Machine learning goes beyond: Time-varying monetary policy and oil price pass-through to inflation expectations

This study investigates the effects of monetary policy shocks on crude oil prices over time, with a particular emphasis on their effect on households’ inflation expectations. Employing a novel machine learning framework, we estimate models with time-varying parameters, incorporating the potential evolution of shock volatility through ridge regressions. This approach allows for flexible adjustment to time variations. Our findings indicate that households have become less sensitive to changes in both inflation and crude oil prices when forming their inflation expectations, particularly since 2000. Consequently, fluctuations in inflation or oil prices have a diminished effect on households’ one-year and five-year ahead inflation expectations. Moreover, our findings, particularly a temporary increase during the early phase of the global financial crisis, may help shed light on the broader “missing disinflation” puzzle observed during that period. Our results provide implications for developing effective monetary policy strategies that can adapt to evolving households’ inflation expectations.