From digital search to deed: forecasting UK housing purchases in Spain using Google Trends across the Brexit disruption

Purpose This study examines whether digitally expressed interest, measured through Google search activity for terms such as “Spain villas” between 2014 and 2024, is associated with residential property purchases in Spain by British nationals. It also explores whether this relationship remains stable in the presence of major institutional shocks, notably Brexit and the COVID-19 pandemic. Design/methodology/approach Using quarterly data for 2014–2024, the paper estimates autoregressive (AR) and AR models augmented with search intensity (ARX). Model specifications with and without Google Trends indicators are compared. The Brexit referendum and the COVID-19 pandemic are treated as structural breaks to assess changes in the relationship between search activity and realized transactions. Several alternative specifications and lag structures are used as robustness checks. Findings Prior to the 2016 Brexit referendum, Google search intensity is positively associated with residential purchases by British nationals and improves short-term forecasting performance. After 2016, this association weakens substantially: search activity remains elevated, while transactions do not follow pre-referendum patterns. The evidence suggests a structural change in the relationship between digitally expressed interest and realized housing transactions, particularly during periods of heightened uncertainty. Practical implications Online search data may complement traditional indicators in forecasting cross-border housing demand. However, their usefulness appears to depend on institutional stability, as predictive performance may decline during periods of political and macroeconomic disruption. Originality/value The paper contributes to the literature on digital behavioral indicators and international housing markets by documenting time variation in the association between online search activity and cross-border housing transactions, highlighting the importance of institutional context in predictive applications.