Work in progress

Abstract: The idea that monetary policy may have non-linear effects has a long history in economics. The literature has provided a plethora of potential sources of non-linearity, but most previous studies have used low dimensional modelling approaches that allow for only one or two channels to operate simultaneously. This study takes a high dimensional “big data” approach, allowing us to systematically evaluate which of many non-linear channels matter for transmission. Using a large, mixed-frequency dataset designed to incorporate many sources of non-linearity in the literature, we establish that monetary policy transmission to asset prices has multi-dimensional state-dependence with economically large non-linear effects. Strong real economic conditions strengthen transmission, but this can be dominated by other effects at times, including from financial variables. As such, transmission is not reducible to low-dimensional stratification, such as recessions vs. expansions. We show that low dimensional approaches can suffer from considerable omitted variable bias.

*The slides from the presentation at the ECB ChaMP WS2 workshop (March 2024) are available here. The first draft is available here.


Abstract: Following subdued inflation in the last decade, inflation rates have surged unexpectedly after the pandemic in many countries, including Europe. Such an environment necessitates a comprehensive understanding of underlying inflation even more than before. This study investigates the dynamics of trend inflation both for the euro area (EA) countries separately and also for the EA aggregate rates. To this aim, trend inflation rates are estimated using various forms of a flexible unobserved components model, allowing for outliers and stochastic volatility. Using granular sector-level data at the country applications provides a finer comparison between the member countries. For the aggregate euro area application, both models exploiting the sector and country levels are considered to understand underlying dynamics more thoroughly. Additionally, the alternative estimates for the euro area are obtained by pooling country-level results. The results show that using multivariate models mostly improves precision and outperforms univariate models at longer horizons. A horse race between these estimates and the other popular measures from the literature, such as PCCI, supercore, etc., shows that no single metric outperforms the others. Therefore, it remains crucial to have various measures in the toolbox of policymakers as they complement each other.

The draft will be available soon.


Publications