Broken Promises of Mobility? Understanding Status Expectations and Political Dissatisfaction through Machine Learning

Opportunities for upward social mobility have stalled in many advanced democracies, leaving younger cohorts with fewer prospects than their parents despite being more highly educated. While earlier generations experienced upward mobility through educational expansion and economic growth, younger cohorts confront fewer opportunities for occupational advancement. This breach of the ‘mobility contract’ has been associated with declining trust and rising political dissatisfaction. To investigate these dynamics, the study applies Machine Learning to longstanding panel data. By capturing complex, non-linear relationships between social origins and occupational outcomes, Random Forest algorithms model intergenerational status expectations in ways that conventional approaches cannot.
The analysis focuses on status discordance, the gap between achieved occupational status and the status reasonably expected on the basis of parental background. Unlike earlier studies, discordance is distinguished into negative (achieving less than expected) and positive (achieving more), allowing for the examination of asymmetries in their association with political dissatisfaction. This distinction provides a fuller account of how intergenerational inequality is associated with political dissatisfaction, considered through its expression in generalised social trust, levels of political interest, and electoral participation.
The empirical analysis draws on the British Household Panel Study (1991-2008) and the UK Household Longitudinal Study (2009-2024). Status expectations are modelled on the parental generation (aged 60-75) using Random Forest algorithms, then applied out-of-sample to their offspring cohort. The analysis focuses on individuals aged 30-45, for whom predicted occupational status is compared with achieved outcomes to derive a continuous measure of status discordance, distinguishing between negative and positive cases. Checks confirm the absence of multicollinearity among class origin, destination, and our mismatch measures. Associations with political dissatisfaction are examined using linear probability models.
Findings indicate only weak associations between status discordance and political dissatisfaction once class of origin, class of destination, and education are considered. Electoral participation and political interest show limited links to negative discordance, while positive discordance is not systematically related to either outcome. By contrast, generalised social trust is consistently lower among those underperforming relative to their parents. This erosion of trust signals a broken social contract: younger generations experiencing downward mobility express diminished confidence in other members of their society.
The most significant consequence of disappointed mobility expectations lies not in electoral disengagement but in the erosion of social trust. By integrating Machine Learning with panel data, the study offers methodological and substantive advances, showing that enduring socio-economic inequalities remain the primary anchors of political discontent.