender, birth year), and identifiable residence; final sample N=75,118 (85% Estonian-speakers; 15% Russian-speakers). Trip/visit construction: Using outbound roaming and domestic data, trips start/end in Estonia; within a trip, multiple country visits possible. For each visit, start/end time determined from first/last abroad call activity; trip length depends on call activity timing. Mobility measures per person computed for 2014–2016: number of trips, number of distinct countries visited, average trip duration, and total days abroad. Visitor group classification (computed annually, by destination country): - Long-term stayer: ≥75% of days in one destination; <25% in Estonia. - Transnational: ≥25% of days in Estonia and 25–75% in destination(s); at least one visit in ≥6 months of the year. - Commuter: <25% of days in destination(s) and ≥25% in Estonia; at least two visits in ≥6 months. - Tourist: all others, residing mainly in Estonia. A person can belong to different groups for different destination countries (e.g., commuter to Finland, tourist to Greece). Group composition (overall): Estonian-speakers: 91% tourists, 6% transnationals, 2% commuters, 1% long-term stayers; Russian-speakers: 92% tourists, 4% transnationals, 3% commuters, 1% long-term stayers. Statistical analysis: - Count outcomes (dependent variables): (1) number of trips (2014–2016), (2) number of distinct countries (2014–2016): zero-truncated negative binomial regression. (3) average trip duration, (4) number of days abroad: negative binomial regression. Overdispersion tested (AER package). Predictors: communication language (Estonian vs Russian), gender, birth-year category, and residence region. Incident rate ratios (IRRs) reported. - Visitor-group membership: separate binary logistic regressions for tourist, commuter, transnational, and long-term stayer (dependent variable 1/0; odds ratios reported). Ethical/data protection: Compliance with EU GDPR; data anonymized; no link to identifiable persons.

• Overall intensity: Russian-speakers exhibit higher cross-border mobility than Estonian-speakers among those who traveled at least once (2014–2016). Descriptive medians: Russian-speakers vs Estonian-speakers: number of trips 6 vs 4; average trip duration 4 vs 3 days; total days abroad 20 vs 15; number of distinct countries visited equal at median (3 vs 3). - Regression-adjusted effects (language): Russian-speakers make 10% more trips (IRR≈1.10, p<0.001), have 14% longer trips on average (p<0.001), and spend 17% more days abroad (p<0.001) than Estonian-speakers, but visit fewer countries. - Gender: Women make fewer trips than men, have shorter trips, and spend less time abroad; no gender differences in number of countries visited. - Age: Largest differences between youngest and oldest groups; across all age groups, Russian-speakers have higher median number of trips and days abroad than Estonian-speakers. - Residence: East Estonia (Russian-speaking regional majority; lowest incomes) shows lower cross-border mobility: fewer trips, shorter trips, fewer countries, and fewer days abroad, controlling for other factors. - Destinations: Estonian-speakers’ top destinations: Latvia (59%), Finland (59%), Sweden (33%). Russian-speakers’ top: Russia (65%), Latvia (49%), Finland (37%). Younger Russian-speakers show increased ties to Finland while maintaining strong links to Russia across all age groups. - Visitor groups (odds): Russian-speakers have higher odds of being commuters (≈+20%) and markedly higher odds of being tourists (≈+88%) compared to Estonian-speakers; no major age differences in commuter or tourist odds; women less likely commuters but more likely tourists. - Destination patterns by group: Long-term stayers (both language groups) concentrate in high-income Nordics (Finland, Sweden, Norway). Transnationals: Estonian-speakers mainly Finland (65%), then Sweden (8%), Norway (8); Russian-speakers Finland (33%), Russia (24%), Sweden (13%). Commuters: Estonian-speakers primarily Finland (44%); Russian-speakers primarily Russia (40%). Tourism differences strongest: Estonian-speakers most often visit Latvia (61%); Russian-speakers Russia (65%); both visit Mediterranean countries, with Italy more common among Estonian-speakers.

Findings substantiate that ethno-linguistic background structures cross-border mobility within a transnational activity space. Despite lower average incomes among Russian-speakers, their higher travel intensity suggests other drivers—social networks, ancestral ties, proximity, and mobility strategies—shape mobility beyond economic resources alone. Persistent travel to ancestral homelands underscores the role of cross-border ties in sustaining identity across generations, while heightened links to Finland among younger Russian-speakers signal evolving, multi-directional transnational connections. Regional disparities (East Estonia) highlight how structural constraints (border regime, Schengen, lower incomes) dampen mobility. Gender differences align with broader mobility patterns. From an activity-space segregation perspective, transnational mobility can both reproduce and disrupt inequalities: work-related commuting and long-term stays may provide economic exits from marginal positions, potentially reducing segregation in origin contexts, while tourist and VFR flows can reinforce cultural-linguistic cohesion across borders. The mobility justice lens suggests that (in)equalities may be transferred not only across activity sites but also across countries, warranting multi-scalar analysis of segregation and opportunity structures beyond state borders.

The study advances segregation research by operationalizing a transnational activity space and quantifying ethno-linguistic differences in cross-border mobilities using large-scale mobile phone data. It shows that Russian-speaking minorities in Estonia travel more frequently, stay longer, and spend more time abroad, with strong ancestral and regional destination patterns, and higher odds of commuter and tourist statuses. Conceptually, the work integrates mobility justice with activity-space segregation, evidencing how cross-border mobility can both sustain identities and offer pathways to mitigate socio-economic marginality. Methodologically, it demonstrates the value of CDRs for distinguishing visitor types and tracking multi-year mobility. Future research should: (1) integrate richer variables (trip purpose, socio-economic indicators, data usage roaming) and multi-operator, multi-country datasets; (2) incorporate immobile populations to assess selection and inequality; (3) examine how transnational mobility transmits advantages/disadvantages between origin and destination contexts; (4) study intra-destination mobility and segregation using cross-country CDRs; and (5) explore policy implications of border regimes and infrastructure on equitable mobility.

• Data limitations: CDRs lack trip motivation/purpose and direct socio-economic measures; outbound country-level roaming may undercount users relying on data services without call/SMS activity; inclusion restricted to users who provided personal attributes and made at least one outbound trip, excluding immobile individuals. - Measurement constraints: Time abroad inferred from first/last call activity can misestimate stay durations; language preference used as a proxy for ethno-linguistic identity may misclassify some users; residence inferred from anchor points may have residual error. - Coverage/scope: Single-operator dataset from one origin country (Estonia) limits generalizability; lack of harmonized data from destination operators limits intra-destination mobility analysis. - Structural context: Border regime and visa differences affect observed mobility; findings predate COVID-19, which significantly altered mobility patterns.