The way central banks define a 'housing boom' is basically a coin flip for whether they can actually see a financial crisis coming.
SSRN · March 16, 2026 · 6418315
Why it matters
Economists often treat the choice of statistical 'filters' as a minor technical detail. This study reveals that only persistent, long-term deviations from housing price trends signal a coming collapse, while short-run price accelerations—often the focus of media panic—are statistically harmless to banking stability.
From the abstract
Not all housing booms carry the same banking crisis risk. Using a global panel of 131 countries and an empirical pipeline combining Double Machine Learning, doubly robust AIPW, Causal Forest, and staggered difference-in-differences, this paper provides evidence that HP-filter booms — which capture persistent deviations from long-run trend — raise next-year banking crisis probability by approximately 3.8 percentage points, while Hamilton-filter booms — which capture short-run price acceleration —