Quantifying Minsky Cycles

We develop a novel sentiment measure derived from survey data to examine the empirical relevance of the Minsky–Kindleberger view on financial crises. Using survey data from multiple countries, we isolate the component of beliefs that reflects the misaggregation of public infor mation relative to a machine benchmark constructed from the same in formation set. The sentiment measure exhibits predictive relationships with financial markets and belief dynamics, with elevated sentiment followed by financial-market repricing and a subsequent downward re vision in survey expectations. We extend this sentiment measure histor ically for a panel of 78 countries using machine learning models trained on BERT embeddings of historical news articles (1903–2020). The back casted sentiment shows that shocks in mediansentiment predict credit booms in the non-tradable corporate sector, which prior research has linked to financial crises, providing new historical evidence consistent with the Minsky–Kindleberger view. The historical backcasting proce dure and the resulting credit-boom evidence prove robust to a wide range of concerns about language drift, model instability, cross-country portability, spurious machine-learning fit, and the influence of particu lar countries, periods, and extreme observations. We further find that this sentiment component is shaped by memory-related dynamics, as the time elapsed since major crises and the share of young-to-old peo ple in the population predict surges in optimism even when recent eco nomic developments are controlled for.