Machine Learning Forecasts Reveal a Concentration Paradox in UK International Student Mobility
Year: 2026
Type: preprint
Abstract: International student mobility (ISM) underpins the financial resilience, soft power, and knowledge production of major destination countries, yet its future trajectory is increasingly uncertain. In the UK, recent policy shifts and geopolitical tensions have coincided with a slowdown in international student growth and mounting concerns over sectoral stability. We use administrative data from the Universities and Colleges Admissions Service (UCAS) covering successful undergraduate applications from 86 origin countries between 2010 and 2024, linked with economic and demographic projections, to forecast international applications to the UK through to 2030. A Poisson-based XGBoost model is trained and evaluated against ARIMA and negative binomial gravity benchmarks on an out-of-sample period spanning Brexit and the COVID-19 pandemic. The machine learning model delivers consistently lower forecast errors and stronger correspondence with observed flows, but gains are modest in the presence of structural shocks. Forecasts indicate that applications from China and Hong Kong, India, and the European Union will together comprise over 60% of international demand by 2030, despite stagnation or decline in absolute numbers from all three blocs. We term this configuration a `concentration paradox', in which reliance on a narrowing set of origins intensifies even as inflows soften, increasing the vulnerability of the UK higher education system to policy and geopolitical disruption.
Source: Research Square
Authors: Ruth Neville, Francisco Rowe, Emilio Zagheni
Language: English
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7164847211